UNIT I SERVICES AND TECHNICAL CHALLENGES PART A 1. Define Mobile radio communication.

n. Mobile radio communications describes any radio communication link between two t erminals of which one or both are in motion or halted at unspecified locations a nd of which one may actually be a fixed terminal such as a base station. 2. What are the types of services provided in wireless communication? (1)Paging systems (2)Cordless Telephone systems (3)Cellular telephone systems 3. What are all effects of small scale Multipath Propogation?. (i) Rapid changes in signal strength over a small travel distance (ii) Random frequency modulation due to varying Doppler shifts on different m ultipath signal (iii) Time dispersion caused by multipath propagation delays 4.What are all the multiple access schemes used in wireless communication? 5. What is the use of Multiple access Schemes? Multiple access schemes are used to allow many mobile users to share sim ultaneously a finite amount of radion spectrum. 6. Find the intermodulation frequencies generated if a base station transmits tw o carrier frequencies at 1930 MHz and 1932 MHz that are amplified by a saturated clippling amplifier. If the mobile radio band is allocated from 1920 MHz to 19 40 MHz, designate the IM frequencies that lie inside and outside the band. Solution: Intermodulation distortion products occur at frequencies mf1 + nf2 all i nteger values of m and n, i.e., - < m, n < . Some of the possible intermodulation frequencies that are produced by a nonlinear device are (2n + 1)f1 2nf2, (2n + 2)f1 (2n + 1)f2, (2n + 1)f1 2nf2, (2n + 2)f2 (2n + 1)f1, etc. for n = 0, 1, 2, .. 7. If a US AMPS cellular operator is allocated 12.5 MHz for each simplex band, a nd if Bt is 12.5 MHz, Bguard is 10 kHz, and BC is 30 kHz, find the number of cha nnels available in an FDMA system. Solution: The number of channels available in the FDMA system is given as

In the US, each cellular carrier is allocated 416 channels. 8. Consider Global System for Mobile, which is a TDMA / FDD system that uses 25 MHz for the forward link, which is broken into radio channels of 200 kHz. If 8 speech channels are supported on a single radio channel, and if no guard band is assumed, find the number of simultaneous users that can be accommodated in GSM.

Solution: The number of simultaneous users that can be accommodated in GSM is give n as

Thus, GSM can accommodate 1000 simultaneous users 9. If W = 1.25 MHz, R = 9600 bps, and a minimum acceptable Eb / N0 is found to b e 10 dB, determine the maximum number of users that can be supported in a single cell CDMA system using (a) omnidirectional base station antennas and no voice a ctivity detection, and (b) three sectors at the base station and activity detect ion with = 3/8. Assume the system is interference limited. Solution: (a) Using Equation

(b) Using Equation for each sector, we can find NS The total number of users is given by 3NS since three sectors exist within a ce ll; therefore N = 3 x 35.7 = 107 users / cell. 10. Define cell-splitting. Cell splitting is the process of subdividing a congested cell into small er cells, each with its own base station and a corresponding reduction in antenn a height and transmitter power.

New cell radius = old cell radius / 2 New cell Area = old cell Area / 4 New traffic load / unit = 4 * old traffic load / Unit area 11 MTSO- Mobile Telephone Switching Office. CO- Central office (Telephone Exchange) CS- Cell Site BS- Base Station. 12. What are the types of Principle Cellular networks? Broadband PCS Narrowband PCS

13. Write the abbreviation for the following terms. MTSO Mobile Telephone Switching Office PCS - Personal communication systems ADC American Digital Cellular JDC Japanese Digital Cellular 14. What are the types of cell splitting? Dyanamic cell splitting

Permanent cell splitting 15. Define Base Station. A fixed station in a mobile radio system used for radio communication wi th mobile stations. Base stations are located at the center or on the edge of a coverage region and consist of radio channels and transmitter and received anten nas mounted on a tower. 16. Define Mobile Station. A station in the cellular radio service intended for use while in motion at unspecified locations. Mobile stations may be hand-held personal units (port ables) or installed in vehicles (mobiles). 17. Define Mobile Switching Center. Switching center which coordinates the routing of cells in a large servi ce area. In a cellular radio system., the MSC connects the cellular base station s and the mobiles to the PSTN. An MSC is also called a mobile telephone switchin g offices(MTSO). 18. Define Page. A brief message which is broadcast over the entire service area, usually in a simulcast fashion by many base stations at the same time. 19. What is mean by Subscriber? A user who pays subscription charges for using a mobile communication sy stem. 20. What is mean by Transceiver? A device capable of simultaneously transmitting and receiving radio sign als. 21. What are the transmission used in mobile radio systems? Mobile radio transmission systems are classified as (i) (ii) (iii) Simplex Half-duplex Full-duplex

22. What are the types of modulation techniques? Analog modulation Digital modulation 23. Define Frequency Reuse or Frequency Planning. Each cellular base station is allocated a group of radio channels to be used within a small geographic called a cell. 24. Define Cluster. The N cells which collectively is the complete set of available frequenc ies is called a cluster. If a cluster is replicated M times, within the system, the total number of duplex channels, C, can be used as a measure of capacity and is given,

C = MkN = MS The capacity of a cellular system is directly proportional to the number of times a cluster is replicated in a fixed service area. The factor N is calle d the cluster size and is typically equal to 4,7, or 12. 25. Define Interference. Sources of interference include another mobile in the same cell, a call in progress in a neighboring cell, other base stations operating in the same fre quency band, or any noncellular system which inadvertently leaks energy into th e cellular frequency band. 26. What are the types of cellular interference systems? Co-channel Interference Adjacent Co-channel Interference 27. Define Co-Channel Interference System. Frequency reuse implies that in a given coverage area there are several cells that use the same set of frequencies. These cells are called co-channel ce lls, and the interference between signals from these cells is called co-channel interference. 28. Define Co-Channel Cell. Frequency reuse implies that in a given coverage area there are several cells th at use the same set of frequencies. These cells are called co-channel cells.

29. Define channel reuse radio. Interference is reduced from improved isolaltion of RF energy from the co-channe l cell. The parameter Q, called the co-channel reuse ratio, is related to the cl uster size. Q = D / R = 3N 30. Define Adjacent Channel Interference or Non Co-Channel Interference System. Interference resulting from signals which are adjacent in frequency to the desired signal is called adjacent channel interference. Adjacent channel int erference results from imperfect receiver filters which allow nearby frequencies to leak into the passband. 31. Define Trunking. Cellular radio systems rely on trunking to accommodate a large number of users in a limited radio spectrum. The concept of trunking allows a large numbe r of users to share the relatives small number of channels in a cell by providin g access to each user, on demand, from a pool of available channel. 32. Define Grade of Service. GOS is a measure of the ability of a user to access a trunked system dur ing the busiest hour. GOS is typically given as the likelihood that a call is bl

ocked, or the likelihood a call experiencing a delay grater than a certain queui ng time. 33. What are the types of cell splitting? Permanent Dynamic 34. Mention the methods to improve coverage and capacity in cellular systems. 1. 2. 3. 4. Cell splitting Sectoring Repeaters for range extension Microcell zone.

PART B 1. Explain the FDMA. Frequency Division Multiple Access (FDMA) Frequency division multiple access (FDMA) assigns individual channels to individual users. It can be seen from Figure that each user is allocated a uni que frequency band or channel. These channels are assigned on demand to users w ho request service. During the period of the call, no other user can share the same channel. In FDD systems, the users are assigned a channel as a pair of freq uencies; one frequency is used for the forward channel, while the other frequenc y is used for the reverse channel. The features of FDMA are as follows: Table: Multiple Access Techniques Used in Different Wireless Communication Syste ms Cellular System Multiple Access Technique Advanced Mobile Phone System (AMPS) FDMA/FDD Global System for Mobile (GSM) TDMA/FDD US Digital Cellular (USDC) TDMA/FDD Pacific Digital Cellular (PDC) TDMA/FDD CT2 (Cordless Telephone) FDMA/TDD Digital European Cordless Telephone (DECT) FDMA/TDD US Narrowband Spread Spectrum (IS 95) CDMA/FDD W-CDMA(3GPP) CDMA/FDD CDMA/TDD edma2000 (3GPP2) CDMA/FDD CDMA/TDD

Figure: FDMD where different channels are assigned different frequency bands The FDMA channel carries only one phone circuit at a time. If an FDMA channel is not in use, then it sits idle and cannot be used by other users to increase or share capacity. It is essentially a wasted resource. After the assignment of a voice channel, the base station and the mobile transmi t simultaneously and continuously. The bandwidths of FDMA channels are relatively narrow (30 kHz in AMPS) as each c hannel supports only one circuit per carrier. That is, FDMA is usually implemen ted in narrow band systems. The symbol time of a narrowband signal is large as compared to the average delay spread. This implies that the amount of intersymbol interference is low and, t hus, little or no equalization is required in FDMA narrowband systems. The complexity of FDMA mobile systems is lower when compared to TDMA systems, th ough this is changing as digital signal processing methods improve for TDMA. Since FDMA is a continuous transmission scheme, fewer bits are needed for overhe ad purposes (such as synchronization and framing bits) as compared to TDMA. FDMA systems have higher cell site system costs as compared to TDMA systems, bec ause of the single channel per carrier design, and the need to use costly bandpa ss filters to eliminate spurious radiation at the base station. The FDMA mobile unit uses duplexers since both the transmitter and receiver oper ate at the same time. This results in an increase in the cost of FDMA subscribe r units and base stations. FDMA requires tight RF filtering to minimize adjacent channel interference. Nonlinear Effects in FDMA: In an FDMA system, many channels share the same anten na at the base station. The power amplifiers or the power combiners, when opera ted at or near saturation for maximum power efficiency, are nonlinear. The nonl inearities cause signal spreading in the frequency domain and generate intermodu lation (IM) frequencies. IM is undesired RF radiation which can interfere with other channels in the FDMA systems. Spreading of the spectrum results in adjace nt channel interference. Intermodulation is the generation of undesirable harmo nics. Harmonics generated outside the mobile radio band cause interference to a djacent services, while those present inside the band cause interference to othe r users in the wireless system [Yac93]. Table: Lists several intermodulation product terms Table: Intermodulation Products n = 0 1930 1928 1932 1934 n = 1 1926 1924 1936 1938 n = 2 1922 1920 1940 1942* n = 3 1918 1916 1944* 1946*

The frequencies in the table marked with an asterisk (*) are the frequencies tha t lie outside the allocated mobile radio band. The first US analog cellular system, the Advanced Mobile Phone System (A MPS), is based on FDMA / FDD. A single user occupies a single channel while the call is in progress, and the single channel is actually two simplex channels wh ich are frequency duplexed with a 45MHz split. When a call is completed, or whe n a handoff occurs, the channel is vacated so that another mobile subscriber may use it. Multiple or simultaneous users are accommodated in AMPS by giving each user a unique channel. Voice signals are sent on the forward channel from the

base station to mobile unit, and on the reverse channel from the mobile unit to the base station. In AMPS, analog narrowband frequency modulation (NBFM) is use d to modulate the carrier. The number of channels that can be simultaneously su pported in a FDMA system is given by

Where Bt is the total spectrum allocation, Bguard is the guard band allo cated at the edge of the allocated spectrum band, and BC is the channel bandwidt h. Note that Bt and BC may be specified in terms of simplex bandwidths where it is understood that there are symmetric frequency allocations for the forward ba nd and reverse band. 2. Explain the TDMA. Time Division Multiple Access (TDMA)

Time division multiple access (TDMA) systems divide the radio spectrum i nto time slots and in each slot only one user is allowed to either transmit or r eceive. It can be seen from figure that each user occupies a cyclically repeati ng time slot, so a channel may be thought of as a particular time slot that reoc curs every frame, where N time slots comprise a frame TDMA systems transmit dat a in a buffer and burst method, thus the transmission for any user is noncontinu ous. This implies that, unlike in FDMA systems which accommodate analog FM, dig ital data and digital modulation must be used with TDMA. The transmission from various users is interlaced into a repeating frame structure as shown in Figure. It can be seen that a frame consists of a number of slots. Each frame is made up of a preamble, an information message, and tail bits. In TDMA/TDD, half of the time slots in the frame information message would be used for the forward li nk channels and half would be used for reverse link channels. In TDMA / FDD sys tems, an identical or similar frame structure would be used solely for either fo rward or reverse transmission, but the carrier frequencies would be different fo r the forward and reverse links. In general, TDMA/FDD systems intentionally ind uce several time slots of delay between the forward and reverse time slots for a particular user, so that duplexers are not required in the subscriber unit. In a TDMA frame, the preamble contains the address and synchronization i

nformation that both the base station and the subscribers use to identify each o ther. Guard times are utilized to allow synchronization of the receivers betwee n different slots and frames. Different TDMA wireless standards have different TDMA frame structures, and some are described. The features of TDMA include the following:

TDMA shares a single carrier frequency with several users, where each user makes use of nonoverlapping time slots. The number of time slots per frame depends o n several factors, such as modulation techniques, available bandwidth, etc. Data transmission for users of a TDMA systems is not continuous, but occurs in b ursts. This results in low batter consumption, since the subscriber transmitter can be turned off when not in use (which is mot of the time). Because of discontinuous transmissions in TDMA, the handoff process is much simp ler for a subscriber unit, since it is able to listen for other base stations du ring idle time slots. An enhanced link control, such as that provided by mobile assisted handoff (MAHO) can be carried out by a subscriber by listening on an i dle slot in the TDMA frame. TDMA uses different time slots for transmission and reception, thus duplexers ar e not required. Even if FDD is used, a switch rather than a duplexer inside the subscriber unti is all that is required to switch between transmitter and recei ver using TDMA. Adaptive equalization is usually necessary in TDMA systems, since the transmissi on rates are generally very high as compared to FDMA channels. In TDMA, the guard time should be minimized. If the transmitted signal at the e dges of a time slot are suppressed sharply in order to shorten the guard time, t he transmitted spectrum will expand and cause interference to adjacent channels. High synchronization overhead is required in TDMA systems because of burst trans missions. TDMA transmissions are slotted, and this requires the receivers to be synchronized for each data burst. In addition, guard slots are necessary to se parate users, and this results in the TDMA systems having larger overheads as co mpared to FDMA. TDMA has an advantage in that it is possible to allocate different number of tim e slots per frame to different users. Thus, bandwidth can be supplied on demand to different users by concatenating or reassigning time slots based on priority . Efficiency of TDMA: The efficiency of a TDMA system is a measure of the percenta ge of transmitted data that contains information as opposed to providing overhea d for the access scheme. The frame efficiency, f, is the percentage of bits per frame which contain transmitted data. Note that the transmitted data may includ e source and channel coding bits, so the raw end user efficiency of a system is generally less than f. The frame efficiency can be found as follows:

The number of overhead bits per frame is [Zie92], bOH = Nr br + Ntbp + Nt bg + Nr bg (1) Where Nr is the number of reference bursts per frame, Nt is the number o f traffic bursts per frame, br is the number of overhead bits per reference burs t, bp is the number of overhead bits per preamble in each slot, and bg is the nu mber of equivalent bits in each guard time interval. The total number of bits p er frame, bT, is bT = Tf R (2) Where Tf is the frame duration, and R is the channel bit rate. The fram e efficiency f is thus given as (3) Number of channels in TDMA system: The number of TDMA channel slots that can be provided in a TDMA system is found by multiplying the number of TDMA slo ts per channel by the number of channels available and is given by (4) Where m is the maximum number of TDMA users supported on each radio chan nel. Note that two guard bands, one at the low end of the allocated frequency b and and one at the high end, are required to ensure that users at the edge of th e band do not bleed over into an adjacent radio service.

3. Explain the CDMA. Code Division Multiple Access (CDMA)

Figure: Spread spectrum multiple access in which each channel is assigned a uniq ue PN code which is orthogonal or approximately orthogonal to PN codes used by o ther users In code division multiple access (CDMA) systems, the narrowband message signal is multiplied by a very large bandwidth signal called the spreading signa l. The spreading signal is a pseudo-noise code sequence that has a chip rate wh ich is orders of magnitudes greater than the data rate of the message. All user s in a CDMA system, as seen from Figure, use the same carrier frequency and may transmit simultaneously. Each user has its own pseudorandom codeword which is a pproximately orthogonal to all other codewords. The receiver performs a time co rrelation operation to detect only the specific desired codeword. All other cod ewords appear as noise due to decorrelation. For detection of the message signa l the receiver needs to know the codeword used by the transmitter. Each user op erates independently with no knowledge of the other users. In CDMA, the power of multiple users at a receiver determines the noise floor after decorrelation. If the power of each user within a cell is not contr olled such that they do not appear equal at the base station receiver, then the near far problem occurs. The near far problem occurs when many mobile users share the same channe l. In general the strongest received mobile signal will capture the demodulator at a base station. In CDMA, stronger received signal levels raise the noise fl oor at the base station demodulators for the weaker signals, thereby decreasing the probability that weaker signals will be received. The combat the near far p roblem, power control is used in most CDMA implementations. Power control is pr ovided by each base station in a cellular system and assures that each mobile wi thin the base station coverage area provides the same signal level to the base s tation receiver. This solves the problem of a nearby subscriber overpowering th e base station receiver and drowning out the signals of far away subscribers. P ower control is implemented at the base station by rapidly sampling the radio si gnal strength indicator (RSSI) levels of each mobile and then sending a power ch ange command over the forward radio link. Despite the use of power control with in each cell, out-of-cell mobiles provide interference which is not under the co ntrol of the receiving base station. The features of CDMA including the followi ng: Many users of CDMA system share the same frequency. Either TDD or FDD may be us ed. Unlike TDMA or FDMA, CDMA has a soft capacity limit. Increasing the number of u sers in a CDMA system raises the noise floor in a linear manner. Thus, there is no absolute limit on the number of users in CDMA. Rather, the system performan ce gradually degrades for all users as the number of users is increased, and imp roves as the number of users is decreased. Multipath fading may be substantially reduced because the signal is spread over a large spectrum. If the spread spectrum bandwidth is greater than the coherenc e bandwidth of the channel, the inherent frequency diversity will mitigate the e ffects of small scale fading. Channel data rates are very high in CDMA systems. Consequently, the symbol (chi p) duration is very short and usually much less than the channel delay spread. Since PN sequences have low autocorrelation, multipath which is delayed by more than a chip will appear as noise. A RAKE receiver can be used to improve recept ion by collecting time delayed versions of the required signal. Since CDMA uses co-channel cells, it can use macroscopic spatial diversity to pr ovide soft handoff. Soft handoff is performed by the MSC, which can simultaneou sly monitot a particular user from two or more base stations. The MSC may chose the best version of the signal at any time without switching frequencies.

 Self jamming is a problem in CDMA system. Self jamming arises from the fact tha t the spreading sequences of different users are not exactly orthogonal, hence i n the despreading of a particular PN code, non-zero contributions to the receive r decision statistic for a desired user arise from the transmissions of other us ers in the system. The near far problem occurs at a CDMA receiver if an undesired user has a high d etected power as compared to the desired user. 4. Explain the SDMA. Space Division Multiple Access (SDMA) Space division multiple access (SDMA) controls the radiated energy for e ach user in space. It can be seen from figure that SDMA serves different users by using spot beam antennas. These different areas covered by the antenna beam may be served by the same frequency (in a TDMA or CDMA system) or different freq uencies (in a FDMA system). Sectorized antennas may be thought of as a primitiv e application of SDMA. In the future, adaptive antennas will likely be used to simultaneously steer energy in the direction of many users at once and appear to be best suited for TDMA and CDMA base station architectures. Figure: A spatially filtered base station antenna serving different users by usi ng spot beams The reverse link presents the most difficulty in cellular systems for se veral reasons [Lib94b]. First, the base station has complete control over the p ower of all the transmitted signals on the forward link. However, because of di fferent radio propagation paths between each user and the base station, the tran smitted power from each subscriber unit must be dynamically controlled to preven t any single user from driving up the interference level for all other users. S econd, transmit power is limited by battery consumption at the subscriber unit, therefore there are limits on the degree to which power may be controlled on the reverse link. If the base station antenna is made to spatially filter each des ired user so that more energy is detected from each subscriber, then the reverse link for each user is improved and less power is required. Adaptive antennas used at the base station (and eventually at the subscr iber units) promise to mitigate some of the problems on the reverse link. In th e limiting case of infinitesimal beam width and infinitely fast tracking ability , adaptive antennas implement optimal SDMA, thereby providing a unique channel t hat is free from the interference of all other users in the cell. With SDMA, al l users within the system would be able to communicate at the same time using th e same channel. In addition, a perfect adaptive antenna system would be able to track individual multipath components for each user and combine them in an opti mal manner to collect all of the available signal energy from each user. The pe rfect adaptive antenna system is not feasible since it requires infinitely large antennas. 5. Explain Capacity of Space division Multiple Access (SDMA. (OR) Explain the capacity of Cellular SDMA Capacity of Space Division Multiple Access For interference limited CDMA operating in an AWGN channel, with perfect power control with no interference from adjacent cells and with omnidirectional antennas used at the base stations, the average bit error rate, Pb, for a user can be found from the Gaussian approximation.

(1) Where K is the number of users in a cell and N is the spreading factor. Q(x) is the standard Q function. Equation (F = f x 100%) assumes that the sign ature sequences are random and that K is sufficiently large to allow the Gaussia n approximation to be valid. To illustrate how directive antennas can improve the reverse link in a s ingle cell CDMA system, consider figure, which illustrates three possible base s tation antenna configurations. The omnidirectional receiver antenna will detect signals from all users in the system, and thus will receive the greatest amount of noise. The sectored antenna will divide the received noise into a smaller v alue and will increase the number of users in the CDMA system (as illustrated). The adaptive antenna shown in figure (c) provides a spot beam for each user, an d it is this implementation which is the most powerful form of SDMA. An ideal a daptive antenna is able to form a beam for each user in the cell of interest, an d the base station tracks each user in the cell as it moves. Assume that a beam pattern, G( ), is formed such that the pattern has maximum gain in the direction of the desired user. Such a directive pattern can be formed at the base station using an N element adaptive array antenna. Assume that a beam pattern, G( ), wit h no variation in the elevation plane, such as shown in figure, can be formed by an array. The pattern, G( ), can be steered through 360 in the horizontal ( ) plane such that the desired user is always in the main beam of the pattern. It is ass umed that K users in the single cell CDMA system are uniformly distributed throu ghout a two dimensional cell (in the horizontal plane, = / 2), and the base stat ion antenna is capable of simultaneously providing such a pattern for all users in the cell. On the reverse link, the power received from the desired mobile si gnals is Pr;0. The powers of the signal incident at the base station antenna fr om K 1interfering users are given by Pr;i for i = 1, k 1. The average total inte rference power, I, seen by a single desired user, (measured in the received sign al at the array port of the base station antenna array, which is steered to the User 0), is given by (2) Where i is the direction of the ith user in the horizontal plane, measure d in the x axis, and E is the expectation operator. No interference from adjace nt cells contributes to total received interference in Equation (2). If perfect power control is applied such that the power incident at the base station anten na from each user is the same, then Pr;i = PC for each of the K users, and the a verage interference power seen by User 0 is given by

Figure: Illustration showing different antenna patterns: (a) An omni directional base station antenna pattern; (b) sectorized base station antenna pattern; (c) Adaptive antenna pattern which provides individual spot beams for each user is t he cell. (3) Assuming that users are independently and identically distributed throug hout the cell, the average total interference seen by a user in the central cell is given by (4) Where D is the directivity of the antenna, given by max (G( )). In typica l cellular installations, D ranges between 3 dB. As the antenna beam pattern is made more narrow, D increases, and the received interference I decreases propor tionally. The average bit error rate for User 0 can thus be given by

Figure: An idealized flat topped pattern with a 60 degree beamwidth and a 6dB si delobe level. The pattern has no variation in the elevation axis to account for users which are close to, and far from, the base station [From[Lib94b] IEEE] (5) Thus, it can be seen that the probability of bit error is dependent on t he beam pattern of a receiver, and there is considerable improvement that is ach ieved using high gain adaptive antennas at the base station. Using the fact that the additional interference from adjacent cells simp ly adds to the interference level, the average probability of error for a partic ular user using directive antennas in a multiple cell environment is given by (6) Where f is the frequency reuse factor described by equation and figure illustrates the average probability of error for different propagation path los

s exponents, where two different types of base station antennas are compared usi ng simulations which considered a single layer of interfering cells using the ge ometry described. In the figure, one set of average probability of error curves is found for a standard omnidirectional base station antenna, and another set o f curves is found for a flat top beam (a beam with constant gain over a specific angular region) having a directivity of about 5.1 dB. The flat top beam is ass umed to have a constant maximum gain lobe spanning an azimuth of 30 degrees, and a constant side lobe level which is 6 dB down from the maximum gain. Furthermo re, it is assumed that K separate flat top beams can be formed by the base stati on and pointed at each of the K users within the cell of interest. Notice that for an average probability of error of 0.001 in a propagation path loss environm ent of n = 4, the flat top beam system will support 350 users, whereas the omni directional antenna will support only 100 users [Lib94b]. This increase in the number of user is roughly equal to the directivity offered by the flat top beam system, and illustrates the promise SDMA offers for improving capacity in wirele ss systems. Note that multipath is not considered. The impact of scattering an d diffuse multipath on the performance of SDMA is currently a topic of research and is certain to impact performance and implementation strategies for emerging SDMA techniques.

Figure: Average probability of error for a user in a CDMA system with one interf ering layer of co-channel cells. Base station of interest using (a) an omnidire ctional base station antenna and (b) SDMA with a flat top beam having D = 5.1 dB and pointed at each of the K users within the cell of interest. Notice that SD MA offers significant capacity gains for a given average probability of error 4. Explain wireless communication system with Examples. Most people are familiar with a number of mobile radio communication sys tem used in everyday life. Garage door openers, remote controllers for home ent ertainment equipment, cordless telephones, hand-held walkie-talkies, pagers (als o called paging receivers or beepers), and cellular telephones are all examples of mobile radio communication systems. However, the cost, complexity, performance , and types of services offered by each of these mobile systems are vastly diffe rent. The term mobile has historically been used to classify and radio termina l that could be moved during operation. More recently, the term mobile is used to describe a radio terminal that is attached to a high speed mobile platform (e .g., a cellular telephone in a fast moving vehicle) whereas the term portable de scribes a radio terminal that can be hand-held and used by someone at walking sp eed (e.g., a walkie-talkie or cordless telephone inside a home). The term subsc riber is often used to describe a mobile or portable user because in most mobile

communication systems, each user pays a subscription fee to use the system, and each users communication device is called a subscriber unit. In general, the co llective group of users in a wireless system are called users or mobiles, even t hough many of the users may actually use portable terminals. The mobiles commun icate to fixed base stations which are connected to a commercial power source an d a fixed backbone network. Table lists definitions of terms used to describe e lements of wireless communication systems. Mobile radio transmission system may be classified as simplex, half-dupl ex or full-duplex. In simplex system, communication is possible in only one dir ection. Pages system, in which message are received but not acknowledges, are s implex system. Half-duplex radio systems allow two-way communication, but use t he same radio channel for both transmission and reception. This means that at a ny given time, a user can only transmit or receive information. Constraints lik e push-to-talk and release-to-listen are fundamental features of half-duplex systems . Full duplex systems, on the other hand, allow simultaneous radio transmission and reception between a subscriber and a base station, by providing two simulta neous but separate channels (frequency division duplex, or FDD) or adjacent time slots on a single radio channel (time division duplex, or TDD) for communicatio n to and from the user. Frequency division duplexing (FDD) provides simultaneous radio transmiss ion channels for the subscriber and the base station, so that they both may cons tantly transmit while simultaneously receiving signals form one another. At the base station, separate transmit and receive antennas are used to accommodate th e two separate channels. At the subscriber unit, however, a single antenna is u sed for both transmission to and reception from the base station, and a device c alled a duplexer is used inside the subscriber unit to enable the same antenna t o be used for simultaneous transmission and reception. To facilitate FDD, it is necessary to separate the transmit and receive frequencies by about 5% of the n ominal RF frequency, so that the duplexer can provide sufficient isolation while being inexpensively manufactured. In FDD, a pair of simplex channels with a fixed and known frequency sepa ration is used to define a specific radio channel in forward channels with chann el used to convey traffic to the mobile user from a base station is called the f orward channel, while the channel used to carry traffic from the mobile user to a base station is called the reverse channel. In the U.S. AMPS standard, the re verse channel has a frequency which is exactly 45 MHz lower than that of the for ward channel. Full duplex mobile radio systems provide many of the capabilities of the standard telephone, with the added convenience of mobility. Full duplex and half-duplex system use transceivers for radio communication. FDD is used e xclusively in analog mobile radio systems and is described. Time division duplexing (TDD) uses the fact that it is possible to share a single radio channel in time, so that a portion of the time is used to transm it from the base station to the mobile, and the remaining time is used to transm it from the mobile to the base station. If the data transmission rate in the ch annel is much greater that the end-users data rate, it is possible to store infor mation bursts and provide the appearance of full duplex operation to a user, eve n though there are not two simultaneous radio transmissions at any instant. TDD is only possible with digital transmission formats and digital modulation, and is very sensitive to timing. It is for this reason that TDD has only recently b een used, and only for indoor or small area wireless applications where the phys ical coverage distances (and thus the radio propagation time delay) are much sma ller than the many kilometers used in conventional cellular telephone systems. Paging Systems Paging system are communication systems that send brief messages to a su

bscriber. Depending on the type of service, the message may be either a numeric message, an alphanumeric message, or a voice message. Paging systems are typic ally used to notify a subscriber of the need to call a particular telephone numb er or travel to a known location to receive further instructions. In modern pag ing systems, news headlines, stock quotations, and faxes may be sent. A message is sent to a paging subscriber via the paging system access number (usually a t oll-free telephone number) with a telephone keypad of modem. The issued message is called a page. The paging system then transmits the page throughout the ser vice area using base stations which broadcast the page on a radio carrier. Paging systems vary widely in their complexity and coverage area. While simple paging systems may cover a limited range of 2 to 5km, or may even be con fined to within individual buildings, wide area paging systems can provide world wide coverage. Though paging receivers are simple and inexpensive, the transmis sion system required is quite sophisticated. Wide area paging systems consist o f a network of telephone lines, many base station transmitters, and large radio towers that simultaneously broadcast a page from each base station (this is call ed simulcasting). Simulcast transmitters may be located within the same service area or inn different cities or countries. Paging systems are designed to prov ide reliable communication to subscribers wherever they are; whether inside a bu ilding, driving on a highway, or highway, or flying in an airplane. This necess itates large transmitter powers (on order of kilowatts) and low data rates (a co uple of thousand bits per second) for maximum coverage from each base station. F igure shows a diagram of a wide area paging system.

Figure: A wide area paging system. The paging control center dispatches pages r eceived from the PSTN throughout several cities at the same time

Cordless Telephone Systems Cordless telephone systems are full duplex communication systems that us e radio to connect a portable handset to a dedicated base station, which is then connected to a dedicated telephone line with a specific telephone number on the public switched telephone network (PSTN). In first generation cordless telepho ne systems (manufactured in the 1980s), the portable unit communicates only to t he dedicated base unit and only over distances of a few tens of meters. Early c ordless telephones operate solely as extension telephones to a transceiver conne cted to a subscriber line on the PSTN and are primarily for in-home use.

Figure: A cordless telephone systems Second generation cordless telephones have recently been introduced whic

h allow subscribers to use their handsets at many outdoor locations within urban centers such as London or Hong Kong. Modern cordless telephones are sometimes combined with paging receivers so that subscriber may first be paged and then re spond to the page using the cordless telephone. Cordless telephone systems prov ide the user with limited range and mobility, as it is usually not possible to m aintain a call if the user travels outside the range of the base station. Typic al second generation base stations provide coverage ranges up to a few hundred m eters. Figure illustrates a cordless telephone system. Cellular Telephone Systems A cellular telephone system provides a wireless connection to the PSTN f or any user location within the radio range of the system. Cellular systems acc ommodate a large number of users over a large geographic area, within a limited frequency spectrum. Cellular radio systems provide high quality service that is often comparable to that of the landline telephone systems. High capacity is a chieved by limiting the coverage of each base station transmitter to a small geo graphic area called a cell so that the same radio channels may be reused by anot her base station located some distance away. A sophisticated switching techniqu e called a handoff enables a call to proceed uninterrupted when the user moves f rom one cell to another. Figure shows a basic cellular system which consists of mobile st ations, base stations and a mobile switching center (MSC). The mobile switching center is sometimes called a mobile telephone switching office (MTSO), since it is responsible for connecting all mobiles to the PSTN bin a cellular system. E ach mobile communicates via radio with one of the base stations and may be hande d-off to any number of base stations throughout the duration of a call. The mob ile station contains a transceiver, an antenna, and control circuitry, and may b e mounted in a vehicle or used as a portable hand-held unit. The base stations consist of several transmitters and receivers which simultaneously handle full d uplex communication and generally have towers which support several transmitting and receiving antennas. The base station serves a bridge between all mobile us ers in the cell and connects the simultaneous mobile calls via telephone lines o r microwave links to the MSC. The MSC coordinates the activities of all of the base stations and connects the entire cellular system to the PSTN. A typical MS C handles 100,000 cellular subscribes and 5,000 simultaneous conversations at a time, and accommodates all billing and system maintenance functions, as well. I n large cities, several MSCs are used by a single carrier.

Figure: A cellular system. The towers represent base station which provide radi o access between mobile users and the mobile switching center (MSC) Communication between the base station and the mobiles is defined by a s tandard common air interface (CAI) that specifies four different channels. The c hannels used for voice transmission from the base station to mobiles are called forward voice channels (FVC), and the channels used for voice transmission from

mobiles to the base station are called reverse voice channels (RVC). The two ch annels responsible for initiating mobile calls are the forward control channels (FCC) and reverse control channels (RCC). Control channels are often called set up channels because they are only involved in setting up a call and moving it to an unused voice channel. Control channels transmit and receive data messages t hat carry call initiation and service requests, and are monitored by mobiles whe n they do not have a call in progress. Forward control channels also serve as b eacons which continually broadcast all of the traffic requests for all mobiles i n the system. As described inn Chapter 11, supervisory and data message are sen t in a number of ways to facilitate automatic channel changes and handoff instru ction for the mobiles before and during a call. 5. Compare the common wireless communication systems. Comparison of Common Wireless Communication Systems Tables illustrate the type of service, level of infrastructure, cost, a nd complexity required for the subscriber segment and base station segment of ea ch of five mobile or portable radio systems discussed earlier in this chapter. For comparison purposes, common household wireless remote devices are shown in t he table. It is important to note that each of five mobile radio systems given in Table use a fixed base station, and for good reason. Virtually all mobile ra dio communication systems strive to connect a moving terminal to a fixed distrib ution system of some sort and attempt to look invisible to the distribution syst em. For example, the receiver tin the garage door opener converts the received signal into simple binary signal which is sent to the switching center of the ga rage motor. Cordless telephones use fixed base stations so they the cordless ph one base station and the portable handset is designed to behave identically to t he coiled cord connecting a traditional wired telephone handset to the telephone carriage. Notice that the expectations vary widely among the services, and the inf rastructure costs are dependent upon the required coverage area. For the case o f low power, hand-held cellular phones, a large number of base station are requi red to insure that any phone is in close range to a base station within a city. If base stations were not within close range, a great deal of transmitter power would be required of the phone, thus limiting the battery life and rendering th e service useless for hand-held users. Because of the extensive telecommunicati ons infrastructure of cooper wires, microwave line-of-sight links, and fiber opt ic cables-all of which are fixed-it is highly likely that future land-based mobi le communication system will continue to rely on fixed base stations which are c onnected to some type of fixed distribution system. However, emerging mobile sa tellite networks will require orbiting base stations. Table: Comparison of Mobile Communication Systems-Mobile Station Service Coverage Range Required Infrastructure Complexity Hardware Cost Carrier Frequency Functionality TV Remote Control Low Low Low Low Infrared Transmitter Garage Door Opener Low Low Low Low <100 MHz Transmit ter Table: Comparison of Mobile Communication Systems-Base Station Service Coverage Range Carrier Frequency TV Remote Control Garage Door Opener Paging System High Required Infrastructure Functionality Low Low Low Low Low Low High High High Complexity Hardware Cost Receiver Receiver

Low Infrared Low <100 MHz <1 GHz Transmitter

Cordless Phone Low Cellular Phone High

Low High

Low High

Moderate 1-3 GHz Transceiver High <2 GHz Transceiver

6. Explain the trends in cellular Radio and personal communications. Trends in Cellular Radio and Personal Communications Since 1989, there has been enormous activity throughout the world to dev elop personal wireless systems that combine the network intelligence of todays PS TN with modern digital signal processing and RF technology. The concept, called personal Communication Service (PCS), originated in the United Kingdom when thr ee companies were given spectrum in the 1800 MHz range to develop Personal Commu nication Networks (PCN) throughout Great Britain [Rap91c]. PCN was seen by the U.K. as a means of improving its international competitiveness in the wireless f ield while developing new wireless systems and services for citizens. Presently , field trials are being conducted throughout the world to determine the suitabi lity of various modulation, multiple-access, and networking techniques for futur e 3G PCN and PCS systems. The terms PCN and PCS are often used interchangeably. PCN refers to a w ireless networking concept where any user can make or receive calls, no matter w here they are, using a light-weight, personalized communicator. PCS refers to n ew wireless systems that incorporate more network features and are more personal ized than existing cellular radio systems, but which do not embody all of the co ncepts of an ideal PCN. Today, cellular and PCS are identical in functionality and differ only in frequency band. Indoor wireless networking products are rapidly emerging and promise to become major par of the telecommunications infrastructure within the next decade . As discussed in Chapter 2, an international standards body, IEEE 802.11, is d eveloping standards for wireless access between computers inside buildings. The European Telecommunications Standard Institute (ETSI) is also developing the 20 Mbps HIPERLAN standard for indoor wireless networks. Breakthrough products suc h as Motorolas 18 GHz Altair WIN (wireless information network) modem, that was n ot commercialized, and Avayas (formerly NCR and Lucent) / ORINOCO WaveLAN compute r modem have been available as wireless Ethernet connections since 1990 [Tuc93] and are beginning to penetrate the business world. As we enter the 21st century , products are emerging that allow users to link their phone with their computer within an office environment, as well as in a public setting, such as an airpor t or train station. A worldwide standard, the Future Public land Mobile Telephone System (FPLMTS)-re named International Mobile Telecommunication 2000 (IMT-2000) in mid-1995-has bee n formulated by the International Telecommunications Union (ITU) which is the st andards body for the United Nations, with headquarters in Geneva, Switzerland. The technical group TG 8/1 standards task group is within the ITUs Radio communic ations Sector (ITU-R). ITU-R was formerly known as the Consultative Committee f or International Radio communications (CCIR). TG 8/1 is considering how worldwi de wireless networks should evolve and how worldwide frequency coordination migh t be implemented to allow subscriber unites to work anywhere in the world. FPLM TS (now IMT-2000) is a third generation universal, multi-function, globally comp atible digital mobile radio system that will integrate paging, cordless, and cel lular systems, as well as low earth orbit (LEO) satellites, into one universal m obile system. A total of 230 MHz in frequency bands 1885 to 2025 MHz was target ed by the ITUs 1992 World Administrative Radio Conference (WARC). In March 1999, ITU-R agreed to additional spectrum allocations that include the frequency band

s 806 to 960 MHz, 1710 to 2200 MHz, and 2520 to 2670 MHz. This additional spect rum allocation was approved in May 2000 at the ITU World Radio Conference (WRC-2 000). The types of modulation, speech coding, and multiple access schemes to be used in IMT-2000 were also locked down by the ITU Radio General Assembly in mid -2000. The selected radio interface for terrestrial wireless service include exp ansions of todays GSM and IS-95 CDMA, as well as new time-code CDMA standard by C hina. Worldwide standards are also required for low earth orbit (LEO) satellite commun ication system that were developed in the 1990s but which failed commercially at the turn of the century. Due to the very large areas on earth which are illumi nated by satellite transmitters, satellite based cellular systems will never app roach the capacities provided by land-based microcellular systems. However, sat ellite mobile systems were touted as offering tremendous promise for paging, dat a collection, and emergency communications, as well as for global roaming. In e arly 1990, the aerospace industry demonstrated the first successful launch of a small satellite on a rocket from a jet aircraft. This launch technique was pion eered by orbital Sciences Corp. and was more than an order of magnitude less ex pensive than conventional ground-based launches and allowed rapid deployment, su ggesting that a network of LEOs could be rapidly launched for wireless communica tions around the globe. While several companies, such as Motorolas Iridium, prop osed systems and service concepts for worldwide paging, cellular telephone, and emergency navigation and notification in the early 1990s [IEE91], the capital ma rkets have not supported mobile satellite systems in general. In emerging nations, where existing telephone service is almost nonexist ent, fixed cellular telephone systems are being installed at a rapid rate. This is due to the fact that developing nations are finding it is quicker and more a ffordable to install cellular telephone systems for fixed home use, rather than install wires in neighborhoods which have not yet received telephone connections to the PSTN. The world is undergoing a major telecommunications revolution that will provide ubiquitous communication access to citizens, wherever they are. The wir eless telecommunications industry requires engineers who can design and develop new wireless systems, make meaningful comparisons of competing systems, and unde rstand the engineering trade-offs that must be made in any system. Such underst anding can only be achieved by mastering the fundamental technical concepts of w ireless personal communications. 7. Explain the concept of Frequency reuse. Frequency Reuse Cellular radio systems rely on an intelligent allocation and reuse of ch annels throughout a coverage region. Each cellular base station is allocated a group of radio channels to be used within a small geographic area called a cell. Base stations in adjacent cells are assigned channel groups which contain comp letely difference channels than neighboring cells. The base station antennas ar ea to design to achieve the desired coverage within the particular cell. By lim iting the coverage area to within the boundaries of a cell, the same group of ch annels may be used to cover different cells that are separated from one another by distances large enough to keep interference levels within tolerable limits. The design process of selecting and allocating channel group for all of the cell ular base stations within a system is called frequency reuse of frequency planni ng. Figure illustrates the concept of cellular frequency reuse, where cells labeled with the same letter use the same group of channels. The frequency reus e plan is overlaid upon a map to indicate where different frequency channels are

used. The hexagonal cell shape shown in Figure is conceptual and is a simplist ic model of the radio coverage for each base station, but it has been universall y adopted since the hexagon permits easy and manageable analysis of a cellular s ystem. The actual radio coverage of a cell is known as the footprint and is det ermined from field measurements or propagation predication models. Although the real footprint is amorphous in nature, a regular cell shape is needed for syste matic system design and adaptation for future growth. While it might seem natur al to choose a circle to represent the coverage area of a base station, adjacent circles cannot be overlaid upon a map without leaving gaps or creating overlapp ing regions. Thus, when considering geometric shapes which cover an entire regi on without overlap and with equal area, there are three sensible choices-a squar e, an equilateral triangle, and a hexagon. A cell must be designed to serve the weakest mobiles within the footprint, and these are typically located at the ed ge of the cell. For a given distance between the center of a polygon and its fa rthest perimeter points, the hexagon has the largest area of the three. Thus, b y using the hexagon geometry, the fewest number of cells can cover a geographic region, and the hexagon closely approximates a circular radiation pattern which would occur for an omni directional base Figure: Illustration of the cellular frequency reuse concept. Cells with the sa me letter use the same set of frequencies. A cell cluster is outlined in bold a nd replicated over the coverage area. In this example, the cluster size, N, is equal to seven, and the frequency reuse factor is 1/7 since each cell contains o ne-seventh of the total number of available channels Station antenna and free space propagation. Of course, the actual cellu lar footprint is determined by the contour in which a given transmitter serves t he mobiles successfully. When using hexagons to model coverage areas, base station transmitters a re depicted as either being in the center of the cell (center-excited cells) or on three of the six cell vertices (edge-excited cells). Normally, omni directio nal antennas are used in center-excited cells and sectored directional antennas are used in corner-excited cells. Practical considerations usually do not allow base stations to be placed exactly as they appear in the hexagonal layout. Mos t system designs permit a base station to be positioned up to one-fourth the cel l radius away the ideal location. To understand the frequency reuse concept, consider a cellular system wh ich has a total of S duplex channels available for use. If each cell is allocat ed a group of k channels (k<S), and if the S channels are divided among N cells into unique and disjoint channel groups which each have the same number of chann els, the total number of available radio can be expressed as S = kN (1)

The N cells which collectively use the complete set of available frequen cies is called cluster. If a cluster is replicated M times within the system, t he total number of duplex channels, C, can be used as a measure of capacity and is given by C = MkN = MS (2)

As seen from Equation (2), the capacity of a cellular system is directly proportional to the number of times a cluster is replicated in a fixed service area. The factor N is called the cluster size and is typically equal to 4, 7, o r 12. If the cluster size N is reduced while the cell size is kept constant, mo re clusters are required to cover a given area, and hence more capacity (a large r value of C) is achieved. A larger cluster size causes the ratio between the c

ell radius and the distance between co-channel cells to decrease, leading to wea ker co-channel interference. Conversely, a small cluster size indicate that co-c hannel cells are located much closer together. The value for N is a function of how much interference a mobile or base station can tolerate while maintaining a sufficient quality of communications. From a design viewpoint, the smallest p ossible value of N is desirable in order to maximize capacity over a given cover age area (i.e., to maximize C in Equation (2)). The frequency reuse factor of a cellular system is given by 1/N, since each cell within a cluster is only assig ned 1/N of the total available channels in the system. Due to the fact that the hexagonal geometry of Figure has exactly six eq uidistant neighbors and that the lines joining the centers of any cell and each of its neighbors are separated by multiples of 60 degrees, there are only certai n cluster sizes and cell layouts which are possible. In order to tessellate-to connect without gaps between adjacent cells-the geometry of hexagons is such tha t the number of cells per cluster, N, can only have values which satisfy Equatio n (3). (3) Where I rs of a hain of . This and j are non-negative integers. To find the nearest co-channel neighbo particular cell, one must do the following: (1) move i cells along any c hexagons and then (2) turn 60 degrees counter-clockwise and move j cells is illustrated figure for i=3 and j=2 (example, N = 19).

Figure: Method of locating co-channel cells in a cellular system. In this exam ple, N = 19 (i.e, 3, j=2). 8. If a total of 33 MHz of bandwidth is allocated to a particular FDD cellular t elephone system which uses two 25kHz simplex channels to provide full duplex voi ce and control channels, compute the Number of channels available per cell if a system user a) four cell reuse b) seven-cell reuse c) 12-cell reuse if 1MHz of t he allocated spectrum is dedicated to control channels, determine an equitable d istribution of control channels and voice channels in each cell for each of the three systems. Solution Given: Total bandwidth = 33 MHz Channel bandwidth = 25 kHz x 2 simplex channels = 50 kHz/duplex channel Total available channels = 33,000/50 = 660 channels (a) For N = 4, Total number of channels available per cell = 660/4 = 165 channels. (b) For N = 7, Total number of channels available per cell = 660/7 = 95 channels (b) For N = 12, Total number of channels available per cell = 660/12 =55 channels. A 1 MHz spectrum for control channels implies that there are 1000/50 = 2 0 control channels out of the 660 channels available. To evenly distribute the control and voice channels, simply allocate the same number of voice channels in each cell wherever possible. Here, the 660 channels must be evenly distributed to each cell within the cluster. In practice, only the 640 voice channels woul d be allocated, since the control channels are allocated separately as 1 per cel l. (a) For N = 4, we can have five control channels and 160 voice channels per cel l. In practice, however, each cell only needs a single control channel (the con

trol channels have a greater reuse distance than the voice channels). Thus, one control channel and 160 voice channels would be assigned to each cell. (b) For N = 7, four cells with three control channels and 92 voice channels, tow cells with three control channels and 90 voice channels, and one cell with two control channels and 92 voice channels could be allocated. In practice, however , each cell would have one control channel, four cells would have 91 voice chann els, and three cells would have 92 voice channels (c) For N = 12, we can have eight cells with two control channels and 53 voice c hannels, and four cells with one control channel and 54 voice channels each. In an actual system, each cell would have one control channel, eight cells would h ave 53 voice channels, and four cells would have 54 channels. 9. Explain different types of channel Assignment strategies. Channel Assignment Strategies For efficient utilization of the radio spectrum, a frequency reuse schem e that is consistent with the objectives of increasing capacity and minimizing i nterference is required. A variety of channel assignment strategies have been d eveloped to achieve these objectives. Channel assignment strategies can be clas sified as either fixed or dynamic. The choice of channel assignment strategy im pacts the performance of the system, particularly as to how calls are managed wh en a mobile user is handed off from one cell to another. In a fixed channel assignment strategy, each cell is allocated a predete rmined set of voice channels. Any call attempt within the cell can only be serv ed by the unused channels in that particular cell. If all the channels in that cell are occupied, the call is blocked and the subscriber does not receive servi ce. Several variations of the fixed assignment strategy exist. In one approach , called the borrowing strategy, a cell is allowed to borrow channels from a nei ghboring cell if all of its own channels are already occupied. The mobile switc hing center (MSC) supervises such borrowing procedures and ensures that the borr owing of a channel does not disrupt or interfere with any of the calls in progre ss in the donor cell. In a dynamic channels assignment strategy, voice channels are not alloca ted to different cells permanently. In stead, each time a call request is made, the serving base station requests a channel from the MSC. The switch then allo cates a channels to the requested cell, the frequency of use of the candidate ch annel, the reuse distance of the channel, and other cost functions. Accordingly, the MSC only allocates a given frequency if that frequency is not presently in use in the cell or any other cell which falls within the min imum restricted distance of frequency reuse to avoid co-channel interference. D ynamic channels assignment reduce the likelihood of blocking, which increases th e trucking capacity of the system, since all the available channels in market ar e accessible to al of the cells. Dynamic channel assignment strategies require the MSC to collect real-time data on channel occupancy, traffic distribution, an d radio signal strength indications (RSSI) of all channels on a continuous basis . This increases the storage and computational load on the system but provides the advantage of increased channel utilization and decreased probability of a bl ocked call. 11. Explain Interference and System capacity. Interference and System Capacity Interference is the major limiting factor in the performance of cellular

radio systems. Sources of interference include another mobile in the same cell , a call inn progress in a neighboring cell, other base stations operating in th e same frequency band, or any no cellular system which inadvertently leaks energ y into the cellular frequency band. Interference on voice channels causes cross talk, where the subscriber hears interference in the background due to an undes ired transmission. On control channels, interference leads to missed and blocke d calls due to errors in the digital signaling. Interference is more severe in urban areas, due to the greater RF noise floor and the large number of base stat ions and mobiles. Interference has been recognized as a major bottleneck in inc reasing capacity and is often responsible for dropped calls. The two major type s of system-generated cellular interference are co-channel interference and adja cent channel interference. Even though interfering signals are often generated within the cellular system, they are difficult to control in practical (due to r andom propagation effects). Even more difficult to control is interference due to out-of-band users, which arises without warning due to front end overload of subscriber equipment or intermittent intermodulation products. In practice, the transmitters from competing cellular carriers are often a significant source of out-of-band interference, since competitors often locate their base stations in close proximity to one another in order to provide comparable coverage to custo mers. Co-channel Interference and System Capacity Frequency reuse implies that in a given coverage area there are several cells that use the same set of frequencies. These cells are called co-channel c ells, and the interference between signals from these cells is called co-channel interference. Unlike thermal noise which can be overcome by increasing the sig nal-to-noise ratio (SNR), co-channel interference cannot be combated by, simply increasing the carrier power of a transmitter. This is because an increase in c arrier transmit power increase, co-channel cells must be physically separated by a minimum distance to provide sufficient isolation due to propagation. When the size of each cell is approximately the same and the base statio ns transmit the same power, the co-channel interference ratio is independent of the transmitted power and becomes a function of the radius of the cell (R) and t he distance between centers of the nearest co-channel cells (D). By increasing the ratio of D/R, the spatial separation between co-channel cells relative to th e coverage distance of a cell is increased. Thus, interference is reduced from improved isolation of RF energy from the co-channel cell. The parameter Q, calle d the co-channel reuse ratio, is related to the cluster size, for a hexagonal ge ometry (1) A small value of Q provides larger capacity since the cluster size N is small, whereas a large value of Q improves the transmission quality, due to a sm aller level of co-channel interference. A trade-off must be made between these two objectives in actual cellular design.

i i i I

= = = =

1, 1, 0, 2,

Cluster j = 1 j = 2 j = 3 j = 2

Size (N) 3 3 7 4.58 9 5.20 12 6

Co-channel Reuse Ratio (Q)

Let i0 be the number of co-channel interfering cells. Then, the signalto-interference ratio (S/I or SIR) for a mobile receiver which monitors a forwar

d channel can be expressed as (2) where S is the desired signal power from the desired base station and Ii is the interference power caused by the ith interfering co-channel cell base station. If the signal levels of co-channel cells are known, then the S!I ratio for the f orward link can be found using Equation(2). Propagation measurements inn a mobile radio channel show that the averag e received signal strength at any point decays as a power law of the distance of separation between a transmitter and receiver. The average received power Pr a t a distance d from the transmitting antenna is approximated by (3) or (4) where P0 is the power received at a close-in reference point in the far field re gion of the antenna at a small distance d0 from the transmitting antenna and n i s the path loss exponent. Now consider the forward link where the desired signa l is the serving baser station and where the interference is due to co-channel b ase stations. If Di is the distance of the ith interferer from mobile, the rece ived power at a given mobile due to the ith interfering cell will be proportiona l to (Di)-n. The path loss exponent typically ranges between two and four in ur ban cellular systems [Rap92b]. When the transmit power of each base station is equal and the path loss exponent is the same throughout the coverage area. S/I for a mobile can be appr oximate as (5) Considering only the first layer of interfering cells, if all the interf ering base stations are equidistant from the desired base station and if this di stance is equal to the distance D between cell centers, then Equation (5) simpli fies to (6) Equation (6) relates S/I to the cluster size N, which in turn determines the overall capacity of the system from Equation (2). For example, assume that the six closet cells are close enough to create significant interference and th at they are all approximately equidistant from the desired base station. For th e U.S. AMPS cellular system which uses FM and 30 kHz channels, subjective tests indicate that sufficient voice quality is provided when S/I is greater than or e qual to 18 dB. Using Equation (6), it can be shown in order to meet this requir ement, the requirement, the cluster size N should be at least 6.49, assuming a p ath loss exponent n=4. Thus a minimum cluster size of seven is required to meet an S/I requirement of 18 dB. It should be noted that Equation (6) is based on the hexagonal cell geometry where all the interfering cells are equidistant from the base station receiver, and hence provides an optimistic result in many case s. For some frequency reuse plans (e.g., N=4), the closest interfering cells va ry widely in their distance from the desired cell. Using an exact cell geometry layout, it can be shown for a seven-cell cl uster, with the mobile unit at the cell boundary, the mobile is a distance D-R f rom the two nearest co-channel interfering cells and is exactly D+R/2, D, D-R/2, and D+R from the other interfering cells inn the first tier, as shown rigorousl

y in [lee86]. Using the approximate geometry shown in Figure, Equation), and as suming n=4, the signal-to-interference ratio for the worst case can be closely a pproximated as (an exact expression is worked out by Jacobs Meyer).

Figure: Illustration of the first tier of co-channel cells for a cluster size of N = 7. An approximation of the exact geometry is shown here, whereas the exact geometry is given in [Lee86]. When the mobile is at cell boundary (point X), i t experiences worst case co-channel interference on the forward channel. The ma rked distances between the mobile and different co-channel cells are based on ap proximations made for easy analysis. For N = 7, the co-channel reuse ratio Q is 46, and the worst case S/I is approximated as 49.56 (17 dB) using Equation (3.11), whereas an exact solution using Equation yields 17.8 dB [Jac94]. Hence for a seven-cell cluster, the S/I ratio is slightly less than 18 dB for the worst case. To design the cellular sy stem for proper performance in the worst case, it would be necessary to increase N to the next largest size, which from Equation is found to be 9 (corresponding to i=0, j=3). This obviously entails a significant decrease in capacity, since 9-cell reuse offers a spectrum utilization of 1/9 within each cell, whereas sev en-cell reuse offers a spectrum utilization of 1/7. In practice, a capacity red uction of 7/9 would not be tolerable to accommodate for the worst case situation which rarely occurs. From the above discussion, it is clear that co-channel in terference determines link performance, which in turn dictates the frequency reu se plan and the overall capacity of cellular systems. Channel Planning for wireless Systems Judiciously assigning the appropriate radio channels to each base statio n is an important process that is much more difficult in practice than in theory . While Equation is a valuable rule of thumb for determining the appropriate fr equency reuse ratio (or cluster size) and the appropriate separation between adj acent co-channel cells, the wireless engineer must deal with the real-world diff iculties of radio propagation and imperfect coverage regions of each cell. Cell ular systems, in practice, seldom obey the homogenous propagation path loss assu mption of Equation. Generally, the available mobile radio spectrum is divided into channels, which are part of an air interface standard that is used throughout a country o r continent. These channels generally are made up of control channels (Vital fo

r initiating, requesting, or paging a call), and voice channels (dedicated to ca rrying revenue-generating traffic). Typically, about 5% of the entire mobile sp ectrum is devoted to control channels, which carry data messages that are very b rief and bursty in nature, while the remaining 95% of the spectrum is dedicated to voice channels. Channels may be assigned by the wireless carrier in any mann er it chooses, since each market may have its own particular propagation conditi ons or particular services it wishes to offer and may wish to adopt its own part icular frequency reuse scheme that fits its geographic conditions or air interfa ce technology choice. However, in practical systems, the air interface standard ensures a distinction between voice and control channels, and thus control chan nels are generally not allowed to be used as voice channels, and vice versa. Fu rthermore, since control channels are vital in the successful launch of any call , the frequency reuse strategy applied to control channels is different and gene rally more conservative (e.g., is afforded greater S/I protection) than for the voice channels. Where the control channels are allocated using 21-cell reuse, wh ereas voice channels are assigned using seven-cell reuse. Typically, the contro l channels are able to handle a great deal of data such that only one control ch annel is needed within a cell. As described in, sectoring is often used to impr ove the signal-to-interference ratio which may lead to a smaller cluster size, a nd in such cases, only a single control channel is assigned to an individual sec tor of a cell. One of the key features of CDMA systems is that the cluster size is N = 1, and frequency planning is not nearly as difficult as for TDMA or firs t generation cellular systems [Lib99]. Still, however, propagation consideratio ns require most practical CDMA systems to use some sort of limited frequency reu se where propagation conditions are particularly ill-behaved in a particular mar ket. For example, in the vicinity of bodies of water, interfering cells on the same channel as the desired serving cell can create interference overload that e xceed that exceeds the dynamic range of CDMA power control capabilities, leading to dropped calls. In such instances, the most popular approach is to use what is called f1/f2 cell planning, where nearest neighbor cells use radio channels t hat are different from its closest neighbor in particular locations. Such frequ ency planning requires CDMA phones to make hard handoffs, just as TDMA and FDMA phones do. In CDMA, a single 1.25 MHz radio channel carries the simultaneous transm issions of the single control channel with up to 64 simultaneous voice channels. Thus, unlike in 30 kHz IS-136 or 200 kHz GSM TDMA systems, where the coverage region and interference levels are well defined when specific radio channels are in use, the CDMA system instead has a dynamic, time varying coverage region whi ch varies depending on the instantaneous number of users on the CDMA radio chann el. This effect, known as a breathing cell, requires the wireless engineer to c arefully plan the coverage and signal levels for the best and worst cases for se rving cells as well as nearest neighbor cells, from both a converge and interfer ence standpoint. The breathing cell phenomenon can lead to abrupt dropped calls resulting from abrupt coverage changes simply due to an increase in the number of users on a serving CDMA base station. Thus, instead of having to make carefu l decisions about the channel assignment schemes for each cellular base station, CDMA engineers must instead make difficult decisions about the power levels and thresholds assigned to control channels, voice channels, and how these levels a nd thresholds should be adjusted for changing traffic intensity. Also, threshol d levels for CDMA handoffs, inn both the soft handoff case and hard handoff case , must be planed and often measured carefully before turning up service. In fac t, the f1/f2 cell planning issue led to the development of TSB-74, which added h ard-handoff capabilities between different CDMA radio channels to the original I S-95 CDMA specification described.

Adjacent Channel Interference Interference resulting from signals which are adjacent in frequency to the desired signal is called adjacent channel interference. Adjacent channel in terference results from imperfect receiver filters which allow nearby frequencie s to leak into the pass band. The problem can be particularly serious if an adj acent channel user is transmitting in very close range to a subscribers receiver, while the receiver attempts to receive a base station on the desired channel. This is referred to as the near-far effect, where a nearby transmitter (which ma y or may not be of the same type as that used by the cellular system) captures t he receiver of the subscriber. Alternatively, the near-far effect occurs when a mobile close to a base station transmits on a channel close to one being used b y a weak mobile. The base station may have difficulty in discriminating the des ired mobile user from the bleed over caused by the close adjacent channel mobile. Adjacent channel interference can be minimized through careful filtering and channel assignments. Since each cell is given only a fraction of the avail able channels, a cell need not be assigned channels which are all adjacent in fr equency. By keeping the frequency separation between each channel in a given ce ll as large as possible, the adjacent channel interference may be reduced consid erably. Thus instead of assigning channels which form a contiguous band of freq uencies within a particular cell, channels are allocated such that the frequency separation between channels in a given cell is maximized. By sequentially assi gning successive channels in the frequency band to different cells, many channel allocation schemes are able to separate adjacent channels in a cell by as many as N channel bandwidths, where N is cluster size. Some channel allocation schem es also prevent a secondary source of adjacent channel interference by avoiding the use of adjacent channels in neighboring cell sites. If the frequency reuse factor is large (e.g., small N), the separation b etween adjacent channels at the base station may not be sufficient to keep the a djacent channel interference level within tolerable limits. For example, if a c lose-in mobile is 20 times as close to the base station as another mobile and ha s energy spill out of its pass band, the signal-to-interference ratio at the at the base station for the weak mobile (before receiver filtering) is approximatel y For a path loss exponent n=4, this is equal to-52 dB. If the intermedia te frequency (IF) filter of the base station receiver has a slope of 20 dB/octav e, then an adjacent channel interferer must be displaced by at least six times t he pass band bandwidth from the center of the receiver frequency pass band to ac hieve 52 dB attenuation. Here, a separation of approximately six channel bandwi dths is required for typical filters in order to provide 0 dB SIR from a close-i n adjacent channel user. This implied more than six channel separations are nee ded to bring the adjacent channel interference to an acceptable level. Tight ba se station filters are needed when close-in and distant users share the same cel l. In practice, base station receivers are preceded by a high Q cavity filter i n order to reject adjacent channel interference. Power Control for Reducing Interference In practical cellular radio and personal communication systems, the powe r levels transmitted by every subscriber unit are under constant control by the serving base stations. This is done to ensure that each mobile transmits the sm allest power necessary to maintain a good quality link on the reverse channel. Power control not only helps prolong battery life for the subscriber unit, but a lso dramatically reduces the reverse channel S/I in the system. As shown in Cha pters 9 and 11, power control is especially important for emerging CDMA spread s pectrum systems that allow every user in every cell to share the same radio chan

nel. 12. Explain about Trunking and Grade of service. Trucking and Grade of Service The concept of trunking allow a large number of users to share the relat ively small number of channels in a cell by providing access to each user, on de mand, from a pool of available channels. In a trunked radio system, each user is allocated a channel on a per call basis, and upon termination of the call, th e previously occupied channel is immediately returned to the pool of available c hannels. Trunking exploits the statistical behavior of users so that a fixed number of ch annels or circuits may accommodate a large, random user community. The telephone company uses trunking theory to determine the number of telephone circuits that need to be allocated for office buildings with hundreds of telephones, and this same principle is used in designing cellular radio systems. There is a trade-of f between the number of available telephone circuits and the likelihood of a par ticular user finding that no circuits are available during the peak calling time . As the number of phone lines decreases, it becomes more likely that all circui ts will be busy for a particular user. In a trunked mobile radio system, when a particular user requests service and all of the radio channels are already in us e, the user is blocked, or denied access to the system. In some systems, a queue may be used to hold the requesting users a channel becomes available. To design trunked radio systems that can handle a specific capacity at s pecific grade of service, it is essential to understand trunking theory and queuin g theory. The fundamentals of trunking theory were developed by Erlang, One Erl ang represents the amount of traffic intensity carried by a channel that is comp letely occupied (i.e. one call-hour per hour or one call-minute per minute ). F or example, a radio channel that is occupied for thirty minutes during an hour c arries 0.5 Erlangs of traffic. The grade of service (GOS) is a measure of the ability of a user to acce ss a trunked system during the busiest hour. The busy hour is based upon custom er demand at the busiest hour during a week, month, or year. The busy hours for cellular radio systems typically occur during rush hours, between 4 p.m. on a T hursday or Friday evening. The grade of service is a benchmark used to define t he desired performance of a particular trunked system by specifying a desired li kelihood of a user obtaining channel access given a specific number of channels available in the system. It is the wireless designers job to estimate the maximu m required capacity and to allocate the proper number of channels in order to me et the GOS. GOS is typically given as the likelihood that a call is blocked, or the likelihood of a call experiencing a delay greater than a certain queuing ti me. A number of definitions listed inn Table are used in trunking theory to make capacity estimates in trunked systems. The traffic intensity offered by each user is equal to the call request rate multiplied by the holding time. That is, each user generates a traffic int ensity of A Erlangs given by A = H (1)

Where H is the average duration of a call and is the average number of call requ ests per unit time for each user. For a system containing U users and an unspec ified number of channels, the total offered traffic intensity A, is given as

A = UA

(2)

Furthermore, in a C channel trunked system, if the traffic is equally distribute d among the channels, then the traffic intensity per channel. Ac, is given as Ac = UA /C (3)

Note that the offered traffic is not necessarily the traffic which is ca rried by the trunked system, only that which is offered to the trunked system. When the offered traffic exceeds the maximum capacity of the system, the carried traffic becomes limited due to the limited capacity (i.e. limited number of cha nnels). The maximum possible carried traffic is the total number of channels, C , in Erlangs. The AMPS cellular system is designed for a GOS of 2% blocking. T his implies that the channel allocations for cell sites are designed so that 2 o ut of 100 calls will be blocked due to channel occupancy during the busies hour. There are two types of trunked systems which are commonly used. The fir st type offers no queuing for call requests. That is, for every user who reques ts service, it is assumed there is no setup time and the user is given immediate access to blocked without access and is free to try again later. This type of trunking is called blocked calls cleared and assumes that calls arrive as determ ined by a Poisson distribution. Furthermore, it is assumed that there are an in finite number of users as well the following: (a) there are memory less arrivals of requests, implying that all users, including blocked users, may request at a channel at any time; (b) the probability of a user occupying a channel is expon entially distributed, (c) there are a finite number or channels available in the trunking pool. This is known as an M/M/m/m queue, and leads to the derivation of the Erlang B formula (also known as the blocked calls cleared formula). The Erlang B formula determines the probability that a call is blocked and is a meas ure of the GOS for a trunked system which provides no queuing for blocked calls. The Erlang B formula is derived in Appendix A and is given by (4) Where C is the number of trunked channels offered by a trunked radio system and A is the total offered traffic. While it is possible to model trunked systems w ith finite users, the resulting expression are much more complicated than the Er lang B result, and added complexity is not warranted for typical trunked systems which have users that outnumber available channels by orders of magnitude. Fur thermore, the Erlang B formula provides conservative estimate of the GOS, as the finite user results always predict a smaller likelihood of blocking. The capac ity of a trunked radio system where blocked calls are lost is tabulated for vari ous values of GOS and numbers of channels in Table. Table: Capacity of an Erlang B system Number of Channels C = 0.01 =0.005 2 0.153 0.105 4 0.869 0.701 5 1.36 1.13 10 4.46 3.96 20 12.0 11.1 24 15.3 14.2 40 29.0 27.3 70 56.1 53.7 100 84.1 80.9 Capacity (Erlangs) for GOS =0.002 =0.001 0.065 0.046 0.535 0.439 0.900 0.762 3.43 3.09 10.1 9.41 13.0 12.2 25.7 24.5 51.0 49.2 77.4 75.2

The second kind of trunked system is one in which a queue is provided to

hold calls which are blocked. If a channel is not available immediately, the c all request may be delayed until a channel becomes available. This type of trun king is called Blocked Calls Delayed, and its measure of GOS is defined as the p robability that a call is blocked after waiting a specific length of time in the queue. To find the GOS, it is first necessary to find the likelihood that a ca ll is initially denied access to the system. The likelihood of a call not havin g immediate access to a channel is determined by the Erlang C formula derived in n Appendix A (5) If no channels are immediately available the call is delayed, and the pr obability that the delayed call is forced to wait more than t seconds is given b y the probability that a call is delayed, multiplied by the conditional probabil ity that the delay is greater than t seconds. The GOS of a trunked system where blocked calls are delayed is hence given by Pr[delay > t] = Pr[delay > 0] Pr[delay > t] [delay > 0] = Pr[delay > 0] exp(-(C-A) t/H) The average delay D for all calls in queued system is given by D = Pr [delay > 0] Where the average delay those calls which are queued is given by H/(C-A). The Erlang B and Erlang C formulas are plotted in graphical form in Figu res These graphs are useful for determining GOS in rapid fashion, although comp uter simulations are often used to determine transient behaviors experienced by particular users in mobile system. 16. Describe the Methods in Improving Coverage and Capacity in Cellular Systems. Improving Coverage and Capacity in Cellular Systems As the demand for wireless service increase, the number of channels assi gned to a cell eventually becomes insufficient to support the required number of users. At this point, cellular design techniques are needed to provide more ch annels per unit coverage area. Techniques such as cell splitting, sectoring, an d coverage zone approaches are used in practice to expand the capacity of cellul ar systems. Cell splitting allows an orderly growth of the cellular system. Se ctoring uses directional antennas to further control the interference and freque ncy reuse of channels. The zone micro cell concept distributes the coverage of a cell and extends the cell boundary to hard-to-reach places. While cell splitt ing increases the number of base stations in order to increase capacity, sectori ng and zone micro cells rely on base station antenna placements to improve capac ity by reducing co-channel interference. Cell splitting and zone micro cell tec hniques do not suffer the trunking inefficiencies experienced by sectored cells, and enable the base station to oversee all handoff chores related to the micro cells, thus reducing the computational load at the MSC. These popular capacity improvement techniques will be explained in detail. Cell Splitting Cell splitting is the process of subdividing a congested cell into smal ler cells, each with its own base station and a corresponding reduction in anten na height and transmitter power. Cell splitting increase the capacity of a cell ular system since it increases the number of times that channels are reused. By defining new cells which have a smaller radius than the original cells and by i (7) (6)

nstalling these smaller cells (called micro cells) between the existing cells, c apacity increases due to the additional number of channels per unit area. Imagine if every cell in Figure were reduced in such a way that the radi us of every cell was cut in half. In order to cover the entire service area wit h smaller cells, approximately four times as many cells would be required. This can be easily shown by considering a circle with radius R. The area covered by such a circle is four times as large as the area covered by a circle with radiu s R/2. The increased number of cells would increase the number of clusters over the coverage region, which is n turn would increase the number of channels, and thus capacity, in the coverage area. Cell splitting allows a system to grow by replacing large cells with smaller cells, while not upsetting the channel alloc ation scheme required to maintain the minimum co-channel reuse ratio Q between c o-channel cells. An example of cell splitting is shown in Figure. In Figure, the base st ations are place at corners of the cells, and the area served by base station A is assumed to be saturated with traffic (i.e., the blocking of base station A ex ceeds acceptable rates). New base stations are therefore needed in the region t o increase the number of channels in the area and to reduce the area served by t he single base station. Note in the figure that the original base station A has been surrounded by six new micro cells. In the example shown in Figure, the sm aller cells were added in such a way as to preserve the frequency reuse plan of the system. For example, the micro cell base station labeled G was place half w ay between two larger stations utilizing the same channel set G. This is also t he case for the other micro cells in the figure. As can be seen from Figure, ce ll splitting merely scales the geometry of the cluster. In this case, the radiu s of each new micro cell is half that of the original cell.

Figure: Illustration of cell splitting For the cells to be smaller in size, the transmit power of these cells m ust be reduced. The transmit power of the new cells with radius half that of th e original cells can be found by examining the received power Pr at the received power Pr at the new and old cell boundaries and setting them equal to each othe r. This is necessary to ensure that the frequency reuse plan for the new micro cells behaves exactly as for the original cells. Pr [at old cell boundary] And Pr [at new cell boundary] Pt2(R/2)-n Pt1 R-n

Where Pt1 and Pt2 are the transmit powers of the larger and smaller cell base st ations, respectively, and n is the path loss exponent. If we take n =4 and set the received powers equal to each other, then

Pt2 = In other words, the transmit power must be reduced by 12 dB inn order to fill in the original coverage area with micro cells, while maintaining the S/I requirement. In practice, not all cells are split at the same time. It is often diff icult for service providers to find real estate that is perfectly situated for c ell splitting. Therefore, different cell sizes will exist simultaneously. In s uch situations, special care needs to be taken to keep the distance between co-c hannel cells at the required minimum, and hence channel assignments become more complicated [Rap97]. Also, handoff issues must be addressed so that high speed and low speed traffic can be simultaneously accommodated (the umbrella cell appr oach of commonly used). When there are two cell sizes in the same region as sho wn in Figure, Equation shows that one cannot simply use the original transmit po wer for all cells, some channels used by the smaller cells would not be sufficie ntly separated from co-channel cells. On the other hand, if the unserved. For this reason, channels in the old cell must be broken down into two channel group s, one that corresponds to the smaller cell reuse requirements and the other tha t corresponds to the larger cell reuse requirements. The larger cell is usually dedicated to high speed traffic so that handoffs occur less frequently. The two channel group sizes depend on the stage on the stage of the spli tting process. At the beginning of the cell splitting process, there will be fe wer channels in the small power groups. However, as demand grows, more channels will be required, and thus the smaller groups will require more channels. This splitting process continues until all the channels in an area are used in the l ower power group, at which point cell splitting is complete within the region, a nd the entire system is rescaled to have a smaller radius per cell. Antenna dow n tilting, which deliberately focuses radiated energy from the base station towa rd the ground (rather than toward the horizon), is often used to limit the radio coverage of newly formed micro cells. Figure: Illustration of cell splitting within a 3km square centered around base station A. Sectoring As shown in cell splitting achieves capacity improvement by essentially rescaling the system. By decreasing the cell radius R and keeping the co-channe l reuse ratio D/R unchanged, cell splitting increases the number of channels per unit area. However, another way to increase capacity is to keep the cell radius unc hanged and seek methods to decrease the D/R ratio. As we now show, sectoring in creases SIR so that the cluster size may be reduced. In this approach, first th e SIR is improved using directional antennas, then capacity improvement is achie ved by reducing the number of cells in a cluster, thus increasing the frequency reuse. However, in order to do this successfully, it is necessary to reduce the relative interference without decreasing the transmit power. The co-channel interference in a cellular system may be decreased by rep lacing a single omni directional antenna at the base station by several directio nal antennas, each radiating within a specified sector. By using directional an tennas, a given cell will receive interference and transmit with only a fraction of the available co-channel cells. The technique for decreasing co-channel int erference and thus increasing system performance by using directional antennas i s called sectoring. The factor by which the co-channel interference is reduced depends on the amount of sectoring used. A cell is normally partitioned into th

ree 120

sector or six 60

sectors as shown in Figure (a) and (b).

Figure: (a) 120 sectoring; (b) sectoring When sectoring is employed, the channels used in a particular cell are b roken down into sectored groups and are used only within a particular sector, as in Figure (a) and (b). Assuming seven-cell reuse, for the case of 120 sectors, the number of interferers in the first tier is reduced from six to two. This is because only two of the six co-channel cells receive interference with a partic ular sectored channel group. Referring to Figure, consider the interference exp erienced by a mobile located in the right-most sector in the center cell labeled 5. There are three co-channel cell sectors labeled 5 to the right of the center ce ll, and three to the left of the center cell. Out of these six co-channel cells , only two cells have sectors with antenna patterns which radiate into the cente r cell, and hence a mobile in the center cell will experience interference on th e forward link from only these two sectors. The resulting S/I for this case can be found using Equation to be 24.2 dB, which is a significant improvement over the omni directional case where the worst case S/I was shown to be 17 dB. This S/I improvement allows the wireless engineer to then decrees the cluster size N in order to improve the frequency reuse, and thus the system capacity. In pract ical system, further improvement in S/I is achieved by down tilting the sector a ntennas such that the radiation pattern in the vertical (elevation) plane has a notch at the nearest co-channel cell distance. The improvement in S/I implies that with 120 sectoring, the minimum requi red S/I of 18 dB can be easily achieved with seven-cell reuse, as compared to 12 -cell reuse for the worst possible situation in the unescorted case. Thus, sect oring reduces interference, which amounts to an increase in capacity by a factor of 12/7, or 1.714. In practice, the reduction in interference offered by secto ring enable planners to reduce the cluster size N, and provides an additional de gree of freedom in assigning channels. The penalty for improved S/I and the res ulting capacity improvement from the shrinking cluster size is an increased numb er of antennas at each base station, and a decrease in trunking efficiency due t o channel sectoring at the base station. Since sectoring reduces the coverage a rea of a particular group of channels, the number of handoffs increases, as well . Fortunately, many modern base stations support sectorization and allow mobile s to be handed off from sector to sector within the same cell without interventi on from the MSC, so that handoff problem is often not a major concern. It is the loss of traffic due to decreased trunking efficiency that caus es some operators to shy away from the sectoring approach, particularly in dense urban areas where the directional antenna patterns are somewhat ineffective in controlling radio propagation. Because sectoring uses more than one antenna per base station, the available channels in the cell must be subdivided and dedicat ed to a specific antenna. This breaks up the available trunked channel pool int o several smaller pools, and decreases trunking efficiency.

Repeaters for Range Extension Often a wireless operator needs to provide dedicated coverage for hard-t o-reach areas, such as within buildings, or in valleys or tunnels. Radio retran

smitters, known as repeaters, are often used to provide such range extension cap abilities. Repeaters are bidirectional in nature, and simultaneously send signa ls to and receive signals from a serving base station. Repeaters work using ove r-the-air signals, so they may be installed anywhere and are capable of repeatin g an entire cellular or PCS band. Upon receiving signals from a base station fo rward link, the repeater amplifies and reradiates the base station signals to th e specific coverage region. Unfortunately, the received noise and interference is also reradiated by the repeater on both the forward and reverse link, so care must be taken to properly place the repeaters, and to adjust the various forwar d and reverse link amplifier levels and antenna patterns. Repeaters can be easi ly thought of as bidirectional bent pipes that retransmit what has been received. In practice, directional antennas or distributed antenna systems (DAS) are conn ected to the inputs or outputs of repeaters for localized spot coverage, particu larly in tunnels or buildings. By modifying the coverage of a serving cell, an operator is able to dedi cate a certain amount of the base stations traffic for the areas covered by the r epeater. However, the repeater does not add capacity to the system-it simply se rves to reradiate the base station signal into specific locations. Repeaters ar e increasingly being used to provide coverage into and around buildings, where c overage has been traditionally weak [Rap96], [Mor00]. Many carriers have opted to provide in-building wireless penetration by installing micro cells outside of large buildings, and then installing many repeaters with DAS networks within th e buildings. This approach provides immediate coverage into targeted areas, but does not accommodate the increases in capacity that will arise due to increased outdoor and indoor user traffic. Eventually, dedicated base stations within bu ildings will be needed to accommodate the large number of in-building cellular u sers. Determining the proper location for repeaters and distributed antenna sy stems within buildings requires careful planning, particularly due to the fact t hat interference levels are reradiated into the building from the base station a nd from the interior of the building back to the base station. Also, repeats mu st be provisioned to match the available capacity from the serving base station. Fortunately, software products, such as Site Planner [Wir01], allow engineers to rapidly determine the best placements for repeaters and the required DAS netw ork while simultaneously computing the available the available traffic and assoc iated cost of the installation. Site Planner is protected by US Patent 6,317,59 9 and other patents. Using Site Planner, engineers can very quickly determine t he proper provisioning for a particular level of range extension. A micro cell Zone Concept The increased number of handoffs required when sectoring is employed res ults in an increased load on the switching and control link elements of the mobi le system. A solution to this problem was presented by lee. This proposal is ba sed on a micro cell concept for seven cell reuse, as illustrated in Figure. In this scheme, each of the three (or possibly more) zone sites (Represented as Tx/Rx in Figure) are connected to a single base station and shar e the same radio equipment. The zones are connected by coaxial cable, fiber opt ic cable, or microwave link to the base station. Multiple zones and a single ba se station make up a cell. As a mobile travels within the cell, it is served by the zone with the strongest signal. This approach is superior to sectoring sin ce antennas are place at the outer edges of the cell, and any base station chann el may be assigned to any zone by the base station. As a mobile travels from one zone to another within the cell, it retains the same channel. Thus, unlike in sectoring, a handoff is not required at the MSC when the mobile travels between zones within the cell. The base station sim

ply switches the channel to a different zone site. In this way, a given channel is active only in the particular zone in which the mobile is traveling, and hen ce the base station radiation is localized and interference is reduced. The cha nnels are distributed in time and space by all three zones and are also reused i n co-channel cells in the normal fashion. This technique is particularly useful along highways or along urban traffic corridors. The advantage of the zone cell technique is that while the cell maintain s a particulars a coverage radius, the co-channel interference in the cellular s ystem is reduced since a large central base station is replace by several lower powered transmitters (zone transmitters) on the edges of the cell. Decreased co -channel interference improves the signal quality and also leads to an increase in capacity without the degradation in trunking efficiency caused by sectoring. As mentioned earlier, an S/I of 18 dB is typically required for satisfactory sy stem performance in narrowband FM. For a system with N=7, a D/R of 4.6 was show n to achieve this. With respect to the zone micro cell system, since transmissi on at any instant is confined to a particular zone, this implies that a Dz/Rz of 4.6 (where Dz is the minimum distance between active co-channel zones and Rz is the zone radius) can achieve the required link performance. In Figure, let eac h individual hexagon represents a zone, while each group of three hexagons repre sents a cell. The zone radius Rz is approximately equal to one hexagon radius. Now, the capacity of the zone micro cell system is directly related to the dist ance between co-channel cells, and not zones. This distance is represented as D in Figure. For a Dz/Rz value of 4.6, it can be seen from the geometry of Figur e that the value of co-channel reuse ratio, D/R, is equal to three, where R is t he radius of the cell and is equal to twice the length of the hexagon radius. U sing Equation, D/R=3 corresponds to a cluster size of N=3. This reduction in th e cluster size from N=7 to N=3 amounts to a 2.33 times increase inn capacity for a system completely based on the zone micro cell concept. Hence for the same S /I requirement of 18 dB, this system provides a significant increase in capacity over conventional cellular planning.

Figure: The micro cell concept [adapted from [Lee91b] IEEE] By examining Figure and using Equation [Lee91b], the exact worst case S/ I of the zone micro cell system can be estimated to be 20 dB. Thus, in the wors t case, the system provides a margin of 2 dB over the required signal-to-interfe rence ratio while increasing the capacity by 2.33 times over a conventional seve n-cell system using omni directional antennas.

Figure: Define D, Dz, R, and RZ for a micro cell architecture with N=7. The sma ller hexagons form zones and three hexagons (outlined in bold) together form a c ell. Six nearest co-channel cells are shown. No loss in trunking efficiency is experience. Zone cell architectures are being adopted in many cellular and personal communication system 17. Assume each base station uses 60 channels, regardless of cell size. If each original cell has a radius of 1 km and each micro cell has a radius of 0.5km, f ind the number of channels contained in a 3 km by 3km square centered around A u nder the following conditions: (a) without the use of micro cells; (b) when the lettered micro cells as shown in Figure 3.9 are used; and (c) if all the origina l base stations are replaced by micro cells. Assume cells on the edge of the sq uare to be contained within the square. Solution

(a) Without the use of micro cells: A cell radius of 1 km implies that the sides of the larger hexagons are also 1 km in length. To cover 1.5 km (1.5 times the hexagon radius) toward the right, left, top, and bottom of base station A. Thus is shown in Figure. From F igure, we see that this area contains five base stations. Since each base stati on has 60 channels, the total number of channels without cell splitting is equal to 5x60=300 channels. (b) With the use of the micro cells shown in Figure: In Figure, the base station A ore, the base station A is surrounded 5+6=11. Since each base station has will be equal to 11x60=660 channels. hen compared to case (a). is surrounding by six micro cells. Theref by six square area under study is equal to 60 channels, the total number of channels This is a 2.2 times increase in capacity w

(c) If all the base stations are replaced by micro cells: From Figure, we see there are a total of 5+12=17 base stations in the sq uare region under study. Since each base station has 60 channels, the total num ber of channels will be equal to 17x60=1020 channels. This is a 3.4 times incre ase in capacity compared to case to case (a). Theoretically, if call were micro cells having half the radius of the original cell, the capacity increase would approach four. 18. If a total of 33 MHz of bandwidth is allocated to a particular FDD cellular telephone system which uses two 25 kHz simplex channels to provide full duplex v oice and control channels, compute the number of channels available per cell if a system uses (a) four-cell reuse, (b) seven-cell reuse, and (c) 12-cell reuse. If 1 MHz of the allocated spectrum is dedicated to control channels, determine a n equitable distribution of control channels and voice channels in each cell for each of the three systems. Solution:Given: Total bandwidth = 33 MHz Channel bandwidth = Channel bandwidth = 25 kHz 2 simplex channels = 50 kHz/duple x channel Total available channels = 33,000/50 = 660 channels (a) For N = 4, total number of channels available per cell = 660/4 165 channels (b) For N = 7, total number of channels available per cell = 660/7 95 channels. (c) For N = 12, total number of channels available per cell = 660/12 55 channels. A 1 MHz spectrum for control channels implies that there are 1000/50 = 2 0 control channels out of the 660 channels available. To evenly distribute the c ontrol and voice channels, simply allocate the same number of voice channels in each cell wherever possible. Here, the 660 channels must be evenly distributed t o each cell within the cluster. In practice, only the 640 voice channels would b e allocated, since the control channels are allocated separately as 1 per cell. (a) For N = 4, we can have five control channels and 160 voice channels per cell. In practice, however, each cell only needs a single control channel (the c ontrol channels have a greater reuse distance than the voice channels). Thus, on e control channel and 160 voice channels would be assigned to each cell.

(b) For N = 7, four cells with three control channels and 92 voice channels, two cells with three control channels and 90 voice channels, and one cell with two control channels and 92 voice channels could be allocated. In practice, howe ver, each cell would have one control channel, four cells would have 91 voice ch annels, and three cells would have 92 voice channels. (c) For N = 12, we can have eight cells with two control channels and 53 voi ce channels, and four cells with one control channel and 54 voice channels each. In an actual system, each cell would have one control channel, eight cells woul d have 53 voice channels, and four cells would have 54 voice channels.

UNIT II WIRELESS PROPOGATION CHANNELS PART A 1. Define large scale propagation. Propagation models that predict the mean signal strength for an arbitrary transm itter-receiver(T-R) separation distance are useful in estimating the radio cover age area of a transmitter and are called large-scale propagation models. 2. Define small scale propagation. Propagation models that characterize the rapid fluctuations of the received sig nal strength over very short travel distances (a few wavelengths) or short time durations (on the order of seconds) are called small-scale or fading models. 3. Define path loss. The path loss, which represents signal attenuation as a positive quantity measur ed in Db, is defined as the difference (in Db) between the effective transmitter power and the received power, and may or may not include the effect of the ante nna gains. 4. Write the basic propagation mechanisms. i)Reflection ii)Diffraction iii)Scattering 5. Define reflection. Reflection occurs when a propagation electromagnetic wave impinges upon an objec t which has very large dimensions when compare to the wavelength of the propagat ing wave. Reflections occur from the surface of the earth and from buildings and walls. 6. Define diffraction. Diffraction occurs when the radio path between the transmitter and receiver is o

bstructed by a surface that has sharp irregularities(edges). 7. Define scattering. Scattering occurs when the medium through which the wave travels consists of obj ects with dimensions that are small compared to the wavelength, and where the nu mber of obstacles per unit volume is large. 8. Define Brewster Angle. The Brewster abgle at which no reflection occurs in the medium of origin. It occ urs when the incident angle B is such that the reflection coefficient || is equa l to zero. The Brewster angle is given by the value of B which satisfies Sin( B ) = 1 / 1 + 2 9. Write the pathloss exponents for different environments. Environment Path Loss Exponents, n Free Space 2 Urban area cellular radio 2.7 to 3.5 Shadowed urban cellular radio 3 to 5 In building line-of-sight 1.6 to 1.8 Obstructed in building 4 to 6 Obstructed in factories 2 to 3 10. What is mean by log-normal shadowing? The log-normal distribution describes the random shadowing effects which occur o ver a large number of measurement locations which have the same T-R separation, but have different levels of clutter on the propagation path. This phenomenon is referred to as log-normal shadowing.

11. Define Coherence Bandwidth? Coherence bandwidth is a stastical measure of the range of frequencies over whic h the channel can be considered flat(i.e a channel which passes all spectral compo nents with approximately equal gain and linear phase) or Coherence bandwidth is the range of frequencies over which two frequency components have a strong pote ntial for amplitude correlation. 12. What is the relationship between Doppler spread and Coherence time? 1) Delay spread and coherence bandwidth are parameters which desc ribe the time dispersive nature of the channel in a local area. 2) Delay spread and coherence time are parameters which describes the time varying nature of the channel in small scale region. 13.Define Brewster angle? The Brewster angle is the angle at which no reflection occurs in th e medium of origin.It occurs when the incident angle B is such that the reflectio n coefficient II is equal to zero.Brewster angle occurs only for vertical(i.e par allel) polarization. 14.What are the effects of small-scale multipath propogation? 1) Rapid changes in signal strength over a small travel distance or time

interval 2) Random frequency modulation dueto varying Doppler shifts on differe nt multipath signals. 3)Time dispersion caused by multipath propogation delays. 15.What are the parameters for the wideband model at 1900 MHz? Transmitter Antenna Height n1 n Low(3.7 m) 9.31 Medium(8.5 m) 7.67 High(13.3 m) 7.94 1900MHz LOS n2 (dB) 2.18 2.17 2.07 3.29 3.36 4.16 (dB) 1900MHz OBS 8.76 7.88 8.77 2.58 2.56 2.69 communication ?

Reflection, Diffraction, Scattering are the ba ic propagation mechani m u ed in wirele communication y tem

18. Define Sector. A number of propagation model are available to predict path lo over irregular terrain. While all the e model aim to predict ignal trength at a particular receiving point or in a pecific local area called a ector.

For the ca e of non-LOS, the y tem grade the problem into one of four categori e : Single diffraction edge Two diffraction edge Three diffraction edge More than three diffraction edge . 20. Define two diffraction Edge. It i u ed to calculate the lo a ociated with tow diffraction edge . In hort , it i the um of two attenuation . The fir t attenuation i the lo at the e cond diffraction edge cau ed by the fir t diffraction edge with the tran mitter a the ource. The econd attenuation i the lo at the receiver cau ed by the econd diffraction edge with the fir t diffraction edge a the ource.

21. Define three diffraction Edge.

19. What are the type

of problem for the non-line of ight?

(1)Fading effect

due to Multipath Time Delay Spread (i)Flat fading (ii)Frequency Selective fading (2)Fading effect due to Doppler pread (i)Fa t fading (ii)Slow fading

17. What are the type

of mall cale fading?

16.What are the ba ic propagation mechani m

u ed in wirele

The outer diffraction edge mu t contain a ingle diffraction edge in between. T hi i detected by calculating the line between the two outer diffraction edge .

The three mo t important effect are : I. Rapid change in ignal trength over a mall travel di tance or time in terval. II. Random frequency modulation due to varying Doppler hift on different m ultipath ignal III. Time di per ion (echoe ) cau ed by multipath propagation delay . 23. Define fading. Fading i cau ed by interference between two or more ver ion of the tran mitted ignal which arrive at the receiver at lightly different time .

Small- cale Fading factor are (i) Multipath propagation (ii) Speed of the mobile (iii) Speed of urrounding object (iv) The tran mi ion bandwidth of the ignal. 25. Define coherence Bandwidth. Coherence bandwidth i a tati tical mea ure of the range of frequencie over wh ich the channel can be con idered flat. Bc 1 / 50

26. Define Doppler Shift. When a pure inu oidal tone of frequency fc i tran mitted, the received ignal pectrum, called the Doppler pectrum, will have component in the range fc fd t o fc + fd , where fd i the Doppler hift.It given by fd, where

28. Define coherence time.

29. What are the type

of mall- cale fading?

The definition of coherence time implie that two ignal arriving with a time eparation greater than Tc affected differently by the channel.

(ii) Direct pul e mea urement (ii) Spread Spectrum liding correlator mea urement (iii) Swept frequency mea urement .

Small-Scale Multiplath Mea urement

are a follow :

27. What are the cla ified

for mall- cale multi path mea urement ?

Fd = 1 / 2

t =

co

Coherence bandwidth which i channel.

related to the pecific multipath tructure of the

24. What are the factor

of mall cale fading?

22. Write the

mall- cale fading effect .

Small Scale Fading(Ba ed on multipath time delay pread) (i)Flat Fading (ii) Frequency Selective Fading Small-Scale Fading(Ba ed on Doppler pread) (i)Fa t Fading (ii)Slow Fading 30. Define flat fading. The mobile radio channel ha a con tant gain and linear pha e re pon e over a ba ndwidth which i greater than the bandwidth of the tran mitted ignal, then the received ignal will undergo flat fading.

31. Define Frequency electing fading. The channel po e e a con tant-gain and linear pha e re pon e over a bandwidth that i maller than the bandwidth of tran mitted ignal, then the channel crea te frequency elective fading. 32. Define Fa t fading. In a fa t fading channel, the channel impul e re pon e change rapidly within th e ymbol duration. That i , the coherence time of the channel i maller than th e ymbol period of the tran mitted ignal.

In a low fading channel, the channel impul e re pon e change at a rate much l ower than the tran mitted ba eband ignal (t). 34. Define Rayleigh fading. The Rayleigh di tribution i commonly u ed to de cribe the tati tical time vary ing nature of the received envelope of a flat fading ignal, or the envelope of an individual multipath component. 35. Define Ricean Fading.

36. Define average fade duration. Average fade duration i defined a the average period of time for which the rec eived ignal i below a pecified level R. 37. Find the far-field di tance for an antenna with maximum dimen ion of 1m and operating frequency of 900 MHz. Given: Large t dimen ion of antenna, D = 1 m Operating frequency f = 900 MHz 38. Define Brew ter Angle.

The Brew ter Angle i

the angle at which no reflection occur in the med

Random multipath component arriving at different angle are tationary dominant ignal.

uperimpo ed on a

33. Define

low Fading.

39. Calculate the Brew ter angle for a wave impinging on ground having permittiv ity of r= 4. Solution:Given: r = 4

PART B 1. Explain about Free pare propagation Model. Free Space propagation Model

RCS dT & dR r.

PT GT

Power Tran mitted Gain of Tran mitting Antenna Wavelength Radar Cro ection Di tance from the cattering object to tran mitter and receive

40. Give the expre ion power of receiver for Radar cro

ium of origin. It occur when the incident angle efficient i equal to zero. The Brew ter angle i given by . 1 = Fir t Medium 2 = Second Medium

B i

uch that the reflection co

ection Model.

The free pace propagation model i u ed to predict received ignal trength whe n the tran mitter and receiver have a clear, unob tructed line-of- ight path bet ween them, Satellite communication y tem and microwave line-of- ight radio lin k typically undergo free pace propagation. A with mo t large- cale radio wav e propagation model , the free pace model predict that received power decay a a function of the T-R eparation di tance rai ed to ome power (i.e. a power l aw function). The free pace power received by a receiver antenna which i epa rated from a radiating tran mitter antenna by a di tance d, i given by the Frii free pace equation. (1) where Pt i the tran mitted power, Pr(d) i the received power which i a functi on of the T-R eparation, Gt i the tran mitter antenna gain, Gr i the receiver antenna gain, d i the T-R eparation di tance in meter , L i the y tem lo not related to propagation (l 1), and i the wavelength in meter . The gain of an antenna i related to it effective aperture, Ae , by (2)

where f i the carrier frequency in Hertz, c i the carrier frequency in radian per econd, and c i the peed of light given in meter / . The value for Pt mu t be expre ed in the ame unit , and Gt and Gr are dimen ionle quantitie . The mi cellaneou lo e L (L 1) are u ually due to tran mi ion line attenuatio n, filter lo e , and antenna lo e in the communication y tem. A value of L = 1 indicate no lo in the y tem hardware. The Frii free pace equation of (1) how that the received power fall off a the quare of the T-R eparation di tance. Thi implie that the receiv ed power decay with di tance at a rate of 20 dB/decade An i otropic radiator i an ideal antenna which radiate power with unit gain uniformly in all direction , and i often u ed to reference antenna gain in wirele y tem . The effective i otropic radiated power (EIRP) i defined a EIRP = Pt Gt (4)

And repre ent the maximum radiated power available from a tran mitter in the di rection of maximum antenna gain, a compared to an i otropic radiator. In practice, effective radiated power (ERP) i u ed in tead of EIRP to d enote the maximum radiated power a compared to a half-wave dipole antenna (in t ead of an i otropic antenna). Since a dipole antenna ha a gain of 1.64 (2.15 d B above an i otrope), the ERP will be 2.15 dB maller than the EIRP for the ame tran mi ion y tem. In practice, antenna gain are given in unit of dBi (dB gain with re pect to an i otropic antenna) or dBd (dB gain with re pect to a hal f-wave dipole) [Stu81]. The path lo , which repre ent ignal attenuation a a po itive quantit y mea ured in dB, i defined a the difference between the effective tran mitted power and the received power, and may or may not include the effect of the ante nna gain . The path lo for the free pace model when antenna gain are includ ed i given by (5)

, and

(3)

The effective aperture Ae i related to the phy ical i related to the carrier frequency by

ize of the antenna

When antenna gain are excluded, the antenna are a umed to have unity gain, an d path lo i given by (6) The Frii free pace model i only a valid predictor for Pr for value o f d which are in the far-field of the tran mitting antenna. The far-field, or F raunhofer region, or a tran mitting antenna i defined a the region beyond the far-field di tance df, which i related to the large t linear dimen ion of the t ran mitter antenna aperture and the carrier wavelength. The Fraunhofer di tance i given by (7a) where D i the large t phy ical linear dimen ion of the antenna. Additionally, to be in the far field region, df mu t ati fy

Furthermore, it i clear that Equation (1) doe not hold for d = 0. For thi rea on, large- cale propagation model u e a clo e-in di tance, d0, a a k nown received power reference point. The received power, Pr(d), at any di tance d > d0, may be related to Pr at d0. The value Pr(d0) may be predicted from Equ ation (1), or may be mea ured in the radio environment by taking the average rec eived power at many point located at a clo e-in radial di tance d0 from the tra n mitter. The reference di tance mu t be cho en uch that it lie in the far-fi eld region, that i , d0 df, and d0 i cho en to be maller than any practical di tance u ed in the mobile communication y tem. Thu , u ing Equation (1), the r eceived power in free pace at a di tance greater than d0 i given by (8) In mobile radio y tem , it i not uncommon to find that Pr may change b y many order of magnitude over a typical coverage area of everal quare kilome ter . Becau e of the large dynamic range of received power level , often dBm or dBW unit are u ed to expre received power level . Equation (8) may be expre ed in unit of dBm by imply taking the logarithm of both ide and multiplyin g by 10. For example, if Pr i in unit of dBm, the received power i given by (9)

The reference di tance d0 for practical y tem u ing low-gain antenna in the 1-2 GHz region i typically cho en to be 1 m in indoor environment and 1 00 m or 1km in outdoor environment , o that the numerator in Equation (8) and (9) i a multiple of 10. Thi make path lo computation ea y in dB unit . 2. If a Tran mitter produce 50 W of power, expre the tran mit power in unit of a) dBM and (b) dBW . If 50w i applied to a unity gain antenna with a 900 MH z carrier frequency, find the received power in dBM at free pace di tance of 10 0m from the antenna. What i Pr (10 km)? A ume unity gain for the receiver An tenna. Solution Given: Tran mitter power, Pt = 50 W Carrier frequency, fc = 900 MHz

Where Pr (d0) i in unit of watt

U ing Equation (a) Tran mitter power, Pt (dBm) = 10log [Pt(mW) / (1 mW)] = 10log [50x103] = 47.0 dBm. (b) Tran mitter power, Pt(dBW) = 10log [Pt(W) / (1 W)] = 10log [50] = 17.0 dBW. The received power can be determined u ing Equation

Pr(dBm) = 10log Pr (mW) = 10log (3.5x10-3 mW) = - 24.5 dBm.

3. Explain Reflection from dielectric .

Figure how an electromagnetic wave incident at an angel i with the plan e of the boundary between two dielectric media. A hown in the figure, part of the energy i reflected back to the fir t media at an angle r. The nature of re flection varie with the direction of polarization of the E-field. The behavior for arbitrary direction of polarization can be tudied by con idering the two d i tinct ca e hown in Figure. The plane of incidence i defined a the plane c ontaining the incident, reflected, and tran mitted ray [Ram65]. In Figure a, t he E-field polarization i parallel with the plane of incidence (that i , the Efield ha a vertical polarization, or normal component, with re pect to the refl ecting urface) and in figure b, the E-field polarization i perpendicular to th e plane of incidence (that i , the incident E-field i pointing out of the page toward the reader, and i perpendicular to the page and parallel to the reflecti ng urface). In Figure, the ub cript i, r, t refer to the incident, reflected, and tran mitted field , re pectively. Parameter 1, 1, 1, and 2, 2, 2 repre ent the permi ttivity, permeability, and conductance of the two media, re pectively. Often, t he dielectric con tant of a perfect (lo le ) dielectric i related to a relati ve value of permittivity, r, uch that = 0 r, where 0 i a con tant given by 8.85x10-12 F/m. If a dielectric material i lo y, it will ab orb power and may be de cri bed by a complex dielectric con tant given by = where = (2) and i the conductivity of the material mea ured in Siemen /meter. The term r a nd are generally in en itive to operating frequency when the material i a good conductor (f< / ( 0 r)). For lo y dielectric , 0 and r are generally con tant with f requency, but may be en itive to the operating frequency, a hown in Table. E lectrical propertie of a wide range of material were characterized over a larg e frequency range by Von Hipple [Von54]. Becau e of uperpo ition, only two orthogonal polarization need be con idered to olve general reflection problem . The reflection coefficient for th e two ca e of parallel and perpendicular E-field polarization at the boundary o 0 r - j (1)

Reflection from Dielectric

The received power at 10km can be expre e d0 = 100m and d=10km

ed in term of dBm u ing Equation, wher

f two dielectric are given by

(5)

The boundary condition from Maxwell equation are u ed to derive Equation (3) an d (4) a well a Equation (6), (7a), and (8b). i = r and Er = Ei Et = (1+ ) Ei (6) (7a) (7b)

where i either || or , depending on whether the E-field i in (vertical) or normal (horizontal) to the plane of incidence. For the ca e which the fir t medium i free pace and 1 = 2, the reflectio n coefficient for the two ca e of vertical and horizontal polarization can be implified to

For the ca e of elliptical polarized wave , the wave may be broken down (depolar ized) into it vertical and horizontal E-field component , and uperpo ition may be applied to determine tran mitted and reflected wave . In the general ca e o f reflection or tran mi ion, the horizontal and vertical

Figure: Axe for orthogonally polarized component . Parallel and perpendicular component are related to the horizontal and vertical patial coordinate . Wave i hown propagating out of the page toward the reader. Axe of the patial coordinate may not coincide with perpendicular and parallel axe of the propagating wave . An angle mea ured counter-clockwi e from the ho rizontal axi i defined a hown in Figure for a propagating wave out of the p[ age (toward the reader) [Stu93]. The vertical and horizontal field component a t a dielectric boundary may be related by (10) where are the depolarized field component in the horizontal and vertical dire ction , re pectively, are the horizontally and vertically polarized component of the incident wave, re pectively, and are time varying component of the E-

where i i the intrin ic the ratio of electric to lar medium. The velocity undary condition at the Figure, i given by

impedance of the ith medium (i = 1,2), and i given by magnetic field for a uniform plane wave in the particu of an electromagnetic wave i given by 1/ , and the bo urface of incidence obey Snell Law which, referring to

field which may be repre ented a pha or . R i a tran formation matrix which m ap vertical and horizontal polarized component to component which are perpend icular and parallel to the plane of incidence. The matrix R i given by

4. Explain Diffraction. Diffraction Diffraction allow radio ignal to propagate around the curved urface of the earth, beyond the horizon, and to propagate behind ob truction . Althoug h the received field trength decrea e rapidly a a receiver move deeper into the ob tructed ( hadowed) region, the diffraction field till exi t often ha ufficient trength to produce a u eful ignal. The phenomenon of diffraction can be explained by Huygen principle, whic h tate that all point on a wave front can be con idered a point ource for the production of econdary wavelet , and that the e wavelet combine to produce ma new wave front in the direction of propagation. Diffraction i cau ed by th e propagation of econdary wavelet into a hadowed region i the vector um of the electric field component of all the econdary wavelet in the pace around the ob tacle. Fre nel Zone Geometry Con ider a tran mitter and receiver eparated in free pace a hown in Figure a. Let an ob tructing creen of effective height h with infinite width ( going into and out of the paper) be placed between them at a di tance d1 from th e tran mitter and d2 from the receiver. It i apparent that the wave propagatin g from the tran mitter to the receiver via the top of the creen travel a longe r di tance than if a direct line-of- ight path (through the creen) exi ted. A uming then difference between the direct path and the diffracted path, called the exce path length ( ), can be obtained from the geometry of Figure b a (1) The corre ponding pha e difference i given by (2)

(3)

where ha unit of radian and i v i convenient becau e it allow

hown in Figure b and Figure c. The parameter to be expre ed in a convenient form.

Equation (2) i often normalized u ing the dimen ionle f diffraction parameter v which i given by

(a proof of Equation (1) and (2) i left a

an exerci e for the reader). Fre nel-Kirchof

where Dxx = x for the ca e of reflection and Dxx Tx = 1+ x for the ca e of tran mi ion [Stu93].

where i the angle between the two et of axe , a rization matrix DC i given by

hown in Figure. The depola

(4) From the above equation it i clear that the pha e difference between a direct line-of- ight path diffracted path i a function of height and po ition of the ob truction, a well a the tran mitter and receiver location.

The concept of diffraction lo a a function of the path difference aro und an ob truction i explained by Fre nel zone . Fre nel zone repre ent ucce ive region where econdary wave have a path length from the tran mitter to r eceiver which are n /2 greater than the total path length of a line-of- ight path. The concentric circle on the plane repre ent the loci of the origin of econ dary wavelet which propagate to the receiver uch that the total path length in crea e by /2 for ucce ive circle . The e circle are called Fre nel zone . T he ucce ive Fre nel zone have the effect of alternately providing con tructiv e and de tructive interference to the total received ignal. The radiu of the nth Fre nel zone circle i denoted by rn and can be expre ed in term of n, , d1 and d2 by (5) Thi approximation i valid for d1, d2

The exce total path length traver ed by a ray pa ing through each cir cle i n /2, where n i an integer. Thu , the path traveling through the malle t circle corre ponding to n = 1 in Figure will have an exce path length of /2 a compared to a line-of ,3 /2, etc. The radii of the concentric circle depend on the location of the plane. The Fre nel zone of Figure will have maximum radii if the plane i midway between the tran mitter or the receiver. Thi effect ill u trate how hadowing i en itive to the frequency a well a the location of ob truction with relation to the tran mitter or receiver. In mobile communication y tem , diffraction lo occur from the blocka ge of econdary wave that only a portion of the energy i diffracted around an ob tacle. That i , an ob truction cau e a blockage of energy from ome of the Fe nel zone , thu allowing only ome of the tran mitted energy to reach the rec eiver. Depending on the geometry of the ob truction, the received energy will b e a vector um of the energy contribution from all unob tructed Fre nel zone .

Figure: Diagram knife-edge geometry Figure: Concentric circle which define the boundarie of ucce ive Fre nel zo ne . A hown in Figure, an ob tacle may block the tran mi ion path, and a f amily of ellip oid can be con tructed between a tran mitter and receiver by joi ning all the point for which the exce path delay i an integer multiple of ha lf wavelength . The ellip oid repre ent Fre nel zone . Note that the Fre nel zone are elliptical in hape with the tran mitter and receiver antenna at their

ht

In practical diffraction problem , it i advantageou to reduce all heig by a con tant, o that the geometry i implified without changing the value of the angel . Thi procedure i hown in figure c.

foci. In Figure, different knife edge diffraction cenario are hown. In gen eral, if an ob truction doe not block the volume contained within the fir t Fre nel zone, then the diffraction lo will be minimal, and diffraction effect ma y be neglected. In fact, a rule of thump u ed for de ign of line-of- ight micro wave link i that a long a 55% of the fir t Fre nel zone i kept clear, then further Fre nel zone clearance doe not ignificantly alter the diffraction lo . Knife-edge Diffraction Model E timating the ignal attenuation cau ed by diffraction of radio wave o ver hill and building i e ential in predicting the field trength in a given ervice area. Generally, it i impo ible to make very preci e e timate of th e diffraction lo e , and inn practice prediction i a proce of theoretical ap proximation modified by nece ary empirical correction . Though the calculation of diffraction lo e over complex and irregular terrain i a mathematically di fficult problem, expre ion for diffraction lo e for many imple ca e have be en derived. A a tarting point, the limiting ca e of propagation over a knifeedge give good in ight into the order of magnitude of diffraction lo . When hadowing i cau ed by a ingle object uch a a hill or mountain, the attenuation cau ed by diffraction can be e timated by treating the ob tructi on a a diffracting knife edge. The i the imple t of diffraction model , and the diffraction lo in thi ca e can be readily e timated u ing the cla ical F re nel olution for the field behind a knife edge (al o called a half-plane). F igure illu trate thi approach. Con ider a receiver at point R, located in the hadowed region (al o cal led the diffraction zone). The field trength at point R in Figure i a vector um of the field due to all of the econdary Huygen ource in the plane above the knife edge. The electric field trength, Ed, of a knife-edge diffraction wa ve i given by

Figure: illu tration of Fre nel zone of different knife-edge diffraction cena rio . (6) where E0 i the free pace field trength in the ab ence of both the ground and the knife edge, and F(v) i the complex Fre nel intgegral. The Fre nel integral , F(v), i a function of the Fre nel-Kitchoff diffraction parameter , and i comm only evaluated u ing table or graph for given value of . The diffraction gain due to the pre ence of a knife edge, a compared to the free pace E-field, i given by

Gd(dB) = 20log

F( )

(7)

In practice, graphical or numerical olution are relied upon to compute diffrac tion gain. A graphical repre entation of Gd (dB) a a function of i given in Fi gure 4.14. An approximate olution for Equation (4.60) provided by Lee [Lee85] i

Figure: illu tration of knife-edge edge diffraction geometry. The receiver R i located in the hadow region

Gd(dB) = 0 (8a) Gd)dB) = 20log (0.5-0.62 ) (8b) -1

-1 0

5. Compute the diffraction lo for the three ca e hown in Figure. A ume = 1 /3 m, d1 = 1km, d2 = 1 km, and (a) h = 25 m, (b) = 0, (c) h = 25 m. Compare you r an wer u ing value from Figure, a well a the approximate olution given by Equation (8a)-(8e). For each of the e ca e , identify the Fre nel zone within which the tip of the ob truction lie . Given: = 1/3 m d1 = 1 km d2 = 1 km (a) h = 25 m U ing Equation (3), the Fre nel diffraction parameter i

To find the Fre nel zone in which the tip of the ob truction lie , we need to co mpute n which ati fie the relation =0/625m, we obtain

(b) h=0m Therefore, the Fre nel diffraction parameter = 0. From figure. The diffraction lo i obtained a 6 dB. U ing the numerical approximation in Equation, the diffraction lo 6 dB.

i equal to

For thi ca e, ince h = 0, we have = 0, and the tip of the ob truction lie in the middle of the fir t Fre nel zone. (c) h=25m

From Figure, the diffraction lo i approximately equal to 1 dB. Diffraction lo i equal to 0 dB. Since the ab olute value of the height h i the ame a part (a), the exce pat h length and hence n will al o be the ame. It hould be noted that although th e tip the ob truction completely block the fir t three Fre nel zone , the diffr action lo e are negligible, ince the ob truction i below the line-of- ight (

U ing Equation, the Fre nel diffraction parameter i

obtained a -2.74.

Therefore, the tip of the ob truction completely block zone .

the fir t three Fre nel

The path length difference between the direct and diffracted ray i

U ing the numerical approximation in equation, the diffraction lo 21.7 dB.

From Figure the diffraction lo

i obtained a

22 dB. i equal to

obtained a

h i negative).

Mo t radio propagation model are derived u ing a combination of analyti cal and empirical method . The empirical approach i ba ed on fitting curve or analytical expre ion that recreate a et of mea ured data. Thi ha the adva ntage of implicitly taking into account all propagation factor , both known and unknown, through actual field mea urement . However, the validity of an empiric al model at tran mi ion frequencie or environment other than tho e u ed to de rive the model can only be e tabli hed by additional mea ured data in the new en vironment at the required tran mi ion frequency. Over time, ome cla ical pro pagation model have emerged, which are now u ed to predict large- cale coverage for mobile communication y tem de ign. By u ing path lo model to e timate the received ignal level a function of di tance, it become po ible to predi ct the SNR for a mobile communication y tem. U ing noi e analy i technique g iven in Appendix B, the noi e floor can be determined. For example, the two-ray model de cribed in Section 4.6 wa u ed to e timate capacity in a pread pectr um cellular y tem, before uch y tem were deployed [Rap92b]. Practical path l o e timation technique are now pre ented.

Both theoretical and mea urement-ba ed propagation model indicate that average received ignal power decrea e logarithmically with di tance, whether i n outdoor radio channel . Such model have been u ed exten ively in the literat ure. The average large- cale path lo for and arbitrary T-R eparation i expr e ed a a function of di tance by u ing a path lo exponent, n where n i the path lo exponent which indicate the rate at which the path lo increa e with di tance; d0 i the clo e-in reference di tance which i determ ined from mea urement clo e to the tran mitter, and d i the T-R eparation d i tance. The bar inn Equation (1) and (2) denote the en emble average of all p o ible of all po ible path lo value for a given value of d. When plotted o n a log-log cale, the modeled path lo i a traight i a traight line with a lope equal to 10n dB per decade. The value of n depend on the pecific propa gation environment. For example, in free pace, n i equal to 2, and when ob tr uction are pre ent, n will have a larger value.

Environment Path lo Exponent, n Free pace 2 Urban area cellular radio 2.7 to 3.5 Shadowed urban cellular radio 3 to 5 In building line-of- ight 1.6 to 1.8 Ob tructed in building 4 to 6 Ob tructed in factorie 2 to 3 It i important to elect a free pace reference di tance that i approp riate for the propagation environment. In large coverage cellular y tem , 1 km reference di tance are commonly u ed [Lee85], wherea in microcellular y tem , much maller di tance ( uch a 100 m or 1 m) are u ed. The reference di tanc e hould alway be in the far field of the antenna o that near-field effect do not alter the reference path lo . The reference path lo i calculated u ing the free pace lo formula give by Equation or through field mea urement at d i tance d0. Table. Li t typical path lo exponent obtained in variou mobile

Table: Path Lo

Exponent for different Environment

Log-di tance path Lo

Model

Practical Link Budget De ign u ing Path Lo

Model

6. De ign practical Link Budget u ing path lo

Model .

radio environment . Log-normal Shadowing The model inn Equation (2) doe not con ider the fact that the urroundi ng environmental clutter may be va tly different at two different location havi ng the ame T-R eparation. Thi lead to mea ured ignal which are va tly dif ferent than the average value predicted by Equation (2). Mea urement have how n that at any value of d, path lo PL(d) at a particular location i random and di tributed log-normally (normal in dB) about the mean di tance-dependent value [Cox84], [Ber87]. That i

The log-normal di tribution de cribe the random hadowing effect which occur over a large number of mea urement location which have the ame T-R epar ation, but different level of clutter on the propagation path. Thi phenomenon i referred to a log-normal hadowing. Simply put, log-normal hadowing impli e that mea ured ignal level at a pecific T-R eparation have a Gau ian (nor mal) di tribution about the di tance-dependent mean of (2), where the mea ured ignal level have value in dB unit . The tandard deviation of the Gau ian di tribution that de cribe the hadowing al o ha unit in dB. Thu , the random effect of hadowing are accounted four u ing the Gau ian di tribution which le nd it elf readily to evaluation ( ee Appendix F).

The clo e-in reference di tance d0, the path lo exponent n, and the t andard deviation , tati tically de cribe the path lo model for an arbitrary lo cation having a pecific T-R eparation, and thi model may be u ed inn computer imulation to provide received power level for random location in communicati on y tem de ign and analy i . In practice, the value of n and are computed from mea ured data, u ing linear regre ion uch that the difference between the mea ured and e timated pa th lo e i minimized in a mean quare error en e over a wide range of mea ure ment location and T-R eparation . The value of in (3 a) i ba ed on either clo e-in mea urement or on a free pace a umption from the tran mitter to d0. An example of how the path lo exponent i determined from mea ured data follo w . Fig. illu trate actual mea ured data in everal cellular radio y tem and demon trate the random variation about the mean path lo (in dB) due to had owing at pecific T-R eparation .

The probability that received ignal level (in dB power unit ) will exce ed a certain value can be calculated from the cumulative den ity function a (5)

Determination of Percentage of Coverage Area

Similarly, the probability that the received ignal level will be below by

i given

Since PL (d) i a random variable with the di tance-dependent mean, o i Pr(d), and (erf) may be u ed to determine the probability ll exceed (or fall below) a particular level.

a normal di tribution in dB about the Q-function or error function that the received ignal level wi The Q-function i defined a

where X i a zero-mean Gau ian di tributed random variable (inn dB) with ard deviation (al o in dB).

tand

It i clear that due to random effect of hadowing, ome location with in a coverage area will be below a particular de ired received ignal thre hold. It i often u eful to compute how the boundary coverage relate to the percent of area covered within the boundary. For a circular coverage area having radiu R from a ba e tation, let there be ome de ired received ignal thre hold . We are intere ted in computing U ( ), the percentage of u eful ervice area (i.e. th e percentage of area with a received ignal that i equal or greater than ), given a known like hood of coverage at the cell boundary. Letting d = r repre ent th e radial di tance from the tran mitter, it can be hown that if Pr [Pr(r) > ] i the probability that the random received ignal at d=r exceed the thre hold wit hin an incremental area dA, then U( ) can be found by [Jak74]

By ub tituting t=a+blog (r/R) in Equation (11), it can be hown that

By choo ing the ignal level uch that Equation (12) may be evaluated for a large number of value of and n, a hown i n Fig. [Reu 74]. For example, of n=4 and =8 dB, and if the boundary i to have 75% boundary coverage (75% of the time the ignal i to exceed the thre hold at the boundary), then the area coverage i equal to 90%. If n=2 and =8 dB, a 75% b oundary coverage 86% area coverage. If n=3 and =9 dB, then 50% boundary coverage provide 71% area coverage. 7. Explain different outdoor propagation model .

Radio tran mi ion in a model communication y tem often take place ove r irregular terrain. The terrain profile of a particular area need to be taken into account for e timating the path lo . The terrain profile may vary from a imple curved earth profile to a highly mountainou profile. The pre ence of t ree , building , and other ob tacle al o mu t be taken into account. A number of propagation model are available to predict path lo over irregular terrain. While all the e model aim to predict ignal trength at a particular receivin g point or in a pecific local area (called a ector), the method vary widely i n their approach, complexity, and accuracy. Mo t of the e model are ba e on a y tematic interpretation of mea urement data obtained in the ervice area. Som e of the commonly u ed outdoor propagation model are now di cu ed. Longley-Rice Model

Outdoor Propagation Model

and Equation (8) may be expre ed a

In order to determine the path lo i clear that

U ing (4), Pr [Pr(r) > ] i

given by

a referenced to the cell boundary (r=R), it

The Longley-Rice model [Ric67], [Lon68] i applicable to point-to-point communication y tem in the frequency range from 40 MHz to GHz, over different kind of terrain. The median tran mi ion lo i predicted u ing the path geom etry of the terrain profile and the refractivity of the tropo phere. Geometric optic technique (primarily the two-ray ground reflection model) are u ed to pr edict ignal trength within the radio horizon. Diffraction lo e over i olat ed ob tacle are e timated u ing the Fre nel-Kirchoff knife-edge model . Forward catter theory i u ed to make tropo catter prediction over long di tance , an d far field diffraction lo e in double horizon path are predicted u ing a mod ified Van der Pol-Bremmer method. The Longley-Rice propagation prediction model i al o referred to a ITS irregular terrain model. The Longley-Rice Model i al o available a a computer program [Lon78] t o calculate large- cale median tran mi ion lo relative to free pace lo ove r irregular terrain for frequencie between 20 MHz. For a given tran mi ion pa th, the program take a it input the tran mi ion frequency, path length, pola rization, antenna height , urface refractivity, effective radiu of earth, grou nd conductivity, ground dielectric con tant, and climate. The program al o oper ate on path- pecific parameter uch a horizon di tance of the antenna , horiz on elevation angle, angular tran -horizon di tance, terrain irregularity, and ot her pecific input . The Longley-Rice method operate in two mode . When a detailed terrain path profiled i available, the path- pecific parameter can be ea ily determine d and the prediction i called a point-to-point mode prediction. On the other h and, if the terrain path profile i not available the Longley-Rice method provid e technique to e timate the path- pecific parameter , and uch a prediction i called an area mode prediction. There have been many modification and correction to the Longley-Rice m odel ince it original publication. One important modification [Lon78] deal wi th radio propagation in urban area , and thi i particularly relevant to mobile radio. Thi modification introduce an exce term a an allowance for the add itional attenuation due to urban clutter near the receiving antenna. Thi extra term, called the urban factor (UF), ha been derived by comparing the predictio n by the original Longley-Rice model with tho e obtained by Okumura [Oku68]. One hortcoming of the Longley-Rice model i that it doe not provide a way of determining due to environmental factor in the immediate vicinity of the mobile receive, or con ider correction factor to account for the effect of bu ilding and foliage. Further, multipath i not con idered.

A cla ical propagation prediction approach imilar to that u ed by Long ley-Rice i di cu ed by Edward and Durkin [Edw69], a well a Dad on [Dad75]. The e paper de cribe a computer imulator, for predicting field trength conto ur over irregular terrain, that wa adopted by the joint Radio Committee (JRC) in the U.K. for the e timation of effective mobile radio coverage area . Althou gh thi imulator only predict large- cale phenomena (i.e.path lo ), it provid e an intere ting per pective in to the nature of propagation over irregular ter rain and the lo e cau ed by ob tacle in a radio path. An explanation of the Edward and Durkin method i pre ented here inn order to demon trate how all of the concept de cribed in thi chapter are u ed in a ingle model. The execution of the Durkin path lo imulator con i t of two part . The fir t part acce a topographic data ba e of a propo ed ervice area and rec

Durkin

Model-A Ca e Study

on truct the ground profile information along the radial joining the tran mitte r to the receiver. The a umption i that the receiving antenna receive all of it energy along that radial and, therefore, experience no multipath propagati on. In other word , the propagation phenomena that i modeled i imply LOS and diffraction from ob tacle along the radial, and exclude reflection from othe r urrounding object and local catterer . The effect of thi a umption i th at the model i omewhat pe imi tic in narrow valley , although it identifie i olated weak reception area rather well. The econd part of the imulation alg orithm calculate the expected path lo along that radial. After thi i done, the imulated receiver location can be iteratively moved to different location in the ervice area to deduce the ignal trength contour.

The topographical data ba e can be thought of a a two-dimen ional array . Each array element corre pond to a point a ervice area map while the actual content of each array element contain the elevation above ea level data a h own in Figure. The e type of digital elevation model (DEM) are readily availa ble from the United State Geological Survey (USGS). U ing thi quantized map o f ervice area height , the program recon truct the ground profile along the ra dial that joint the tran mitter and the receiver. Since the radial may not alwa y pa through di crete data point , interpolation method are u ed to determin e the approximate height that are ob erved when looking along that radial. Fig ure a how the topographic grid with arbitrary tran mitter and receiver locatio n , the radial between the tran mitter and receiver, and the point with which t o u e diagonal linear interpolation. Figure b al o how what a typical recon t ructed radial terrain profile might look like. In actuality, the value are not imply determined by one interpolation routine, but by a combination of three f or increa ed accuracy. Therefore, each point of the recon tructed profile con i t of an average of the height obtained by diagonal, vertical (row), and horiz ontal (column) interpolation method . Figure: Illu tration a two-dimen ional array of elevation information

Figure: illu tration of terrain profile recon truction u ing diagonal interpolat ion From the e interpolation routine , a matrix of di tance from the receiver and co rre ponding height along the radial i generated. Now the problem i reduced t o a one-dimen ional point-to-point link calculation. The e type of problem ar e well-e tabli hed and procedure for calculating path lo u ing knife-edge dif fraction technique de cribed previou ly are u ed. At thi point, the algorithm mu t make deci ion a to what the expected tran mi ion lo hould be. The fir t tep i to decide whether a line-of- ig ht (LOS) path exi t between the tran mitter and the receiver. To do thi , the program computer the difference, , between the height of the line joining the tr an mitter and receiver antenna from the height of the ground profile for each p oint along the radial ( ee Figure)

If any i found to be po itive along the profile, it i concluded that a LOS path a clear LOS, otherwi e it can be concluded that a LOS path doe exi t . A uming the path ha a clear LOS, the algorithm then check to ee whether f ir t Fre nel zone clearance i achieved. A hown earlier, if the fir t Fre nel

Figure: illu tration of line-of- ight (LOS) deci ion making proce

zone of the radio path i unob tructed, then the re ulting lo mechani m i ap proximately that of free pace. If there i an ob truction that ju t barely tou che the line joining the tran mitter and the receiver then the ignal trength at the receiver i 6 dB le than the free pace value due to energy diffracting off the ob truction and away from the receiver. The method for determining fir t Fre nel zone clearance i dome by fir t calculating the Fre nel diffraction p arameter , defined in Equation (4.59), for each of the j ground element . If j -0.8 for all j=1,..n, then free pace propagation condition are domi nant. For thi ca e, the received power i calculated u ing the free pace tran mi ion formula given in Equation (4.1). If the terrain profile failed the fir t Fre nel zone te t (i.e. any j-0.8), then there are two po ibilitie : a. Nom-LOS b. LOS, but with inadequate fir t Fre nel-zone clearance.

For both of the e ca e , the program calculate the free pace power u i ng Equation and the received power u ing the plane earth propagation equation gi ven by Equation. The algorithm then elect the maller of the power calculate d with Equation a the appropriate received power for the terrain profile. If t he profile i LOS with inadequate fir t Fre nel zone clearance, the next tep i to calculate the additional lo due to inadequate Fre nel lo i calculated b y Equation. For the ca e of non-LOS, the y tem grade the problem into one of four categorie :

The method te t for each ca e equentially until it find the one that find the one that fit the given profile. A diffraction edge i detected by co mputing the angle between the line joining the tran mitter and receiver antenna and the line joining the receiver antenna to each point on the recon tructed terrain profile. The maximum of the e angle i located and labeled by the prof ile point (di, hi). Next, the algorithm tep through the rever e proce of ca lculating the angle between the line joining the tran mitter and receiver anten na and the line joining the tran mitter antenna to each point on the recon tru cted terrain profile. The Fre nel parameter, j, a ociated with thi edge can be determined from the length of the ob tacle above the line joining the tran mitt er and receiver antenna . The lo can then be evaluated by calculating PL u in g the Equation. Thi extra lo cau ed by the ob tacle i then added to either the free pace or plane earth lo , whichever i greater. If the condition for a ingle diffraction edge i not ati fied, then th e check for two diffraction edge i executed. The te t i imilar to that for a ingle diffraction edge, with the exception that the computer look for two ed ge in ight of each other ( ee Figure). The Edward and Durkin [Edw69] algorithm u e the Ep tein and Peter onn method [Ep 53] to calculate the lo a ociated with two diffraction edge . In hort, it i then um of two attenuation . The fir t attenuation i the lo at the econd diffraction edge cau ed by the fir t diffraction edge with the tran mitter a the ource. The econd attenuation i the lo at the receiver cau ed by the econd diffraction edge with the fir t diffraction edge a the ource. The two attenuation um to give the additional lo cau ed by the ob tacle tha t i added to the free pace lo or the plane earth lo , whichever i larger.

a. b. c. d.

Single diffraction edge Two diffraction edge Three diffraction edge More than three diffraction edge .

For three diffraction edge , the outer diffraction edge mu t contain a ingle diffraction edge in between. Thi i detected by calculating the line be tween the two outer diffraction edge . If an ob tacle between the two outer edg e pa e through the line, then it i concluded that a third diffraction edge e xi t ( ee Figure). Again, the Ep tein and Peter on method i u ed to calculate the hadow lo cau ed by the ob tacle . For all other ca e of more than thre e diffraction edge , the profile between the outer two ob tacle i approximated by a ingle, virtual knife edge. A after the approximation, the problem i tha t of a three edge calculation.

Thi method i very attractive becau e it can read in a digital elevatio n map and perform a ite- pecific propagation computation on the elevation data. It can produce a ignal trength contour that ha been reported to be good wit hin a few dB. The di advantage are that it cannot adequately predict propagati on effect due to foliage, building , other man-made tructure , and it doe not account for multipath propagation other than ground reflection, o additional l o factor are often included. Propagation prediction algorithm which u e ter rain information are typically u ed for the de ign of modern wirele y tem . Okumura Model Okumura model i one of the mo t widely u ed model for ignal predictio n in urban area . Thi model i applicable up to 3000 MHz) and di tance of 1 k m to 100 km. It can be u ed for ba e tation antenna height ranging from 30 m to 1000m. Okumura developed a et of curve giving the median attenuation relative to free pace (Amu), in an urban area over a qua i- mooth terrain with a ba e tation effective antenna height (h e) of 200 m and a mobile antenna height (hre) o f 3m. The e curve were developed from exten ive mea urement u ing vertical om ni directional antenna at both the ba e and mobile, and are plotted a a functi on of frequency in the range 100 MHz to 1920 MHZ and a a function of di tance f rom the ba e tation in the range 1 km to 100 km. To determine path lo u ing Okumura model, the free pace path lo between the point of intere t i fir t determined, and then the value of Amu (f,d) (a read from the curve ) i added t o it along with correction factor to account for the type of terrain. The mode l can be expre ed a L50 (dB) = LF+Amu (f,d) G(hte)-G(hre)-GAREA (1)

Where L50 i the 50th percentile (i.e., median) value of propagation path lo , Lf i the free pace propagation lo . Amu i the median attenuation relative t o free pace. G(hte) i the ba e tation antenna height gain factor. G (hre) i the mobile antenna height gain factor, and GAREA i the gain due to the type o f environment. Note that the antenna height gain are trictly a function of he ight and have nothing to do with antenna pattern . Plot of Amu (f,d) and GAREA for a wide range of frequencie are hown i n Figure. Furthermore, Okumura found that G(hte) varie at a rate of 20 dB/deca de and G(hte) varie at a rate of 10 dB/decade for height le than 3 m

Other correction may be applied to Okumura model. Some of the importan t terrain related parameter are the terrain undulation height ( h), i olated ridg e height, average lope of the terrain and the mixed land- ea parameter. Once t he terrain related parameter are calculated, the nece ary correction factor c an be added or ubtracted a required. All the e correction factor are al o av

Figure: illu tration of multiple diffraction edge

Figure: Median attenuation relative to free pace (Amu(f,d), over a qua i- mooth terrain [from[Oku68] IEEE].

Okumura model i wholly ba ed on mea ured data and doe not provide any analytical explanation. For many ituation , extrapolation of the derived curv e can be made to obtain value out ide the mea urement range, although the vali dity of uch extrapolation depend on the circum tance and the moothne of t he curve in que tion. Okumura model i con idered to be among the imple t and be t in term o f accuracy in path lo prediction for mature cellular and land mobile radio y tem in cluttered environment . It i very practical and ha become a tandard for y tem planning in modern land mobile radio y tem in Japan. The major di advantage with the model i it low re pon e to rapid change in terrain, there fore the model i fairly good in urban and uburban area , but not a good in ru ral area . Common tandard deviation between predicted and mea ured path lo value are around 10 dB to 14 dB. Hata Model The Hata model [Hat90] i an empirical formulation of the lo data provided by Okumura, and i valid from 150 MHz to 1500 ented the urban area propagation lo a a tandard formula and tion Equation for application to other ituation . The tandard dian path lo in urban area i given by graphical path MHz. Hata pre upplied correc formula for me

L50(urban) (dB) = 69.55+26.16 logfc-13.82loghte-a(hre) + (44.9-6.55loghte) log d (3) where fc i the frequency (in MHz) from 150 MHz to 1500 MHz, the i the effective tran mitter (ba e tation) antenna height (in meter ) ranging from m t o 200 m, hre i the effective receiver (mobile) antenna height (in meter ) rangi ng from 1 m to 10m, d i the T-R eparation di tance (in km), and a(hre) i the correction factor for effective mobile antenna height which i a function of the ize of the coverage area. For a mall to medium ized city, the mobile antenn a correction factor i given by a(hre) = (1.1logfc-0.7) hre-(1.56logfc-0.8) dB and for a large city, it i given by a(hre) = 8.29 (log 1.54hre)2 1.1 dB for fc a(hre) = 3.2 (log 11.75 hre)2 -4.97 dB for fc

(4)

300 MHz 300 MHz

(5.a) (5.b)

L50 (dB) = L50 (urban) 2 [log(fc/28)]2-5.4 (6)

To obtain the path lo n (2) i modified a

in a uburban area, the tandard Hata formula in Equatio

Figure: Correction factor, AREA, for different type EE]

of terrain 9from [Oku68] IE

ailable a

Okumura curve .

L50(dB) = L50 (urban) 4.78 (logfc)2 +18.33logfc-40.94

Although Hata model doe not have any of the path- pecific correction wh ich are available in Okumura model, the above expre ion have ignificant pract ical value. The prediction of the Hata model compare very clo ely with the ori ginal Okumura model, a long a d exceed 1 km. Thi model i well uited fro l arge cell mobile y tem , but not per onal communication y tem (PCS) which ha ve cell on the order of 1 km radiu . PCS Exten ion to Hata Model The European Cooperative for Scientific and Technical re earch (EURO-COS T) formed the COST-231 working committee to develop an extended ver ion of the H ata model. COST-231 propo ed the following formula to extend Hata model to 2 GH z. The propo ed model for path lo i [EUR91]. L50 (urban) = 46.3+33.9 logfc 13.82 log the a(hre) +(44.9-6.55loghte) log d +CM where a(hre) i defined in Equation (4), (5.a.) and (5.b) and CM= 9)

f :1500 MHz to 2000 MHz the :30 m to 200 m hre : 1 m to 10 m d : 1 km to 20 km Walfi ch and Bertoni Model

S = P0 Q2 P1

(10)

The factor Q2 give the reduction in the rooftop ignal due to the row o f building which immediately hadow the receiver at treet level. The P1 term i ba ed upon ba ed upon diffraction and determine the ignal lo from the roo ftop to the treet, (12)

Where L0 repre ent free pace lo , Lrt repre ent the rooftop-to- treet diffra ction and catter lo , and Lm denote diffraction lo due to the row of build ing [Xia92]. Fig. illu trate the geometry u ed in the Walfi ch Bertoni model

S(dB) = L0+Lrt +Lm

Where P0 repre ent

free pace path lo

between i otropic antenna

given by (11)

A model developed by Walifi ch and Bertoni [Wa188] con ider the impact or rooftop and building height by u ing diffraction to predict average ignal trength at treet level. The model con ider the path lo , S, to be a product of three factor .

The COST-231 exten ion of the Hata model i parameter :

re tricted to the following range of

3 dB for metropolitan center

0 dB for medium

ized city and uburban area

(7)

and for path lo

in open rural area , the formula i

modified a

(8)

[Wal88], [Mac93]. Thi model i being con idered for u e by ITU-R in the IMT-20 00 tandard activitie . Wideband PCS Micro cell Model Work by Feuer tein et al. in 1991 u ed a 20MHz pul ed tran mitter at 190 0 MHz to mea ure path lo , outage, and delay pread in typical microcellular y tem in San Franci co and Oak land. U ing ba e tation antenna height of 3.7 m, 8.5, and 13.3 and a mobile receiver with an antenna height of 1.7 m above gro und, tati tic for path lo , multipath, and coverage area

Figure: Propagation geometry for model propo ed by Walfi ch and Bertoni [from [W al88] IEEE]. Were developed from exten ive mea urement in line-of- ight (LOS) and ob tructed (OBS) environment {Feu.94]. Thi work revealed that a two-ray ground reflection model ( hown in Figure.) i a good e timate for path lo in LOS mic ro cell , and a imple log-di tance path lo model hold well for OBS micro cel l environment . For a flat earth ground reflection model, the di tance df at which the f ir t Fre nel zone ju t become ob tructed by the ground (fir t Fre nel zone clea rance) i given by

For LOS ca e , a double regre ion path lo model that u e a regre io n breakpoint at the fir t Fre nel zone clearance wa hown to fit well to mea ur ement . The model a ume omnidirectional vertical antenna and predict averag e path lo a

where P1 i equal to (the path lo in decibel at the reference di tance of d 0=1m), d i in meter and n1, n2 are path lo exponent which are a function of tran mitter height, a given in Figure. It can ea ily be hown that at 1900 MH z, p1=38.0 dB.

Where n i the OBS path lo exponent given in Figure a a function of t ran mitter height. The tandard deviation (in dB) of the log-normal hadowing c omponent about the di tance-dependent mean wa found from mea urement u ing the technique de cribed. The log-normal hadowing component i al o li ted a funct ion of height for both the LOS and OBS micro cell environment . Figure indicate that the log-normal hadowing component i between 7 and 9 dB regardle of an tenna height. It can be een that LOS environment provide lightly le path l o than the theoretical two-ray theoretical two-ray ground reflected model, whi ch would predict n1=2 and n2=4. Tran mitter Antenna Height 1900 MHz LOS n1 n2 (dB) 1900 MHz OBS n (dB) Low (3.7m) 2.18 3.29 8.76

2.58

9.31

For the OBS ca e, the path lo lo law of equation (3.a)

wa

found to fit the tandard log-di tance path

Medium (8.5 m) 2.17 High (13.3 m) 2.07

3.36 4.16

7.88 8.77

2.56 2.69

7.61 7.94

8. Find the median path lo u ing Okumura model for d=50 km, the=100m, hre=10 m in a uburban environment. If the ba e tation tran mitter radiate an EIRP of 1 kW at a carrier frequency of 900 MHz, find the power at the receiver (a ume a unity gain receiving antenna). Solution

Amu (900 MHz(50 km) = 43 dB And GAREA = 9 dB. U ing Equation (2.a) and (2.c), we have

G(the) = 20log G(hre) = 20log

L50 (dB) = LF+Amu(f,d)-G(the)-G(hre) GAREA = 125.5dB+43 dB-(-6) dB-10.46 dB-9 dB

Pr(d) = EIRP (dBm) L50 (dB)+Gr (dB) =60 dBm-155.04 dB+0 dB = 95.04 dBm. 9. Explain about Indoor propagation Model .

With the advent of Per onal Communication Sy tem (PCS), there i a great deal of intere t inn characterizing radio propagation in ide building . The in door channel differ from the traditional mobile radio channel in two a pect -th e di tance covered are much maller, and the variability of the environment i much greater for a much maller range of T-R eparation di tance . It ha been ob erved that propagation within building i trongly influenced by pecific fe

Indoor Propagation Model

Therefore, the median received power i

U ing Equation (1), the total mean path lo

From the Okumura curve

The free

pace path lo

LF can be calculated u ing Equation (4.6) a

Figure: Parameter EEE]

for the wideband micro cell model at 1900 MHz [from [Feu94] I

ature uch a the layout of the building, the con truction material , and the b uilding type. Thi ection outline for path lo within building . Indoor radio propagation i dominated by the ame mechani m a outdoor: reflection, diffraction, and cattering. However, condition are much more var iable. For example, ignal level vary greatly depending on whether interior do or are open or clo ed in ide a building. Where antenna are mounted al o impac t large- cale propagation. Antenna mounted at de k level in a partitioned off ice receive va tly ignal than tho e mounted on the ceiling. Al o, the maller propagation di tance make it more difficult in ure far-field radiation for all receiver location and type of antenna . The field or indoor radio propagation i relatively new, with the fir t wave of re earch occurring in the early 1980 . Cox [Cox83b] at AT & T Bell Labo ratorie and Alexander [Ale82] at Briti h Telecom were the fir t to carefully t udy indoor path lo in and around a large number of home and office building . Excellent literature urvey are available on the topic indoor propagation . In general, indoor channel may be cla ified either a line-of- ight (L OS) or ob tructed (OBD), with varying degree of clutter [Rap89]. Some of the k ey model which have recently emerged are now pre ented.

Building have a wide variety of partition and ob tacle which form the internal and external tructure. Hou e typically u e a wood frame partition w ith pla ter board to form internal wall and have wood or nonreinforced concrete between floor . Office building , on the other hand, often have large open are a (open plan) which are con tructed by u ing moveable office partition o that the pace may be reconfigured ea ily, and u e metal reinforced concrete between floor . Partition that are formed a part of the building tructure are calle d hard partition , and partition that may be moved and which do not pan to the ceiling are called oft partition . Partition vary widely in their phy ical a nd electrical characteri tic , making it difficult to apply general model to p ecific indoor in tallation . Neverthele , re earcher have formed exten ive da ta ba e of lo e for a great number of partition , a hown in Table. Table: Average Signal Lo Mea urement Reported by variou Re earcher for Radi o Path Ob tructed by common Building Material Material Type Lo (dB) Frequency All metal 26 815MHz Aluminum Siding 20.4 815MHz Foil in ulation 3.9 815MHz Concrete block wall 13 1300MHz Lo from one floor 20-30 1300MHz Lo from one floor and one wall 40-50 1300MHz Fade ob erved when tran mitter turned a right angle corner in a corridor 10-15 1300MHz Light textile inventory 3-5 1300MHz Chain-like fence in area 20 ft high containing tool , inventory and people 5-12 1300MHz Metal blanket 12 q ft 4-7 1300MHz Metallic hopper which hold crap metal for recycling 10 q ft 3-6 1300MHz Small metal pole 6 diameter 3 1300MHz Metal pulley y tem u ed to hoi t metal inventory 4 q ft 6 1300MHz Light machinery 10 q ft 1-4 1300MHz General machinery 10-20 q ft 5-10 1300MHz Heavy machinery > 20 q ft 10-12 1300MHz Metal Catwalk/ tair 5 1300MHz Light textile 3-5 1300MHz

Partition Lo e

( ame floor)

Heavy textile inventory 8-11 1300MHz Area where worker in pect metal fini hed product for defect 3-12 1300MHz Metallic inventory 4-7 1300MHz Large 1-beam-16-20 8-10 1300MHz Metallic inventory rack -8 q ft 4-9 1300MHz Empty cardboard inventory boxe 3-6 1300MHz Concrete block wall 13-20 1300MHz Ceiling duct 1-8 1300MHz 2.5 m torage rack with mall metal part (loo ely packed) 4-6 1300MHz 4 m metal box torage 10-12 1300MHz 5 m torage rack with product (loo ely packed) 2-4 1300 MHz 5m torage rack with large paper product 9tightly packed) 6 1300 MHz 5m torage rack with large metal part (tightly packed) 20 1300 MHz Typical N/C machine 8-10 1300 MHz Semi-automated a embly line 5-7 1300 MHz 0.6m quare reinforced concrete pillar 12-14 1300 MHz Stainle teel piping for cook-cool proce 15 1300 MHz Concrete wall 8-15 1300 MHz Concrete floor 10 1300 MHz Commercial ab orber 38 9.6 GHz Commercial ab orber 51 28.8 GHz Commercial ab orber 59 57.6 GHz Sheetrock (3/8in)-2 heet 2 9.6 GHz Sheetrock (3/8in)-2 heet 2 28.8 GHz Sheetrock (3/8in)-2 heet 5 57.6 GHz Dry plywood (3/4 in) -1 heet 1 9.6 GHz Dry plywood (3/4 in) -1 heet 4 28.8 GHz Dry plywood (3/4 in) -1 heet 8 57.6 GHz Dry plywood (3/4 in) -2 heet 4 9.6 GHz Dry plywood (3/4 in) -2 heet 6 28.8 GHz Dry plywood (3/4in) -2 heet 14 57.6 GHZ Wet plywood (3/4in) 1 heet 19 9.6 GHZ Wet plywood (3/4in) 1 heet 32 28.8 GHz Wet plywood (3/4in) 1 heet 59 57.6 GHz Wet plywood (3/4 in) 2 heet 39 9.6 GHz Wet plywood (3/4 in) 2 heet 46 28.8 GHz Wet plywood (3/4 in) 2 heet 57 57.6 GHz Aluminum (1/8 in) 1 heet 47 9.6 GHz Aluminum (1/8 in) 1 heet 46 28.8 GHZ Aluminum (1/8 in) 1 heet 53 57.6 GHz Table: Total Floor Attenuation Factor and Standard Deviation (dB) for Three Buil ding . Each Point Repre ent the Average Path Lo Over a 20 Mea urement Track [ Sei92a] Building (dB) (dB location MHz FAF(dB) (dB) Of Location Walnut Creek One Floor Two Floor SF PacBell One Floor Two Floor Three Floor Four Floor 915MHz FAF Number of 1900 Number

33.6 44.0 13.2 18.1 24.0 27.0

3.2 4.8 9.2 8.0 5.6 6.8

2.5 39 16 10 10 10

31.3 38.5 26.2 33.4 35.2 38.4

4.6 4.0 10.5 9.9 5.9 3.4

110 29 21 21 20 20

Five Floor San Ramon One Floor Two Floor Three Floor

27.1 29.1 36.6 39.6

6.3 5.8 6.0 6.0

10 93 81 70

46.4 35.4 35.6 35.2

3.9 6.4 5.9 3.9

17 74 41 27

The lo e between floor of a building are determined by the external d imen ion and material of the building, a well the type of con truction u ed to create the floor the external urrounding [Sei92a], [Sei92b]. Even the number of window in a building and the pre ence of tinting (which attenuate radio en ergy) can impact the lo between floor . Table Illu trate value for floor at tenuation factor (FAF) in three building in San Franci co [Sei92a]. It can be een that for all three building , the attenuation between on floor of the buil ding i greater than the incremental attenuation cau ed by each additional floor . Table illu trate very imilar tendencie . After about five or ix floor ep aration , very little additional path lo i experienced.

7.0 2.8 1.7 1.5 2.9 5.4 7.2

52 9 9 9 21 21 21

Indoor path lo ha been hown by many re earcher to obey the di tance power l aw in Equation (14)

Eric on Multiple Breakpoint Model The Eric on radio model wa obtained by mea urement in a multiple floo r office building [Ake88]. The model ha four breakpoint and con ider both an upper and lower bound on the path lo . The model al o a ume that there i 30 dB attenuation at d0 = 1 m, which can be hown to be accurate for f = 900 MHz a nd unity gain antenna . Rather than a uming a log-normal hadowing component, the Eric on model provide a determini tic limit on the range of path lo at a particular di tance. Bernhardt [Ber89] u ed a uniform di tribution to generate

where the value of n depend on the urrounding and building type, and nt a normal random variable in dB having a tandard deviation of dB. at Equation (14) i identical in form to the log-normal hadowing model ion (3.a). Typical value for variou building are provided in Table.

Log-di tance Path Lo

Model

Building FAF (dB) Office Building 1: Through One Floor 12.9 Through Two Floor 18.7 Through Three Floor 24.4 Through Four Floor 27.0 Office Building 2: Through One Floor 16.2 Through Two Floor 27.5 Through Three Floor 31.6

(dB) Number of location

repre e Notice th of Equat [And94].

Table: Average Floor Attenuation Factor in dB for One, Two Three and Four Floor in Two Office Building [Sei92b]

Partition Lo e

between Floor

path lo value within the maximum and minimum range a a function of di tance for in-building imulation. Fig. how a plot of in-building path lo ba ed o n the Eric on model a a function of di tance.

Building Frequency (MHz) Retail Store 914 2.2 Grocery Store 914 1.8 Office, hard partition 1500 Office, oft partition 900 Office, oft partition 1900 Factory LOS Textile/Chemical 1300 Textile/Chemical 4000 Paper/Cereal 1300 1.8 Metalworking 1300 1.6 Suburban Home 900 3.0 Indoor Street Factory OBS Textile/Chemical 4000 Metalworking 1300 3.3

n 8.7 5.2 3.0 2.4 2.4 2.0 2.0 6.0 5.8 7.0 2.1 6.8

(dB) 7.0 9.6 9.6 3.0 7.0

9.7

Attenuation Factor Model An in-building ite- pecific propagation model that include the effect of building type a well a the variation cau ed by ob tacle wa de cribed by Seidel [Sei92b] and ha been u ed to accurately deploy indoor and campu network [Mor00, Ski96]. Thi model provide flexibility and wa hown to reduce the tandard deviation between mea ured and predicted path lo to around 4 dB, a co mpared to 13 dB when only a log-di tance model wa u ed in two different buildin g . The attenuation factor model i given by

Table illu trate

typical value of n for a wide range of location in m

where nMF denote a path lo oor .

exponent ba ed on mea urement

through multiple fl

where nSF repre ent the exponent value for the ame floor mea urement, FAF repre ent a floor attenuation factor for a pecified number of building floor , and P AF repre ent the partition attenuation factor for a pecific ob truction encoun tered by a ray draw between the tran mitter and receiver in 3-D. Thi technique of drawing a ingle ray between the tran mitter and receiver i called primary ray tracing. Summing the cumulative partition lo e along the primary ray ha been hown to yield excellent accuracy while offering tremendou computational e fficiency [Mor00, Ski96, Rap00]. Thu , if a good e timate for n exi t on the ame floor, then the path lo on a different floor can be predicted by adding an appropriate value of FAF (e.g., elected from Table.), and then umming the par tition lo e elected from Table. Alternatively, in Equation (15), FAF may be replaced by an exponent which already con ider the effect of multiple floor e paration

Figure: Eric on in-building path lo

model [from[Ake88] IEEE].

Table: Path Lo [And94]

Exponent and Standard Deviation Mea ured in Different Building

any building . Thi table al o illu trate how the tandard deviation decrea e a the average region become maller and more ite pecific. Scatter plot i llu trating actual mea ured path in two multifloored office building are hown in figure.

where i the attenuation con tant for the channel with unit of dB per meter (dB /m). Table provide typical value of a a function of frequency a mea ured in [Dev90b].

n (dB) Location All Building : All location 3.14 Same floor 2.76 Through One Floor Through Two Floor Through Three Floor Grocery Store 1.81 Retail Store 2.18 Office Building 1: Entire Building 3.54 Same Floor 3.27 We t Wing 5th Floor Central Wing 5th Floor We t Wing 4th Floor Office Building 2: Entire Building 4.33 Same Floor 3.25

Number of 16.3 12.9 4.19 5.04 5.22 5.2 8.7 12.8 11.2 2.68 4.01 3.18 13.3 5.2 634 501 5.1 6.5 6.7 89 137 320 238 8.1 4.3 4.4 100 37

73 30 30

104 120 120

Location Frequency -Attenuation (dB/m) Building 1:4 tory 850 MHz 0.62 1.7 GHz 0.57 4.0 GHz 0.47 Building 2:2 tory 850 MHz 0.48 1.7 GHz 0.35 4.0 GHz 0.23

The ignal trength received in ide of a building due to an external tra n mitter i important for wirele y tem that hare frequencie with neighborin g building or with outdoor y tem . A with propagation mea urement between f loor , it i difficult to determine exact model for penetration a only a limit ed number of experiment have been publi hed, and they are ometime difficult t

Signal Penetration into Building

Table: Free Space Plu

Linear path Attenuation Model [Dev90b]

Table: Path Lo [Sei92b].

Exponent and Standard Deviation for Variou

Type of Building

Deva e pace plu ce, a hown po ible to

irvatham et al. [Dev90b] found that in-building path lo obey fre an additional lo factor which increa e exponentially with di tan in Table. Ba ed on thi work in multi-floor building , it would be modify Equation (15) uch that

o compare. However, ome generalization can be made from the literature. In me a urement reported to date, ignal trength received in ide a building increa e with height. At the lower floor of building , the urban clutter induce the level of penetration. At higher floor , a LOS path may exi t, thu cau ing a t ronger incident ignal at the exterior wall of the building.

RF penetration ha been found to be a function of frequency a well a h eight within the building . The antenna pattern in the elevation plane al o pla y an important role in how much ignal penetrate a building from the out ide. Mo t mea urement have con idered outdoor tran mitter with antenna height far e le the maximum height of the building under te t. Mea urement in Liverpool [Tur87] howed that penetration lo decrea e with increa ing frequency. Spec ifically, penetration attenuation value of 16.4dB, 11.6dB dB, and 7.6 dB were m ea ured on the ground floor of a building at frequencie of 441 MHz, 896.5 MHz, and 1400 MHz, re pectively. Mea urement by Turkmani [Tur92] howed penetration lo of 14.2 dB, 13.4 dB, and 12.8 dB for 900 MHz, 6 dB le penetration lo o n average than did mea urement made in part of the building without window . Walker [Wal92] mea ured radio ignal into fourteen different building in Chicago from even external cellular tran mitter . Re ult howed that build ing penetration lo decrea ed at a rate of 1.9 dB per floor from then ground le vel up to the fifteenth floor and then began increa ing above the fifteen floor. The increa e in penetration lo at higher floor wa attributed to hadowing effect of adjacent building . Similarly, Turkmani [Rur87] reported penetration lo decrea ed at a rate of 2 dB per floor from the ground level up to the nint h floor and then increa ed above the ninth floor. Similar re ult were al o rep orted by Durante [Dur73]. Mea urement have hown that the percentage of window , when compared wi th the building face urface area, impact the level of RF penetration lo , a doe the pre ence of tinted metal in the window . Metallic tint can provide fr om 3 dB to 30 dB RF attenuation in a ingle pane of gla . The angle of inciden ce of the tran mitted wave upon the face of the building al o ha a trong impac t on the penetration lo , a wa hown by Horiki hi [Hor86]. Ray Tracing and ite pecific Modeling In recent year , the computational and vi ualization capabilitie of com puter have accelerated rapidly. New method for predicting radio ignal covera ge involve the u e of Site Specific (SISP) propagation model and graphical info rmation y tem (GIS) databa e [Ru 93]. SISP model upport ray tracing a a mea n of determini tically modeling any indoor or outdoor propagation environment. Through the u e of building databa e , which may be drawn or digitized u ing t andard graphical oftware package , wirele y tem de igner are able to includ e accurate repre entation of building and terrain feature . For outdoor propagation prediction, ray tracing are u ed in conjunction with aerial photograph o that three-dimen ional (3-d) repre entation of build ing may be integrated with oftware that carrie out reflection, diffraction, a nd cattering model . Photogrametric technique are u ed to convert aerial or atellite photograph of citie into u able 3-D databa e for the model [Sch92], [Ro 93], [Wag94], [Rap00]. In indoor environment , architectural drawing prov ide a ite pecific repre entation for propagation model [Val93], [Sei94] [Kre9 4], [Ski96], [Mor00]. A building databa e become prevalent, wirele y tem will be developed u ing computer aided de ign tool that provide determini tic, rather than tati tical, prediction model of mall- cale and large- cale path lo in a wide ran ge of operating environment . For example, in the SitePlanner ha become a popul ar re earch and teaching tool for tudent and engineer new to the wirele fie

ld. Some day uch ite pecific modeling technique may be downloadable into wi rele phone and u ed to determine in tantaneou air interface parameter . 10. Li t out the effect and Factor influencing mall cale Fading. Small-Scale Multipath Propagation

In built-up urban area , fading occur becau e the height of the mobile antenna are well below the height of urrounding tructure , o there i no in gle line-of- ight path to the ba e tation. Even when a line-of- ight exi t , m ultipath till occur due to reflection from the ground and urrounding tructu re . The incoming radio wave arrive from different direction with different p ropagation delay . The ignal received by the mobile at any point in pace may con i t of a large number of plane wave having randomly di tributed amplitude , pha e , and angle of arrival. The e multipath component combine vectorially at the receiver antenna, and can cau e the ignal received by the mobile to di t ort or fade. Even when a mobile receiver i tationary, the received ignal may fade due to movement of urrounding object in the radio channel. If object in the radio channel are tatic, and motion i con idered to be only due to that of the mobile, then fading i purely a patial phenomenon. The patial variation of the re ulting ignal are een a temporal variation b y the receiver a it move through the multipath field. Due to the con tructive and de tructive effect of multipath wave umming at variou point in pace, a receiver moving at high peed can pa through everal fade in a mall period of time. In a more eriou ca e, a receiver may top at a particular location at which the received ignal i in a deep fade. Maintaining good communication can then become very difficult, although pa ing vehicle or people walking in the vicinity of the mobile can often di turb the field pattern, thereby dimini h ing the likelihood of the received ignal remaining in a deep null for a long pe riod of time. Antenna pace diver ity can prevent deep fading null , a hown i n Chapter 6. Fig. how typical rapid variation in the received ignal level d ue to mall- cale fading a a receiver i moved over a di tance a few meter . Due to the relative motion between the mobile and the ba e tation, each multipath wave experience an apparent hift in frequency. The hift in receiv ed ignal frequency due to motion i called the Doppler hift, and i directly p roportional to the velocity and direction of motion of the mobile with re pect t o the direction of arrival of the received multipath wave. Factor Influencing Small-Scale Fading

Multipath Propagation: The pre ence of reflecting object and catterer in the channel create a con tantly changing environment that di ipate the ignal in amplitude, pha e, and time. The e effect re ult ver ion of the tran mitted i gnal that arrive at the receiving antenna, di placed with re pect to one another in time and patial orientation. The random pha e and amplitude of the differ ent multipath component cau e fluctuation in ignal trength, thereby inducing mall- cale fading, ignal di tortion, or both. Multipath propagation often le ngthen the time required for the ba eband portion of the ignal to reach the re

Many phy ical factor in the radio propagation channel influence mallcale fading the e include the following

ignal

Time di per ion (echoe ) cau ed by multipath propagation delay .

Rapid change in ignal trength over a mall travel or time interval Random frequency modulation due to varying Doppler hift on different multipath

Multipath in the radio channel create three mo t important effect are:

mall- cale fading effect . The

ceiver which can cau e ignal mearing due to inter ymbol interference.

Speed of urrounding object : If object in the radio channel are in motion, the y induce a time varying Doppler hift on multipath component . If the urroundi ng object move at a greater rate than the mobile, then thi effect dominate th e mall- cale fading. Otherwi e, motion of urrounding object may be ignored, and only the peed of the mobile need be con idered. The coherence time define the taticne of the channel, and i directly impacted by the Doppler hift. The tran mi ion bandwidth of the ignal: If the tran mitted radio ignal bandwi dth i greater than the bandwidth of the multipath channel, the received ignal wi ll be di torted, but the received ignal trength will not fade much over a loca l area (i.e., the mall- cale ignal fading will not be ignificant). A will b e hown, the bandwidth of the channel can be quantified by the coherence bandwid th which i related to the pecific multipath tructure of the channel. The coh erence bandwidth i related to the pecific multipath tructure of the channel. The coherence bandwidth i a mea ure of the maximum frequency difference for wh ich ignal are till trongly correlated in amplitude. If the tran mitted ign al ha a narrow bandwidth a compared to the channel, the amplitude of the of m all- cale ignal trength and the likelihood of ignal mearing appearing over mall cale di tance are very much related to the pecific amplitude and delay of the multipath channel, a well a the bandwidth of the tran mitted ignal. 13. Con ider a tran mitter which radiate a inu oidal carrier frequency of 1850 MHz. For a vehicle moving 60 mph, compute the received carrier frequency if th e mobile i moving i moving (a) directly toward the tran mitter, (b) directly a way from the tran mitter, and (c) in a direction which i perpendicular to the d irection of arrival of the tran mitted ignal. Solution Given: Carrier frequency fc = 1850 MHz Therefore, wavelength = c/fc =

(a) The vehicle i moving directly toward the tran mitter. The Doppler hift in thi ca e i po itive and the received frequency i gi ven by Equation (b) The vehicle i moving directly away from the tran mitter. The Doppler hift in thi ca e i negative and hence the received Frequency i given by

(c) The vehicle i moving perpendicular to the angle of arrival of the tran mitt ed ignal. In thi ca e, =90 , co =0, and there i no Doppler hift, The received ignal frequency i the ame a the tran mitted frequency of 1 850 MHz.

Vehicle peed V=60 mph=26.82 m/

Speed of the mobile: The relative motion between the ba e tation and the mobile re ult in random frequency modulation due to different Doppler hift on each o f the multipath component . Doppler hift will be po itive or negative dependin g on whether the mobile receiver i moving toward or away from the ba e tation.

11. Explain Impul e Re pon e Model of a Multipath Channel. Impul e Re pon e Model of a Multipath Channel The mall- cale variation of a mobile radio ignal can be directly rela ted to the impul e re pon e of the mobile radio channel. The impul e re pon e i a wideband channel characterization and contain all information nece ary to imulate or analyze any type of radio tran mi ion through the channel. Thi t em from the fact that a mobile radio channel may be modeled a a linear filter with a time varying impul e re pon e, where the time variation i due to receive r motion in pace. The filtering nature of the channel i cau ed by the ummati on of amplitude and delay of the multiple arriving wave at any in tant of tim e. The impul e re pon e i a u eful characterization of the channel, ince it m ay be u ed to predict and compare the performance of many different mobile commu nication y tem and tran mi ion bandwidth for a particular mobile channel con dition. To how that a mobile radio channel may be modeled a a linear filter wi th a time varying impul e re pon e, con ider the ca e where time variation i du e trictly to receiver motion in pace. Thi i hown in Fig. In Fig. the receiver move along the ground at ome con tant velocity . F or a fixed po ition d, the channel between the tran mitter and the receiver can be modeled a a linear time invariant y tem. However, due to the different mul tipath wave which have propagation delay which vary over different patial loc ation of the receiver, the impul e re pon e of the linear time invariant channe l hould be a function of the po ition of the receiver. That i , the channel im pul e re pon e can be expre ed a h(d,t). Let x(t) repre ent the tran mitted ignal, then the received ignal (d,t) at po ition d can be expre ed a a convol ution of x(t) with h (d.t). Since the receiver move along the ground at a con tant velocity , the po ition of the receiver can by expre ed a D= t (3)

Figure: The mobile radio channel a a function of time and pace. Sub tituting (3) in (2), we obtain

Since i con tant, y( t,t) i ju t a function of t. Therefore, Equation (4) can b e expre ed a

From Equation (5), it i clear that the mobile radio channel can be modeled a a linear time varying channel, where the channel change with time di tance. Since may be a umed con tant over a hort time (or di tance) interval, we may let x(t) repre ent the tran mitted bandpa waveform, y(t) the received w aveform, and h(t, ) the impul e re pon e of the time varying multipath radio chan nel. The impul e re pon e h(t, ) completely characterize the channel and i a fu nction of both t and . The variable t repre ent the time variation due to moti on, wherea repre ent the channel multipath delay for a fixed value of t. One may think of a being a venire adju tment of time. The received ignal y(t) ca n be expre ed a a convolution of the tran mitted ignal x(t) with them channel

impul e re pon e.

Figure: (a) Bandpa channel impul e re pon e model: (b) ba eband equivalent ch annel impul e re pon e model. If the multipath channel i a umed to be a bandlimited bandpa channel , which i rea onable, then h(t, ) may be equivalently de cribed by a complex ba e band impul e re pon e hb(t, ) with the input and output being the complex envelope repre entation of the tran mitted and received ignal , re pectively.

The factor of in Equation (7) i due to the propertie of the complex en velope, in order to repre ent the pa band radio y tem at be ba eband. The lo wpa characterization remove the high frequency variation cau ed by the carri er, making the ignal analytically ea ier to handle. It i hown by Couch [Cou9 3] that the average power of a bandpa ignal x2(t) i equal to 0.5 c(t) 2, where t he or ergodic tocha tic ignal. It i u eful to di cretize the multipath delay axi of the impul e re po n e into equal time delay egment called exce delay bin , where each bin ha a time delay width equal to i+1- i where 0 i equal to 0, and repre ent the fir t a rriving ignal at the receiver. Letting i=0, it i een that 1- 0 i equal to the time delay bit width given by . For convention, 0=0, 1 = , and i=i , for i=0 to N-1, N repre ent the total number of po ible equally- paced multipath component , including the fir t arriving component. Any number of multipath ignal receive d within the ith bin are repre ented by a ingle re olvable multipath component having delay i, Thi technique of quantizing the delay bin determine the time delay re olution of the channel model, and the u eful frequency pan of the mode l can be hown to be 2/ . That i , the mode may be u ed to analyze tran mitted RF ignal having bandwidth which are le than 2/ . Note that 0=0 i the exce time delay of the fir t arriving multipath component, and neglect the propagation d elay between the tran mitter and receiver. Exce delay i the relative delay o f the in multipath component a compared to the fir t arriving component and i given by i. The maximum exce delay of the channel i given by N . Since the received ignal in a multipath channel con i t of a erie of attenuated, time delayed, pha e hifted replica of the tran mitted ignal; the ba eband impul e re pon e of a multipath channel can be expre ed a

where ai (t, ) and i(t) are the real amplitude and exce delay , re pectively, of ith multipath component at time t[Tur72]. The pha e term 2 fc i(t)+ i(t, ) in (10) rep re ent the pha e hift due to free pace propagation of the ith multipath compo nent, plu any additional pha e hift which are encountered in the channel. In general, the pha e term i imply repre ented by a ingle variable i (t, ) which l ump together all the mechani m for pha e hift of a ingle multipath componen t within the ith exce delay bin. Note that ome exce delay bin may have no multipath at ome time t and delay i and delay i, ince ai(t, ) may be zero. In Eq uation (10), N i the total po ible number of multipath component (bin ), and () i the unit impul e function which determine the pecific multipath bin that have component at time t and exce delay i, Figure illu trate an example of d ifferent nap hot of hb(t, ), where t varie into the page, and the time delay bi n are quantized to width of . Modern wirele communication y tem have recent ly u ed patial filtering to increa e capacity and coverage, and often it i u e ful to modify Equation (10) to include the effect of angle of arrival of each m ultipath component.

It i important to not that depending on the choice of and the phy ical c hannel delay propertie , there may be two or more multipath ignal that arrive within an exce delay bin that are unre olvable and that vectorially combine to yield the in tantaneou amplitude and pha e of a ingle modeled multipath compo nent. Such ituation cau e the multipath amplitude within an exce delay bin to fade over the local area. However, when only a ingle multipath component ar rive within an exce delay bin, the amplitude over the local area for that par ticular time delay will generally not fade ignificantly.

Figure: An example of the time varying di crete-time impul e re pon e model for a multipath radio channel. Di crete model are u eful in imulation where modula tion data mu t be convolved with the channel impul e re pon e [Tra02]. If the channel impul e re pon e i a umed to be time invariant, or i a t lea t wide en e tationary over a mall- cale time or di tance interval, then the channel impul e re pon e may be implified a The a umption of time invariance over a local area i valid when the time delay re olution of the channel impul e re pon e model accurately and uniquely re olv e very multipath component over local area.

i u ed to ound the channel to determine hb( ).

where the bar repre ent the average over the local area and many nap h ot of hb(t; ) 2 are typically averaged over a local ( mall- cale) area to provide a ingle time-invariant multipath power delay profile P( ). The gain k in Equation (13) relate the tran mitted power in the probing pul e p(t) to the total power received in a multipath delay profile.

Ba ed on maller than the o be deconvolved ultipath ignal given by

work by Cox [Cox72], [Cox75], if p(t) ha a time duration much impul e re pon e of the multipath channel, p(t) doe not need t from the received ignal r(t) inn order to determine relative m trength . The received power delay profile in a local area i

For mall- cale channel modeling, the power delay of the channel i d by taking the patial average of hb(t; ) 2 over a local area. By making cal area mea urement of hb(t; ) 2 inn different location , it i po ible n en emble of power delay profile , each one repre enting a po ible mallmultipath channel tate [Rap91a].

foun everal lo to build a cale

When mea uring or predicting hb( ), a probing pul e p( ) which approximate a delta function i u ed at the tran mitter. That i ,

12. Explain How Mea urement are Made for Small-Scale propagation.

Becau e of the importance of the multipath tructure in determining the mall- cale fading effect , a number of wideband channel ounding technique hav e been developed. The e technique may be cla ified a direct pul e mea uremen t , pread pectrum liding correlate mea urement , and wept frequency mea urem ent . Direct RF Pul e Sy tem A imple channel ounding approach i the direct RF pul e y tem. Thi technique allow engineer to determine rapidly the power delay profile of any c hannel, a demon trated by Rappaport and Seidel [Rap89], [Rap90]. E entially a wideband pul ed bi tatic radar, thi y tem tran mit a repetitive pul e of wid th Tbb , and u e a receiver with a wide bandpa filter (BW=2/TbbHz). The ig nal i then amplified, detected with an envelope detector, and di played and to red on a high peed o cillo cope. Thi give an immediate mea urement of the q uare of the channel impul e re pon e convolved with the probing pul e. If the o cillo cope i et on averaging mode, then thi y tem can provide a local avera ge power delay profile. Another attractive a pect of thi y tem i y tem i t he lack of complexity, ince off-the- helf equipment may be u ed.

The minimum re olvable delay between multipath component i equal to th e probing pul e width Tbb. The main problem with thi y tem i that it i ubj ect to interference and noi e, due to the wide pa band filter required for mul tipath time re olution. Al o, the pul e y tem relie on the ability to trigger the o cillo cope on the fir t arriving ignal. If the fir t arriving ignal i blocked or fade , erver fading occur , and it i po ible the y tem may not t rigger properly. Another di advantage i that the pha e of the individual mult ipath component are not received, due to the u e of an envelope detector. Howe ver, u e of a coherent detector permit mea urement of the multipath pha e u ing thi technique. Figure: Direct RF channel impul e re pon e mea urement y tem. Spread pectrum Sliding Correlator Channel Sounding The ba ic block diagram of a pread pectrum channel ounding y tem i hown in Fig. The advantage of a pectrum y tem i that, while the probing ig nal may be wideband, it i po ible to detect the tran mitted ignal u ing a nar rowband receiver preceded by a wideband mixer, thu improving the dynamic range of the y tem a compared to direct RF pul e y tem. In a pread pectrum channel ounder, a carrier ignal i pread over a la rge bandwidth by mixing it with a binary p eudo-noi e (PN) equence having a chi p duration Tc and a chip rate Rc equal to 1/Tc Hz. The power pectrum envelope of the tran mitted pread pectrum ignal i given by [Dix84] a

The pread pectrum ignal i then received, filtered, and de pread u in g a PN equence generator identical to the u ed at the tran mitter. Although th e two PN equence are identical, the tran mitter chip clock i run at a lightl y fa ter rate than the receiver chip clock. Mixing the chip equence inn thi fa hion implement a liding correlator [Dix84]. When the PN code of the fa ter chip clock catche up with the PN code of the two chip equence will be virtua

Small-Scale Multipath Mea urement

lly identically aligned, giving maximal correlation. When the two equence are not maximally correlated, mixing the incoming pread pectrum ignal with the u n ynchronized receiver chip equence will pread thi ignal into a bandwidth at lea t a large a the receiver reference PN equence. In thi way, the narrowb and filter that follow the correlator can reject almo t all of the incoming ig nal power. Thi i how proce ing gain i realized in a pread pectrum receive r and how it can reject pa band interference, unlike the direct RF pul e oundi ng y tem.

where Tbb = 1/Rbb, i the period of the ba eband information. For the c a e of a liding correlator channel ounder, the ba eband information rate i eq ual to the frequency off et of the PN equence clock at the tran mitter and rec eiver. When the incoming ignal i correlated with the receiver equence, the ignal i collap ed back to the original bandwidth (i.e., de pread), envelope detec ted, and di played on an o cillo cope. Since different incoming multipath will have different time delay , they will maximally correlate with the receiver PN equence at different time . The energy of the e individual path will pa thro ugh the correlator depending on the time delay. Therefore, after envelope detec tion, the channel impul e re pon e convolved with the pul e hape of a ingle ch ip i di played on the o cillo cope. Cox [Cox72] fir t u ed thi method to mea ure channel impul e re pon e in outdoor uburban environment at 910 MHz. Deva irvatham [Dev86], [Dev90a] ucce fully u ed a direct equence pread pectrum channel ounder to mea ure time delay pread of multipath component and ignal level mea urement in office and re idential building at 850 MHz. Builtitude [ Bul89] u ed thi technique for indoor and microcellular channel ounding work, a did Landron [Lan92], while Newhall and Saldanha mea ured campu e and train ya rd [New96a]. A detailed de cription of a practical liding correlator i given in [New96b].

The liding correlation proce given equivalent time mea urement that are updated every time the two equence are maximally correlated. The time bet ween maximal correlation ( T) can be calculated from Equation (5)

The lide factor i defined a the radio between the tran mitter chip cl ock rate and the difference between the tran mitter and receiver chip clock rate [Dev86]. Mathematically, thi i expre ed a

where

Tc= chip period ( ) Rc=chip rate (Hz) = lide factor (dimen ionle ) l = equence length (chip )

a In be th

In other word , the y tem can re olve two multipath component a long they are equal to or greater than two chip duration , or 2Tc econd apart. actuality, multipath component with interarrival time maller than 2Tc can re olved ince the rm pul e width of a chip i maller than the ab olute wid of the triangular correlation pul e, and i on the order of Tc.

The time re olution ( ) of multipath component u ing a pread h liding correlation i

pectrum wit

Proce ing gain (PG) i given a

l = 2n 1

(7) equence generator [Dix84].

Since the incoming pread pectrum ignal i mixed with a receiver PN e quence that i lower than the tran mitter equence, the ignal i e entially d own-converted (collap ed) to a low-frequency narrowband ignal. In other word , t he relative rate of the two code lipping pa t each other i the rate of inform ation tran ferred to the o cillo cope. Thi narrowband ignal allow narrowband proce ing, eliminating much of the pa band noi e and interference. The proc e ing gain of Equation (3) i then realized u ing a narrowband filter (BW=2( - )). The equivalent time mea urement refer to the relative time of multipat h component a they are di played on the o cillo cope. The ob erved time cale on the o cillo cope u ing a liding correlator related to the actual propagation time cale by

Thi effect i due to the relative rate of information tran fer in the liding correlator. For example, T of Equation (5) i an ob erved time mea ured o n an o cillo cope and not actual propagation time. Thi effect, known a time d ilation, occur in the liding correlator y tem becau e the propagation delay are actually expanded in time by the liding correlator. Caution mu t be taken to en ure that the equence length ha a period wh ich i greater than the longe t multipath propagation delay. The PN equence pe riod i PN eq = Tcl (9)

The equence period give an e timate of the maximum unambiguou range o f incoming multipath ignal component . Thi range i found by multiplying the peed of light with PN eq in Equation (9).

There are everal advantage to the pread pectrum channel ounding y tem. One of the key pread pectrum modulation characteri tic i the ability t o reject pa band noi e, thu improving the coverage range for a given tran mit ter power. Tran mitter and receiver PN equence ynchronization i eliminated b y the liding correlator. Sen itivity i adju table by changing the liding fac tor and the po t-correlator filter bandwidth. Al o, required tran mitter power can be con iderably lower than comparable direct pul e y tem due to the inher ent Proce ing gain of pread pectrum y tem . A di advantage of the pread pectrum y tem, a compared to the direct pul e y tem, i that mea urement are not made in real time, but they are compl ied a the PN code lide pa t one another. Depending on y tem parameter and mea urement objective , the time required to make power delay profile mea uremen t may be exce ive. Another di advantage of the y tem de cribed here i that a nocoherent detector i u ed, o that pha e of individual multipath component can not be mea ured. Even if coherent detection i u ed, the weep time of a pread pectrum ignal induce delay uch that the pha e of individual multipath component with different time delay would be mea ured at ub tantially differ

where n i the number of hift regi ter in the

For a maximal length PN equence, the

where

= tran mitter chip clock rat (Hz) = receiver chip clock rate 9Hz) equence length i

ent time , during which the channel might change. Frequency Domain Channel Sounding Becau e of the dual relation hip between time domain and frequency domai n technique , it i po ible to mea ure the channel impul e re pon e in the freq uency domain. Figure hown a frequency domain channel ounder u ed for mea urin g channel impul e re pon e . A vector network analyzer control a ynthe ized f requency weeper, and an S-parameter te t et i u ed to monitor the frequency r e pon e of the channel. The weeper can a particular frequency band (centered on the carrier) by tepping through di crete frequencie . The number and paci ng of the e frequency tep impact the time re olution of the impul e re pon e mea urement. For each frequency tep, the S-parameter te t et tran mit a know n ignal level at port 1 and monitor the received ignal level at port 2. The e ignal level allow over the mea ured frequency range. The tran mi ivity re pon e i a frequency domain repre entation of the channel impul e re pon e. Thi re pon e i then converted to the time domain u ing inver e di crete Fourier t ran form (IDFT) proce ing, given a band-limited ver ion of the impul e re pon e . In theory, thi technique work well and indirectly provide amplitude and ph a e information in the time domain. However, the y tem require careful calibr ation and hardwired ynchronization between the tran mitter and receiver, making it u eful only for very clo e mea urement (e.g., indoor channel ounding). An other limitation with thi y tem i the non-real-time nature of the mea urement . For time varying channel , the channel frequency re pon e can change rapidly, given an erroneou impul e re pon e mea urement. To mitigate thi effect, fa t weep time are nece ary to keep the total wept frequency re pon e mea uremen t interval a hort a po ible. A fa ter weep time can be accompli hed by red ucing the number of frequency tep , but thi acrifice time re olution and exc e delay range in the time domain.

13. Determine the parameter of Mobile Multipath channel.

Many multipath channel parameter are derived from the power delay profi le, given by Equation. Power delay profile are mea ured u ing the technique d i cu ed in Section 5.4 and are generally repre ented a plot of relative recei ved power a a function of exce delay with re pect to a fixed time delay refer ence. Power delay profile are found by averaging in tantaneou power delay pro file mea urement over a local area in order to found by averaging in tantaneou power delay profile mea urement over a local area in order to determine an ave rage mall- cale power delay profile. Depending on the time re olution of the p robing pul e and the type of multipath channel tudied, re earcher often choo e to ample at patial eparation of a quarter of a wavelength and over receive

Parameter of Mobile Multipath Channel

Figure: Frequency domain channel impul e re pon e mea urement

y tem

r movement no greater than 6 m in outdoor channel and no greater than m in ind oor channel in the 450 MHz -6 GHz range. Thi mall- cale ampling avoid larg e- cale averaging bia in the re ulting mall- cale tati tic . Fig. how typi cal power delay profile plot from outdoor and indoor channel , determined from a large number of clo ely ampled in tantaneou profile . In order to compare different multipath channel and to develop ome gen eral de ign guideline for wirele y tem , parameter which gro ly quantify t he multipath channel are u ed. The mean exce delay, rm delay pread, and exc e delay pread (X dB) are multipath channel parameter that can be determined from a power delay profile. The time di per ive propertie of wide band multipa th channel are mo t commonly quantified by their mean exce delay and rm de lay pread ( ). The mean exce delay i the fir t moment of the power delay profi le and i defined to be

The rm delay pread i the quare root of the econd central moment of the power delay profile and i defined to be

The e delay are mea ured relative to the fir t detectable ignal arrivi ng at the receiver at 0=0. Equation (1)-(3) do not rely on the ab olute power le vel of P( ), but only the relative amplitude of the multipath component within P ( ). Typical value of rm delay pread are on the order of micro econd in outdo or mobile radio channel and on the order of nano econd in indoor radio channel . Table how the typical mea ured value of rm delay pread. It i important to note that the rm delay pread and mean exce delay are defined from a ingle power delay profile which i the temporal or patial a verage of con ecutive impul e re pon e mea urement collected and averaged over a local area. Typically, many mea urement are made at many local area in orde r to determine a tati tical range of multipath channel parameter for a mobile communication y tem over a large- cale area [Rap90].

The maximum exce delay (X dB) of the power delay profile i defined to be the time delay during which multipath energy fall to X dB below the maximum . In other word , the maxmum exce delay i defined a X- 0, where 0 i the fir t arriving ignal and X i the maximum delay at which a multipath component i with in X dB of the tronge t arriving multipath ignal (which doe not nece arily a rrive at 0). Fig. illu trate the computation of the maximum exce delay for mu ltipath component within 10 dB of the maximum. The maximum exce delay (X dB) define the temporal extent of the multipath that i above a particular thre ho ld. The value of X i ometime called the exce delay pread of a power delay profile, but in all ca e mu t be pecified with a thre hold that relate the mu ltipath noi e floor to the maximum received multipath component. Table: Typical Mea ured Value of RMS Delay Spread

10-25 Wor t ca e San Franci co [Rap90] 910 200-310 n Averaged typical ca e [Cox72] 910 1960-2110 n Averaged extreme ca e [Cox72] 10-50 n 25 n median Office building [Sal87] 270 n max. Office building [Dev90a] 70-94 n avg. 1470 n max. Three San Franci co building

Environment Urban 910 [Cox75] Urban 892 Suburban Suburban Indoor 1500 Indoor 850 Indoor 1900 [Sei92a]

Frequency (MHz) RMS Delay Spread ( t) Note 1300 n avg. 600 n t. dev. 3500 n max

Reference New York City

In practice, value for and depend on the choice of noi e thre hold u e d to proce P( ). The noi e thre hold i u ed to differentiate between received multipath component and thermal noi e. If the no e thre hold i et too low, t hen noi e will be proce ed a multipath, thu giving ri e to value of It hould be noted the power delay profile and the magnitude frequency r e pon e (the pectral re pon e) of a mobile radio channel are related through th e Fourier tran form. It i therefore po ible to obtain an equivalent de cripti on of the channel in the frequency domain u ing it frequency re pon e character i tic . Analogou to the delay pread parameter in the time domain, coherence bandwidth i u ed to characterize the channel in the frequency domain. The rm delay pread and coherence bandwidth are inver ely proportional to one another, although their exact relation hip i a function of the exact multipath tructure . Coherence Bandwidth While the delay pread i a natural phenomenon cau ed by reflected and cattered propagation path in the radio channel, the coherence bandwidth, Bc, i a defined relation derived from the rm delay pread. Coherence bandwidth i a tati tical mea ure of the range of frequencie over which the pread. Coheren ce bandwidth i a tati tical mea ure of the range of frequencie over which the channel can be con idered flat (i.e., a channel which pa e all pectral compone nt with approximately equal gain and linear pha e). In other word , coherence bandwidth i the range of frequencie over which two frequency component have a trong potential for amplitude correlation. Two inu oid with frequency epar ation greater than Bc are affected quite differently by the channel. If the coh erence bandwidth i defined a the bandwidth over which the frequency correlatio n function i above 0.9, then the coherence bandwidth i approximately [Lee89b]

It i important to note that an exact relation hip between coherence ban dwidth and rm delay pread i function of pecific channel impul e re pon e an d applied ignal , and Equation (4) and (5) are ball park e timate . In general, pectral analy i technique and imulation are required to determine the exact impact that time varying multipath ha on a particular tran mitted ignal [Chu 8 7], [Fun93]. For thi rea on, accurate multipath channel model mu t be u ed in the de ign of pecific modem for wirele application [Rap91a], [Woe94].

Doppler Spread and Coherence Time Delay pread and coherence bandwidth are parameter which de cribe the time di per ive nature of the channel in a local area. However, they do not off er information about the time varying nature of the channel cau ed by either rel ative motion between the mobile and ba e tation, or by movement of object in t he channel. Doppler pread and coherence time are parameter which de cribe the time varying nature of the channel in a mall- cale region.

Doppler pread BD i a mea ure of the

pectral broadening cau ed by the

If the definition i relaxed o that the frequency correlation function i 0.5, then the coherence bandwidth i approximately

above

Coherence time TC i the time domain dual of Doppler pread and i u ed to characterize the time varying nature of the frequency depre ivene of the c hannel in the time domain. The Doppler pread and coherence time are inver ely proportional to one another. That i ,

Coherence time i actually a tati tical mea ure of the time duration ov er which the channel impul e re pon e i e entially invariant, and quantifie t he imilarity of the channel re pon e at different time . In other word , coher ence time i the time duration over which two received ignal have a trong pot ential for amplitude correlation. If the reciprocal bandwidth of the ba eband ignal i greater than the coherence time of the channel, then the channel will c hange during the tran mi ion of the ba eband me age, thu cau ing di tortion a t the receiver. If the coherence time i defined a the time over which the tim e correlation function i above 0.5, then the coherence time i approximately.

The definition of coherence time implie that two ignal arriving with a time eparation greater than TC are affected differently by the channel. For example, for a vehicle traveling 60 mph u ing a 900 MHz carrier, a con ervative value of TC can be hown to be 2.22 m from Equation (6.a). If a digital tran m i ion y tem i u ed, then a long a the ymbol rate i di tortion could re ul t from multipath time delay pread, depending on the channel impul e re pon e). U ing the practical formula of (6.c), TC= 6.77 m the ymbol rate mu t exceed 1 50 bit / in order to avoid to di tortion due to frequency di per ion.

Type of Small-Scale Fading The type of fading experienced by a ignal propagating through a mobile radio channel depend on the nature of the tran mitted ignal with re pected ig nal with re pect to the characteri tic of the channel. Depending on the relati on between the ignal parameter ( uch a bandwidth, ymbol period, etc.) and th e channel parameter ( uch a rm delay pread and Doppler pread), different tr an mitted ignal will undergo different type of fading. The time di per ion an d frequency di per ion mechani m in a mobile radio channel lead to four po ibl e di tinct effect , which are manife ted depending on the nature of the tran mit ted ignal, the channel, and the velocity. While multipath delay pread lead t

14. Explain about the tati tical Model

for multipath Fading channel .

where fm i the maximum Doppler hift given by fm a time duration during which a Rayleigh fading d (6.b) i often too re trictive. A popular rule munication i to define the coherence time a the a) and (6.b). That i ,

= / . In practice, (6.a) ugge t ignal may fluctuate wildly, an of thump for modem digital com geometric mean of Equation (6.

time rate of change of the mobile radio channel and i defined a the range of f requencie over which the received Doppler pectrum i e entially non-zero. Wh en a pure inu oidal tone of frequency fc i tran mitted, the received ignal p ectrum, called the Doppler pectrum, will have component in the range fc-fd to fc+fd, where fd i the Doppler hift. The amount of pectral broadening depend on fd which i a function of the relative velocity of the mobile, and the angle between the direction of motion of the mobile and direction of arrival of the cattered wave . If the ba eband ignal bandwidth i much greater than BD, the e ffect of Doppler pread are negligible at the receiver. Thi i a low fading channel.

o time di per ion and frequency elective fading, Doppler pread lead to freque ncy di per ion and time elective fading. The two propagation mechani m are in dependent of one another. Fig. how a tree of the four different type of fadi ng.

Flat fading If the mobile radio channel ha a con tant gain and linear pha e re pon e over a bandwidth which i greater than the bandwidth of the tran mitted ignal , then the received ignal will undergo flat fading. Thi type of fading i hi torically the mo t common type of fading de cribed in the technical literature. In flat fading, the multipath tructure of the channel i uch that the pectra l characteri tic of the tran mitted ignal are pre erved at the receiver. Howe ver the trength of the received ignal change with time, due to fluctuation , in the gain of the channel cau ed by multipath. The characteri tic of a flat f ading channel are illu trated in Figure. It can be een from figure that if the channel gain change over time, a change of amplitude occur in the received ignal. Over time, the received ig nal r(t) varie in gain, but the pectrum of the tran mi ion i pre erved. In a flat fading channel, the reciprocal bandwidth Small-Scale Fading (Ba ed on multipath time delay pread)

Small-Scale Fading (Ba ed on Doppler pread)

Fa t Fading Slow Fading 1. High Doppler pread pler pread 2. Coherence time < Symbol period

1. Low Dop 2. Coherence t

Flat Fading cy Selective Fading 1. BW of ignal < BW of channel > BW of channel 2. Delay pread < Symbol period elay pread > Symbol period

Frequen 1. BW of ignal 2. D

Time di per ion due to multipath cau e o either flat or frequency elective fading.

Fading Effect

Due to multipath Time Delay Spread the tran mitted ignal to underg

ime> Symbol period 3. Channel variation fa ter than ba e riation lower than band ignal variation ba eband ignal variation Figure: Type of mall- cale fading

3. Channel va

where TS i the reciprocal bandwidth (e.g., ymbol) and BS i the bandwidth, re pectively of the tran mitted modulation, and and BC are the rm delay pread and coherence bandwidth, re pectively, of the channel.

Frequency Selective Fading If the channel po e e a con tant-gain and linear pha e re pon e over a bandwidth that i maller than the bandwidth of tran mitted ignal, then the c hannel create frequency elective fading on the received ignal. Under uch co ndition , the channel impul e re pon e ha a multipath delay pread which i gre ater than the reciprocal bandwidth of the tran mitted me age waveform. When th i occur , the received ignal include multiple ver ion the tran mitted wavefo rm which are attenuated (faded) and delayed in time, and hence the received ign al i di torted. Frequency elective fading i due to time di per ion of the tr an mitted ymbol within the channel. Thu the channel induce inter ymbol inte rference (ISI). Viewed in the frequency domain, certain frequency component in the received ignal pectrum have greater gain than other .

Figure: Flat fading channel characteri tic

To

ummarize, a ignal undergoe flat fading if

Of the tran mitted ignal i much larger than the multipath time delay pread of the channel, and hb(t, ) can be approximated a having no exce delay (i.e., a ingle delta function with =0). Flat fading channel are al o known a amplitude varying channel and are ometime referred to a narrowband channel , ince the bandwidth of the applied ignal i narrow a compared to the channel flat fading bandwidth. Typical flat fading channel cau e deep fade , and thu may require 20 or 30 dB more tran mitter power to achieve low bit error rate during time of deep fade a compared to y tem operating over non-fading channel . The di tribution of the in tantaneou gain of flat fading channel i important for de igning radio link , and the mo t common amplitude di tribution i the Rayleigh di tribution. The Reyleigh flat fading channel model a ume that the channel i nduce and amplitude which varie in time according to the Rayleigh di tribution .

Frequency elective fading channel are much more difficult to model tha n flat fading channel ince each multipath ignal mu t be modeled and the chann el be con idered to be a linear filter. It i for thi rea on that wideband mul tipath mea urement are made, and model are developed from the e mea urement . When analyzing mobile communication y tem , tati tical impul e re pon e model to be made uch a the two-ray Rayleigh fading model (which con ider the impu l e re pon e to be made up of two delta function which independently fade and h ave ufficient time delay between them to induce frequency elective fading upon the applied ignal), or computer generated or mea ured impul e re pon e , are g enerally u ed for analyzing frequency elective mall- cale fading. Fig. illu tr ate the characteri tic of a frequency elective fading channel. For frequency elective fading, the pectrum S(f) of the tran mitted ig nal ha a band width which i greater than the coherence bandwidth BC of the cha nnel. Viewed in the frequency domain, the channel become frequency elective, where the gain i different for different frequency component . Frequency elec tive fading i cau ed by multipath delay which

Approach or exceed the ymbol period of the tran mitted ymbol. Frequency elec tive fading channel are al o known a wideband channel ince the bandwidth of the ignal S(t) i wider than the bandwidth of the channel impul e re pon e. A time varie , the channel varie in gain and pha e acro the pectrum of S(t), re ulting in time varying di tortion in the received ignal r(t). To ummarize, a ignal undergoe frequency elective fading if BS > BC and TS < (10) A common rule of thumb i that a channel i flat fading if TS 10 and a channel i frequency elective if TS < 10 , although thi i dependent on the pecific type o f modulation u ed. Chapter 6 pre ent imulation re ult which illu trate the i mpact of time delay pread on bit error rate (BER). (9)

Fa t Fading Depending on how rapidly the tran mitted ba eband ignal change a comp ared to the rate of change of the channel, a channel may be cla ified either a a fa t fading or low fading channel. In a fa t fading channel, the channel im pul e re pon e change rapidly within the ymbol duration. That i , the coheren ce time of the channel i maller than the ymbol period of the tran mitted ign al. Thi cau e frequency di per ion (al o called time elective fading) due to Doppler preading, which lead to ignal di tortion . Viewed in the frequency d omain, ignal di tortion due to fa t fading increa e with increa ing Doppler p read relative to the bandwidth of the tran mitted ignal. Therefore, a ignal un dergoe fa t fading if TS > TC and BS < BD (12) (11)

It hould be noted that when a channel i pecified a a fa t or low fa ding channel, it doe not pecify whether the channel i flat fading or frequenc

Fading Effect

Due to Doppler Spread

Figure: Frequency

elective fading channel characteri tic

y elective in nature. Fa t fading only deal with the rate of change of the ch annel due to motion. In the ca e of the flat fading channel, we can approximate the impul e re pon e to be imply a delta function (no time delay). Hence, a f a ter than the rate of the fading channel i a channel in which the amplitude of the delta function varie fa ter than the rate of change of the tran mitted ba eband ignal. In the ca e of a frequency elective, fa t fading channel, the a mplitude , pha e , and time delay of any one of the multipath component vary f a ter than the rate of change of the tran mitted ignal. In practice, fa t fadi ng only occur for very low data rate . Slow Fading

In a low fading channel, the channel impul e re pon e change at a rate much lower than the tran mitted ba eband ignal (t). In thi ca e, the chann el may be a umed to be tatic over one or everal reciprocal bandwidth interval . In the frequency domain, thi implie that the Doppler pread of the channel i much le than the bandwidth the ba eband ignal. Therefore, a ignal under goe low fading if

It hould be clear that the velocity of the mobile (or velocity of objec t in the channel) and the ba eband ignaling determine whether a ignal underg oe fa t fading or low fading. The relation between the variou multipath parameter and the type of fa ding experienced by the ignal are ummarized in Fig. Over the year , ome auth or have confu ed the term fa t and low fading with the term large- cale and mall fading. It hould be empha ized that fa t and low deal with the relation hip between the time rate of change in the channel and the tran mitted ignal, and not with propagation path lo model .

Several multipath model have been ugge ted to explain the ob erved ta ti tical natural of a mobile channel. The fir t model pre ented by O ana wa b a ed on interference of wave incident and reflected from the flat ide of rand omly located building . Although o ana model predict flat fading power pectr a that were in agreement with mea urement in uburban area , it a ume the exi tence of a direct path between the tran mitter and receiver, and i limited to a re tricted range of reflection angle . O ana model i therefore rather infle xible and inappropriate for urban area where the direct path i almo t alway b locked by building or other ob tacle . Clarke model [Cla68] i ba ed on catte ring and i widely u ed. Clarke Model for Flat Fading

Clarke [Cla68] developed a model where the tati tical characteri tic o f the electromagnetic field of the received ignal at the mobile are deduced fr om cattering. The model a ume a fixed tran mitter with a vertically polarize d antenna. The field incident on the mobile antenna i a umed to be compri ed on N azimuthal plane wave with arbitrary carrier pha e , arbitrary azimuthal ang le of arrival, and each wave having equal average amplitude. It hould be note d that the equal average amplitude a umption i ba ed on the fact that in the a b ence of a direct line-of- ight path, the cattered component arriving at a re ceiver will experience imilar attenuation over mall- cale di tance .

Stati tical Model for Multipath Fading Channel

15. Explain about the

tati tical Model

for multipath fading Channel .

Figure how a diagram of plane wave incident on a mobile traveling at a velocity , in the x-direction. The angle of arrival i mea ured in the x-y plan e with re pect to the direction of motion. Every wave that i incident on the m obile undergoe a Doppler hift due to the motion of the receiver and arrive at the receiver at the ame time. That i , no exce delay due to multipath i a umed for any of the wave . For the nth wave arriving at an angle n to the x-axi , the Doppler hift in Hertz i given by

The vertically polarized plane wave arriving at the mobile have E and f ield component given by where E0 i the real amplitude of local average E-field (a umed con tant), Cn i a real random variable repre enting the amplitude of individual wave , i the intrin ic impedance of free pace (377 ), and fc i the carrier frequency. The ra ndom pha e of the nth arriving component n i given by

The amplitude of the E-and H-field are normalized uch that the en embl e average of the CnS i given by

Both TC(t) and TS(t) are Gau ian random proce which are denoted a TC and TS, re pectively, at any time t. TC and TS are uncorrelated zero-mean Gau ian random variable with an equal variance given by

where the over bar denote the en emble average. The envelope of the received E-field, Ez(t), i given by

Since TC and TS are Gau ian random variable , it can be hown through a Jacobean tran formation [Pap91] that the random received ignal envelope r ha a Rayleigh di tribution given by

Ga [Gan72] developed a

pectrum analy i for Clarke model. Let p( ) d

Spectral Shape Due to Doppler Spread in Clarke

Model den

Since the Doppler hift i very mall when compared to the carrier frequ ency, the three field component may be modeled a narrow band random proce e . The three component Ez, Hx, and Hy can be approximated a Gau ian random var iable if N i ufficiently large. The pha e angle are a umed to have a unifo rm probability den ity function (pdf) on the interval (0.2 ). Ba ed on the analy i by Rice [Ric48] the E-field can be expre ed in an in-pha e and quadrature fo rm

where

the wavelength of incident wave.

Figure: Illu trating plane wave arriving at random angel

ote the fraction of the total incoming power within d of the angle , and let A den ote the average received power with re pect to an i otropic antenna. A N p ( ) a pproache a continuou , rather than a di crete, di tribution. If G( ) i the azim uthal gain pattern of the mobile antenna a a function of the angle of arrival, the total received power can be expre ed a where AG( ) p ( ) d i the differential variation of received power with angle. If th e catter ignal i a CW ignal of frequency fc, then the in tantaneou frequenc y of the received ignal component arriving at an angle i obtained u ing Equati on. (1)

If S(f) i the power pectrum of the received ignal, the differential v ariation of received power with frequency i given by

Equating the differential variation of received power with frequency to the differential variation in received power with angle, we have

Differentiating Equation n(14), and rearranging the term , we have

The pectrum i centered on the carrier frequency and i zero out ide th e limit of fc fm. Each of the arriving wave ha it own carrier frequency (du e to it direction of arrival) which i lightly off et from the center frequenc y. For the ca e of a vertical /4 antenna (G( )=1.5), and a uniform di tribution p( ) =1/2 , the output pectrum i given by (20) a

In Equation (22), the power pectral den ity at f=fc fm i infinite, i.e., Doppler component arriving at exactly 0 and 180 have an infinite power pectral den ity. Thi i not a problem ince i continuou ly di tributed and the probab ility of component arriving at exactly the e angle i zero. Figure how the power pectral den ity of the re ulting RF ignal due t o Doppler fading. Smith [Smi75] demon trated an ea y way to imulate Clarke mod el u ing a computer imulation a de cribed. After envelope detection of the Doppler- hifted ignal, the re ulting ba eband pectrum ha a maximum frequency of 2fm. It can be hown [Jak74] that th e electric field produce a ba eband power pectral den ity given by

Sub tituting Equation (17) and (19) into both ide (16), the power ity S(f) can be expre ed a

U ing Equation (14),

can be expre ed a a function of f a

pectral den

where fm i the maximum Doppler hift. It hould be noted that F( ) i nction of ., (i.e., f( ) = f (- ).

an even fu

The pectral hape of the Doppler pread determine the time domain fadi ng waveform and dictate the temporal correlation and fade lope behavior . Ray leigh fading imulator mu t u e a fading pectrum uch Equation (22) in order t o produce reali tic fading waveform that have proper time correlation.

Figure: Doppler power pectrum for an unmodulated CW carrier [from [Gan72] IEEE] . Figure: Ba eband power pectral den ity of a CW Doppler ignal after envelope de tection. Simulation of Clarke and Gan Fading Model

A hown in Figure (b), two independent Gau ian low pa ource are u ed to produce in-pha e and quadrature fading branche . Each Gau ian ource may be formed by umming two independent Gau ian random variable which are orthog onal (i.e., g = a+jb, where a and b are real Gau ian random variable and g i complex Gau ian). By u ing the pectral filter defined by Equation (22) to ha pe the random ignal in the frequency domain, accurate time domain waveform of Doppler fading can be produced by u ing an inver e fa t Fourier tran form (IFFT) at the la t tage of the imulator. Smith [Smi75] demon trated a imple computer program that implement Fig .(b). Hi method u e a complex Gau ian random number generator (noi e ource) to produce a ba eband line pectrum with complex weight in the po itive freque ncy band. The maximum frequency component of the line pectrum i fm. U ing th e property of real ignal , the negative frequency component are con tructed by imply conjugating the complex Gau ian value obtained for the po itive freque ncie . Note that the IFFT of each complex Gau ian ignal hould be a purely re al Gau ian random proce in the time domain which i u ed in each of the quadr ature arm hown in Figure. The random valued line pectrum i then multiplied with a di crete frequency repre entation of having the ame number of point a the noi e ource. To handle the ca e where Equation (22) approache infinity at the pa band edge, Smith truncated the value of SEz(f) by computing the lop e of the function at the pa band edge. Simulation u ing the architecture in Figure are u ually implemented in the frequency domain u ing complex Gau ian li ne pectra to take advantage of ea y implementation of Equation (22). Thi , in turn, implie that the low pa Gau ian noi e component are actually a erie of frequency component (line pectrum from fm to fm), which are equally paced a nd each have a complex Gau ian weight. Smith methodology i hown in Figure.

It i often u oftware. A popular e modulation path to pectral and temporal

eful to imulate multipath imulation method u e the produce a imulated ignal characteri tic very clo e

fading channel in hard ward or concept of in-pha e and quadrat repre enting Equation (7) with to mea ured data.

where not when band

K[] i the complete elliptical integral of the fir t kind. Equation (23) i intuitive and i re ult of the temporal correlation of the received ignal pa ed through a nonlinear envelope detector. Figure illu trate the ba e pectrum of the received ignal after envelope detection.

Figure: Simulator u ing quadrature amplitude modulation with (a) RF Doppler fil ter and (b) ba eband Doppler filter.

5.23(221) Figure: Simulator u ing quadrature amplitude modulation with (a) RF Doppler filt er and (b) ba eband Doppler filter.

5.24(222)

Figure: Frequency domain implementation of a Rayleigh fading imulator at ba eba nd.

1. Specify the number of frequency domain point (N) u ed to repre ent an d the maximum Doppler frequency hift (fm). The value u ed for N i u ually a p ower of two. 2. Compute the frequency pacing between adjacent pectral line a f=2fm / (N-1). Thi define the time duration of a fading waveform, T=1/ f. 3. Generate complex Gau ian random variable for each of the N/2 po itive frequency component of the noi e ource. 4. Con truct the negative frequency component of the noi e ource by conju gating po itive frequency value and a igning the e at negative frequency value . 5. Multiply the in-pha e and quadrature noi e ource by the fading pectrum 6. Perform an IFFT on the re ulting frequency domain ignal from the in-ph a e and quadrature arm to get two N-length time erie , and add the quare of each ignal point in time to create an N-point time erie like under the radica l of Equation (11). Note that each quadrature arm hould be a real ignal after the IFFT to model Equation (7). 7. Take the quare root of the um obtained in Step 6 to obtain an N-point

To implement the imulator hown in Figure, the following tep

are u ed:

Several Rayleigh fading imulator may be u ed in conjunction with varia ble gain and time delay to produce frequency elective fading effect . Thi i hown in Figure. By making a ingle frequency component dominant in amplitude within an d at f=0, the fading i changed from Rayleigh to Ricean. For a multipath fading imulator with many re olvable component , thi method can be u ed to alter the probability di tribution of the individual multipath component in the imulat or of Fig. One mu t take care to properly implement the IFFT uch that each arm of Figure produce a real time domain ignal a given by TC(f) and TS(t) in Equ ation (8) and (9).

Figure: A ignal may be applied to a Rayleigh fading imulator to determine perf ormance in a wide range of channel condition . Both flat and frequency electiv e fading condition may be imulated, depending on gain and time delay etting . To determine the impact of flat fading on an applied ignal S(t), one me rely need to multiply the applied ignal by r(t), the output of the fading imu lator. To determine the impact of more than one multipath component, convolutio n mu t be performed a hown in Figure.

Rice computed tati tic for a mathematical problem which i imilar to Clarke fading model [Cla68], and thereby provided imple expre ion for computi ng the average number of level cro ing and the duration of fade . The level cr o ing rate (LCR) and average fade duration of an Rayleigh fading ignal are two important tati tic which are u eful for de igning error control code and div er ity cheme to be u ed in mobile communication y tem , ince it become po ible to relate the time rate of change of the received ignal to the ignal leve l and velocity of the mobile. The level cro ing rate (LCR) i defined a the expected rate at which t he Rayleigh fading envelope, normalized to the local rm ignal level, cro e a pecified level in a po itive-going direction. The number of level cro ing pe r econd i given by where i the time derivative or r(t) (i.e., the lope). P(R, ) i the joint den ity function of r and at r=R,fm i the maximum Doppler frequency, a nd = R/Rrm i the value of the pecified level R, normalized to the local rm a mplitude of the fading envelope [Jak74]. Equation (24) give the value of NR, t he average number of level cro ing per econd at pecified R. The level cro i ng rate i a function of the mobile peed a i apparent from the pre ence of fm in Equation (24). There are few cro ing at both high and low level , with the maximum rate occurring at = 1/ (i.e., at a level 3 dB below the rm level). T he ignal envelope experience very deep fade only occa ionally, but hallow fa

Level Cro ing and Fading Stati tic

time erie of a imulated Rayleigh fading ignal with the proper Doppler and time correlation.

pread

de are frequent. The average fade duration i defined a the average period of time for w hich the received ignal i below a pecified level R. For a Rayleigh fading i gnal, thi i given by

where i i the duration of the fade and T i the ob ervation interval of the fadi ng ignal. The probability that the received ignal r i le than the thre hol d R i found from the Reyleigh di tribution a where P(r) i the pdf of a Rayleigh di tribution. Thu , u ing Equation (23), (2 4), and (25), the average fade duration a a function of and fm can be expre ed a The average duration of a ignal fade help determine the mo t likely nu mber of ignaling bit that may be lo t during a fade. Average fade duration pr imarily depend upon the peed of the mobile, and decrea e a the maximum Doppl er frequency fm become large. If there i a particular fade margin built into the mobile communication y tem, it i appropriate to evaluate the receiver perf ormance by determining the rate at which the input ignal fall below a given le vel R, and how long it remain below the level, on average. Thi i u eful for relating SNR during a fade to the in tantaneou BER which re ult . Two-ray Rayleigh Fading Model Clarke model and the tati tic for Rayleigh fading are for flat are for flat fading condition and do not con ider multipath time delay. In modern mob ile communication y tem with high data rate, it ha become nece ary to model the effect of multipath delay pread a well a fading. A commonly u ed multip ath model i an independent Rayleigh fading two-ray model (which i a pecific i mplementation of the generic fading imulator hown in Figure). Figure how a block diagram of the two-ray independent Rayleigh fading channel model. The imp ul e re pon e of the model i repre ented a

where 1 and 2 are independent and Rayleigh di tributed, 1 and 2 are independent and uniformly di tributed over [0,2 }, and i the time delay between the two ray . By etting 2 equal to zero, the pecial ca e of a flat Rayleigh fading channel i o btained a By varying , it i po ible to create a wide range of frequency elective fading effect . The proper time correlation propertie of the Rayleigh random v ariable 1 and 2 are guaranteed by generating two independent waveform , each prod uced from the inver e Fourier tran form of the pectrum de cribed.

where Pr[r given by

R] i

the probability that the received ignal r i

le

than R and i

Figure: Two-ray Rayleigh fading model. Saleh and Valenzueia Indoor Stati tical Model Saleh and Valenzuela {Sal87] reported the re ult of indoor propagation mea urement between two vertically polarized omni directional antenna located on the ame floor of a medium ized office building. Mea urement were made u i ng 10n , 1.5 GHz, radar-like pul e . The method involved averaging the quare l aw detected pul e re pon e while weeping the frequency of the tran mitted pul e . U ing thi method, multipath component within 4n were re olvable. The re ult obtained by Saleh and Valenzuela how that; (a) the indoor c hannel i qua i tatic or very lowly time varying, and (b) the tati tic of the channel impul e re pon e are independent of tran mitting and receiving antenna polarization, if there i no line-of- ight path between them. They reported a m aximum multipath delay pread of 100n to 200n within the room of a building, and 300n in hallway . The mea ured rm delay pread within room had a median of 25n and a maximum of 50n . The large- cale path lo with no line-of- ight path wa found to vary a 60 dB range and obey a log-di tance power law ( ee Equa tion.) With an exponent between three and four. Saleh and Valenzuela developed a imple multipath model for indoor chann el ba ed on mea urement re ult . The model a ume that the multipath componen t arrive in clu ter . The amplitude of the received component are independen t Reyleigh random variable with variance that decay exponentially with clu ter delay a well a exce delay within a clu ter. The corre ponding pha e angle are independent uniform random variable over [0.2 ]. The clu ter and multipath component within a clu ter from Poi on arrival proce e with different rate . The clu ter and multipath component within a clu ter have exponentially di tributed interarrival time . The formation of the clu ter i related to the b uilding tructure, while the component within the clu ter are formed by multipl e reflection from object in the vicinity of the tran mitter and the receiver.

Rappaport and Seidel [Rap91a] reported mea urement at 1300 MHz in five factory building and carried out ub equent mea urement in other type of buil ding . The author developed an elaborate, empirically derived tati tical mode l to generate mea ured channel ba ed on the di crete impul e re pon e channel m odel of Equation. And wrote a computer program called SIRCIM (Simulation of Indo or channel impul e re pon e mea urement [Rap91a]. Sub equent work by Huang pro duced SMRCIM (Simulation of Mobile Radio Channel Impul e re pon e Model ), a im ilar program that generate mall- cale urban cellular and microcellular channel impul e re pon e [Rap93a]. The e program are currently in u e at over 100 in titution throughout the world, and have been update to include angle of arrival information for micro cell, indoor channel [Nuc99], [Lib99]. By recording power delay profile impul e re pon e at /4 interval on a 1 m track at many indoor mea urement location , the author were able to characte

SIRCIM and SMRCIM Indoor and Outdoor Stati tical Model

rize local mall- cale fading of individual multipath component , and the mallcale variation in the number and arrival time of multipath component within a local area. Thu , the re ulting tati tical model are function of multipath time delay bin i, the mall- cale receiver pacing, Xi, within a 1 m local area, the topography Sm which i either i either line-of- ight (LOS) or ob tructed, t he large- cale T-R eparation di tance Dn, and the particular mea urement locati on Pn. Therefore, each individual ba eband power delay profile i expre ed in a manner imilar to Equation. Expect the random amplitude and time delay are r andom variable which depend on the urrounding environment. Pha e are ynthe ized u ing a p eudo-determini tic model which provide reali tic re ult , o tha t a complete time varying complex ba eband channel impul e re pon e hb (t; i) may be obtained over a local area through imulation. In Equation (31), i the average multipath receiver power within a di crete ex ce delay interval of 7.8125 n . The mea ured multipath delay in ide open-plan building ranged from 40 n to 800 n . Mean multipath delay and rm delay pread value ranged from 30 n to 300 n , with median value of 96 n in LOS path and 105 n in ob tructed pat h . Delay pread were found to be uncorrelated with T-R eparation but were aff ected by factory inventory, building con truction material , building age, wall location , and ceiling height . Mea urement in a food proce ing factory that manufacture dry-good and ha con iderably le metal inventory than other fac torie had an rm delay pread that wa half of tho e ob erved inn factorie pro ducing metal product . Newer factorie which incorporate teel beam and teel r einforced concrete in the building tructure have tronger multipath ignal and le attenuation than older factorie which u ed wood and brick for perimeter w all . The data ugge ted that radio propagation in building may be de cribed b y a hybrid geometric/ tati tical model that account for both reflection from w all and ceiling and random cattering from inventory and equipment. By analyzing the mea urement from fifty local area in many building , it wa found that the number of multipath component , Np, arriving at a certain location of X, Sm, and Pn, and almo t alway ha a Gau ian di tribution. The a verage number of multipath component range between 9 and 36, and i generated b a ed on an empirical fit to mea urement . The probability that a multipath comp onent will arrive at a receiver at a particular exce delay Ti in a particular environment Sm i denoted a PR (Ti,Sm). Thi wa found from mea urement by co unting the total number of detected multipath component at a particular di cret e exce delay time, and dividing by the total number of po ible multipath comp onent for each exce delay interval. The probabilitie for multipath arriving at a particular exce delay value may be modeled a piecewi e function of ex ce delay, and are given by where S1 corre pond to the LOS topography, and S2 corre pond to ob tru cted topography. SIRCIM u e the probability of arrival di tribution de cribed by Equation (32) or (33) along with the probability di tribution of the number of multipath component , NP(X,Sm,Pn), to imulate power delay profile over mal l- cale di tance. A recur ive algorithm repeatedly compare Equation (32) or (3 3) with a uniformly di tributed random variable until the proper NP i generated for each profile [Hua91], [Rap91a]. Figure how an example of mea ured power delay profile at 19 di crete receiver location along a 1 m track, and illu trate accompanying narrowband in formation which SIRCIM compute ba ed on ynthe ized pha e for each multipath c omponent [Rap91a]. Mea urement reported in the literature provide excellent ag reement with impul e re pon e predicted by SIRCIM.

U ing imilar tati tical modeling technique , urban cellular and microc ellular multipath mea urement data from [Rap90], [Sei91a] were u ed to develop S MRCIM. Both large cell and microcell model were developed. Fig. how an exam ple of SMRCIM output for an outdoor micro cell environment.

16. A ume a receiver i located 10 km from a 50 W tran mitter. The carrier freq uency i 900 MHz, free pace propagation i a umed, Gt = 1, and Gr = 2, find (a ) the power at the receiver, (b) the magnitude of the E-field at the receiver an tenna, (c) the rm voltage applied to the receiver input a uming that the recei ver antenna ha a purely real impedance of 50 and i matched to the receiver. Solution:Given: Tran mitter power, Pt = 50 W Carrier frequency, fc = 900 MHz Tran mitter antenna gain, Gt = 1 Receiver antenna gain, Gr = 2 Receiver antenna re i tance = 50

17. Determine the proper patial ampling interval required to make mall cale propagation mea urement which a ume that con ecutive ample are highly correl ated in time. How many ample will be required over 10 m travel di tance if fc = 1900 MHz and v = 50 m/ . How long would it take to make the e mea urement , a uming they could be made in real time from a moving vehicle? What i the Dopple r pread BD for the channel? Solution:For correlation, en ure that the time between ample i equal to TC/2, and u e the malle t value of TC for con ervative de ign.

TC = 565

(c) The applied rm voltage at the receiver input i

(b) The magnitude of the received E-field i

(a) The power received at di tance d = 10 km i

The Doppler pread

UNIT III WIRELESS TRANSCEIVERS PART A 1. Define Modulation. Modulation i the proce of encoding information from me age ource in a manne r uitable for tran mi ion. It generally involve tran lating a ba eband me ag e ignal to bandpa ignal at frequencie that are very high when compare to ba nd frequency. The band pa ignal i called a Modulating ignal. 2. Define Demodulation. Demodulation i the proce of extracting the ba eband me age from the carrier o that it may be proce ed and interpreted by the intended receiver. 3. Define Frequency Modulation.

Frequency Modulation i the mo t popular analog modulation technique u e d in mobile y tem . In FM, the amplitude of the modulated carrier ignal i kep t con tant while it frequency i varied by the modulating me age ignal. Thu , FM ignal have all their information in the pha e or frequency of the carrier. 4. Define Amplitude Modulation. In Amplitude Modulation cheme , there i a linear relation hip between the quality of received ignal and the power of the received ignal ince AM ig nal uperimpo e the exact relative amplitude of the modulating ignal onto the carrier. Thu , AM ignal all their information in the amplitude of the carrier . 5. Define off et QPSK?

A modified form of QPSK, called off et QPSK(OQPSK) or le u ceptile to the e deleteriou effect and upport more fication. That i , OQPSK en ure there are fewer ba eband ignal ied to the RF amplifier, which help eliminate pectrum regrowth tion.

taggered QPSK i efficient ampli tran ition appl after amplifica

The time taken to make thi mea urement i

equal to

Therefore, the number of

ample required over a 10 m travel di tance i

Taking time ample pling interval of

at le

than half TC, at 282.5

corre pond to a patial am

6.What i /4 QPSK?

7.What are constant envelo e modulation? Binary Frequency Shift Keying Minimum Shift Keying Gaussian Minimum Shift Keying 8.What are the conditions of constant envelo e modulation ? 1)Power efficient class C am lifiers can be used without in troducing degradation in the s ectrum occu ancy of the transmitted signal. 2)Low out-of-band radiation of the order of -60 dB to -70 dB can be achieved. 3)Limiter-discriminator detection can be used, which sim lifies receiver design and rovides high immunity against random FM noise and si gnal fluctuations due to Rayleigh fading. 9.What are the advantages of constant envelo e modulation? 1)They occu y larger bandwidth than linear modulation sche mes. 2)In situations where bandwidth efficiency is more im orta nt than ower efficiency,constant envelo e modulation is not well-suited. 10. Define BFSK? In binary frequency shift keying(BFSK),the frequency of a constant am litude carrier signal is switched between two values according to t he two ossible message states(called high and low tones),corres onding to a bin ary 1 or 0.De ending on how the frequency variations are im arted into transmitt ed waveform,the FSK signal will have a either a discontinuous hase or continous hase between bits. 11. Define Ca ture Effect. The ca ture effect is direct result of the ra id nonlinear im rovement i n received quality for an increase in received ower. If two signals in the same frequency band are available at an FM receiver, the one a earing at the higher received signal level is acce ted and demodulated, while the weaker one is reje cted. 12. Define Frequency discriminators. The RF signal received, am lified, and filtered at the carrier and then converted to an intermediate frequency (IF) which contains the same s ectrum as the original received signal. 13. Draw the block diagram of slo e detector in FM demodulator.

The /4 shifted QPSK ique which offers a com romise ximum hase transition. It may ion. In /4 QPSK, the maximum PSK and 90 for OQPSK.

modulation is a quadrature hase shift keying techn between OQPSK and QPSK in terms of the allowed ma be demodulated in a coherent or non coherent fash hase change is limited to 135,as com ared to 180 for Q

14. What are the advantage of digital modulation? Digital Modulation offers many advantages over analog modulation. Some a dvantages include greater noise immunity and robustness to channel im airments, easier multi lexing of various forms of information(voice, data, and video) and greater security.

Power efficiency describes the ability of a modulation technique to res erve the fidelity of the digital message at low level ower. In a digital commun ication system, in order to increase noise immunity, it is necessary to increase the signal ower. 16. Define Band width efficiency. Bandwidth efficiency describes the ability of a modulation schemes to ac commodate data within a limited bandwidth. 17. Define absolute band width. The absolute bandwidth of a signal is defined as the range of frequencie s over which the signal has a non zero power pectral den ity. For ymbol repre ented a rectangular ba eband pul e , the PSD ha a ( in f)2/ f2 profile which extend over an infinite range of frequencie , and ha an ab olute bandwidth of infinity. 18. Define Null to Null Band width. A impler and more widely accepted mea ure of bandwidth i the fir t nul l to null bandwidth. The null to null bandwidth i equal to the width of the mai n pectral lobe. 19. Define Half power Band width. The half power bandwidth i defined a the interval between frequencie at which the PSD ha dropped to half power, or 3 dB below the peak value. Half p ower bandwidth i al o called a the 3dB bandwidth. 20. Define Digital Modulation type .

21. Define Linear Modulation technique . linear Modulation Technique are bandwidth efficient and hence are very attractive for u e in wirele communication y tem where there i an increa in g demand to accommodate more and more u er within a limited pectrum.

ed a

S(t)= Re [Am(t) exp(j2 fct)]

The Linear modulation cheme, the tran mitted ignal

(t) can be expre

(i)

Linear Modulation Technique . (ii) Non linear Modulation Technique

15. Define

ower efficiency.

22. Define QPSK.

23. Define MSK. Minimum hift keying ( MSK ) i a pecial type of continuou pha e frequ ency hift keying (CPFSK) wherein the peak frequency deviation i equal to the b it rate. MSK i continuou pha e FSK with a modulation index of 0.5.

PART B 1. Explain Minimum Shift Keying. Minimum Shift Keying (MSK) Minimum hift keying (MSK) i a pecial type of continuou pha e-frequen cy hift keying (CPFSK) wherein the peak frequency deviation i equal to the bi t rate. In other word , MSK i continuou pha e FSK with a modulation index of 0.5. The modulation index of an FSK ignal i imilar to the FM modulation inde x, and i defined a kFSK = (2 F)/Rb, where F i the peak RF frequency deviation an d Rb i the bit rate. A modulation index of 0.5 corre pond to the minimum freq uency pacing that allow two FSK ignal to be coherently orthogonal, and the na me minimum hift keying implie the minimum frequency eparation (i.e., bandwidt h) that allow orthogonal detection. Two FSK ignal H(t) and L(t) are aid to be orthogonal if

of BPSK,

Quadrature pha e hift keying (QPSK) ha twice the bandwidth efficiency ince 2 bit are tran mitted in a ingle modulation ymbol.

Figure: Block diagram of nocoherent FSK receiver. MSK i ometime referred to a fa t FSK, a the frequency pacing u ed i only half a much a that u ed in conventional nonoherent FSK [Xio94]. MSK i pectrally efficient modulation cheme and i particularly attrac tive for u e in mobile radio communication y tem . It po e e propertie uc h a con tant envelope, pectral efficiency, good BER performance, and elf- ync hronizing capability. An MSK ignal can be thought of a a pecial form of OQPSK where the ba eband rectangular pul e are replaced with half- inu oidal pul e [Pa 79]. The e pul e have hape like the St. Loui arch during a period of 2Tb. Con ider th e OQPSK ignal with the bit tream off et a hown in figure. If half- inu oid al pul e are u ed in tead of rectangular pul e , the modified ignal can be def ined a MSK and for an N-bit tream i given by

and where mli (t) and m Qi (t) are the odd and even bit of the bipolar erial data tream which have value of 1 and which feed the in-pha e and quadrate arm of t he modulator at a rte of Rb/2. It hould be noted that there are a number of va riation of MSK that exi t in the literature [Sun86]. For example, while one ve r ion of MSK u e only po itive half- inu oid a the ba ic pul e hape, another ver ion u e alternating po itive and negative half- inu oid a the ba ic pul e hape. However, all variation of MSK are continuou pha e FSK employing diff erent technique to achieve pectral efficiency [Sun86]. The MSK waveform can be een a a pecial type of a continuou pha e FSK if Equation (2) i rewritten u ing trigonometric identitie a

where k i 0 or depending on whether m1(t) i or-1. From Equation (4), it can be deduced that MSK ha con tant amplitude. Pha e continuity at the bit tran ition period i en ured by choo ing the carrier frequency to be an integral multiple of one fourth the bit rate, 1/4T. Comparing Equation (4) with Equation, it can be concluded that the MSK ignal i an FSK ignal with binary ignaling frequen cie of fc+1/4T and fc-1/4T. It can further be een from Equation (4) that the pha e of the MSK ignal varie linearly during the cour e of each bit period [Pr o94, Chapter 10]. MSK Power Spectrum From Equation, the RF power pectrum i obtained by frequency hifting t he magnitude quared of the Fourier tran form of the ba eband pul e- haping func tion. For MSK, the ba eband pul e haping function i given by

Thu , the normalized power pectral den ity for MSK i given by [Pa 79]

Figure how the power pectral den ity of an MSK ignal. The pectral den ity of QPSK and OQPSK are al o drawn for compari on. From Figure, it i ee n that the MSK pectrum ha lower ide lobe than QPSK and OQPSK. Ninety-nine p

ercent of the MSK power i contained within a bandwidth B=1.2/T, while for QPSK and OQPSK, the 99 percent bandwidth B i equal to 8/T. The fa ter roll off the MSK pectrum i due to the fact that moother pul e function are u ed. Fig. al o how that the main lobe of MSK i wider that that of QPSK and OQPSK, and hen ce when compared in term of fir t null bandwidth, MSK i le pectrally effici ent than the pha e- hift keying technique [Pa 79].

Since there i no abrupt change in pha e at bit tran ition period , band limiting the MSK ignal to meet required out-of-band pecification doe not ca u e the envelope to go through zero. The envelope i kept i kept more or le con tant even after band limiting. Any mall variation in the envelope level c an be removed by hard limiting at the receiver without rai ing the out-of-band r adiation level . Since the amplitude i kept con tant, MSK ignal can be ampli fied u ing efficient nonlinear amplifier . The continuou pha e property make it highly de irable for highly reactive load . In addition to the e advantage , MSK ha imple demodulation and ynchronization circuit . It i for the e rea on that MSK i a popular modulation cheme for mobile radio communication .

MSK Tran mitter and Receiver Figure how a typical MSK modulator. Multiplying a carrier ignal with co [ t/2T] produce two pha e-coherent ignal at fc+1/4T and fc-1/4T. The e tw o FSK ignal are eparated u ing two narrow bandpa filter and appropriately combined to form the in-pha e and quadrate carrier component x(t) and y(t), re pectively. The e carrier are multiplied with the odd and even bit tream , ml( t) and mQ(t), to produce the MSK modulated ignal SMSK(t). The block diagram of an MSK receiver i hown in. The received ignal S MSK(t) (in the ab ence of noi e and interference) i multipath by the re pective in-pha e and quadrate carrier x(t) and y(t). The output of the multiplier ar e integrated over two bit period and dumped to a deci ion circuit at the end of each two bit period . Ba ed on the level of the ignal at the output of the in tegrator, the thre hold detector decide whether the ignal i a 0 or a 1. The output data tream corre pond to mI(t) and mQ (t), which are off et combined to obtain the demodulated ignal.

6.39(317)

Figure: Power pectral den ity of MSK ignal al .

a compared to QPSK and OQPSK ign

Figure: Block diagram of an MSK tran mitter. Note than ml(t) and mQ(t) are off e t by Tb.

Figure: Block diagram of an MSK receiver. 2. Explain Gau ian Minimum Shift Keying Modulation Tech. Gau ian Minimum Shift Keying (GMSK) GMSK i a imple binary modulation cheme which may be viewed a a deriv ate of MSK. In GMSK, the ide lobe level of the pectrum are further reduced b y pa ing the modulating NRZ data waveform through a premodulation Gau ian pul e- haping filter [Mur81] ( ee Section 6.63). Ba eband Gau ian pul e haping m oothe the pha e trajectory of the MSK ignal and hence tabilize the in tantan eou frequency variation over time. Thi ha the effect of con iderably reduci ng the ide lobe level in the tran mitted pectrum.

Premodulation Gau ian filtering convert the full re pon e me age ign al (where each ba eband ymbol occupie a ingle bit period T) into a partial re pon e cheme where each tran mitted ymbol pan everal bit period . However, ince pul e haping doe not cau e the pattern-averaged pha e trajectory to dev iate from that of imple MSK, GMSK can be coherently detected ju t a an MSK ig nal, or noncoherently detected a imple FSK. In practice, GMSK i mo t attract ive for it excellent power efficiency (due to the con tant envelope) and it ex cellent pectral efficiency. The premodulationn Gau ian filtering introduce I SI in the tran mitted ignal, but it can be hown that the degradation i not e vere if the 3 dB-bandwidth-bit duration product (BT) of the filter i greater th an 0.5. GMSK acrifice the irreducible error rate cau ed by partial re pon e ignaling in exchange for extremely good pectral efficiency and con tant envelop e propertie . The GMSK premodulation filter ha an impul e re pon e given by

and the GMSK filter may be completely defined from B and the ba eband ymbol dur ation T. It i therefore cu tomary to define GMSK by it BT product. Figure how the imulated RF power pectrum of the GMSK ignal for vari ou value of BT. The power pectrum of MSK, which i equivalent to GMSK with a BT product of infinity, i al o hown for compari on purpo e . It i clearly een from the graph that a the BT product decrea e , the ide lobe level fall o ff very rapidly. For example, for a BT=0.5, the peak of the econd lobe i more

The parameter

related to B, the 3 dB ba eband bandwidth of HG(f), by

than 30 dB below the main lobe, wherea for imple MSK, the econd lobe i only 20 dB below the main lobe. However, reducing BT increa e the irreducible erro r rate produced by the low pa filter due to ISI. A mobile radio channel ind uce an irreducible error rate due to mobile velocity, o a long a the GMSK. T able how occupied bandwidth containing a given percentage of power in a GMSK ignal a a function of the BT product [Mur81]. Table: Occupied RF Bandwidth (for GMSK and MSK a a fraction of Rb) Containing a Given Percentage of Power [Mur81]. Notice that GMSK i Spectrally Tighter than MSK BT 90% 0.2 GMSK 0.25 GMSK 0.5 GMSK MSK 0.78 99% 0.52 0.57 0.69 1.20 99.9% 0.79 0.86 1.04 2.76 99.99% 0.99 1.22 1.09 0.52 1.33 2.08 6.00

While the GMSK pectrum become more and more compact with decrea ing BT value, the degradation due to ISI increa e . It wa hown by I hizuka [I h80] that the BER degradation due to ISI cau ed by filtering i minimum for a BT valu e of 0.5887, where the degradation in the required Eb/N0 i only 0.14 dB from th e ca e of no ISI. GMSK Bit Error Rate The bit error rate for GMSK wa fir t found in [Mur81] for AWGN channel, and wa hown to offer performance within 1 dB of optimum MSK when BT=0.25. Th e bit error probability i a function of BT, ince the pul e haping impact ISI . The bit error probability for GMSK i given by

GMSK Tran mitter and Receiver The imple t way to generate a GMSK ignal i to pa a NRZ me age bit tream through a Gau ian ba eband filter having an impul e re pon e given in Eq uation (7), followed by an FM modulator. Thi modulation technique i hown in figure and i currently u ed in a variety of analog and digital implementation for the US Cellular Digital packet Data (CDPD) y tem a well a for the Global Sy tem for Mobile (GSM) y tem. Figure may al o be implemented digitally u ing a tandard I/Q modulator. GMSK ignal can be detected u ing orthogonal coherent detector a hown in figure, or with imple no coherent detector uch a tandard FM di criminat or . Carrier recovery i ometime performed u ing a method ugge ted by de Bud a [deB72], where the um of the two di crete frequency component contained at t he output of a frequency doubler i divided by four. De Buda method i imilar to the Co ta loop and i equivalent to that of a PLL with a frequency doubler. Thi type of receiver can be ea ily implemented u ing digital logic a hown in Figure. The two D flip-floop act a a quadrature product demodulator and the XOR gate act a ba eband multiplie . The mutually orthogonal reference carrier are generated u ing two D flip-flop , and the VCO center frequency i et equa l to four time the carrier center frequency. A nonptimum, but highly effective method of detecting GNSK ignal i to imply ample the output of an FM demodul ator.

Figure: Block diagram of a GMSK tran mitter u ing direct FM generation.

Figure: Block diagram of a GMSK receiver. 3. Explain about M-ray PSK modulation Tech. M-ray Pha e Shift Keying (MPSK) In M-ray PSK, the carrier pha e take on one of M po ible value , namely, i=2(i1) /M, where i=1,2,.,M. The modulated waveform can be expre ed a

Equation, can be u ed to compute the probability of ymbol error for MPS K y tem in an AWGN channel. From the geometry of Figure, it i ea ily een th at the di tance between adjacent ymbol i equal to . Hence, the average ymb ol error probability of an M-ary PSK y tem i given by

Since there are only two two dimen ional. The M-ary me adiu centered at the origin. l et i illu trated in Figure. t envelope ignal when no pul e

ba i ignal , the con tellation of M-ary PSK i age point are equally paced on a circle of r The con tellation diagram of an 8-ary PSK igna It i clear from Figure. that MPSK i a con tan haping i u ed.

Defined over the interval 0

By choo ing orthogonal ba i

ignal t

1(t)=

TS, the M-ray PSK ignal et can be expre ed a

where ES=(log2M) Eb i the energy per ymbol and T =(log2M) Tb i iod. The above equation can be rewritten in quadrature form a

the

ymbol per

Figure: Con tellation diagram of an M-ary PSK y tem (M-8. Ju t a in BPSK and QPSK modulation, M-ary PSK modulation i either i e ither coherently detected or differentially encoded for noncoherent differential detection. The ymbol error probability of a differential M-ary PSK y tem in AWGN channel for M 4 i approximated by [Hay94] Power Spectra of M-ary PSK The power pectral den ity (PSD) of an M-ary PSK ignal can be obtained in a manner imilar to that de cribed for BPSK and QPSK ignal . The ymbol dur ation TS of an M-ary PSK ignal i related to the bit duration Tb by TS=Tblog2 M (16) The PSD of the M-ary PSK ignal with rectangular pul e i given by

Table: Bandwidth and Power Efficiency of M-ary PSK Signal M 2 4 8 16 32 64 B = Rb/B* 0.5 1 1.5 2 2.5 3 Eb/NO for BER=10-6 10.5 10.5 14 18.5 23.4 28.5 *B: Fir t null bandwidth of M-ary PSK ignal The PSD of M-ary PSK y tem for M=8 and M=16 are hown in Figure. A cl early een from Equation (17) and Figure, the fir t null bandwidth of M-ary PSK ignal decrea e a M increa e while Rb i held con tant. Therefore, a the va lue M increa e , the bandwidth efficiency al o increa e . That i , for fixed Rb , B increa e and B decrea e a M i increa ed. At the ame time, increa ing M implie that the con tellation i more den ely packed, and hence the power effic iency (noi e tolerance) i decrea ed. The bandwidth and power efficiency of M-P SK y tem u ing ideal Nyqui t pul e haping inn AWGN for variou value of M are li ted in Table. The e value a ume no timing jitter or fading, which have a large negative effect on bit error rate a M increa e . In general, imulation mu t be u ed to determine bit error value in actual wirele communication chan nel , ince interference and mulipath can alter the in tantaneou pha e of an MP SK ignal, thereby creating error at the detector. Al o, the particular implem entation of the receiver often impact performance. In practice, pilot ymbol or equalization mu t be u ed to exploit MPSK in mobile channel , and thi ha not been a popular commercial practice.

4. Explain about M-ary QAM Modulation Tech. M-ary Quadrature Amplitude Modulation (QAM) In M-ary PSK modulation, the amplitude of the tran mitted ignal wa con trained to remain con tant, thereby yielding a circular con tellation. By allo wing the amplitude to al o vary with the pha e, a new modulation cheme called q uadrature amplitude modulation (QAM) i obtained. Figure. how the con tellati on diagram of 16-ary QAM. The con tellation con i t of a quare lattice of ig nal point . The general form of an M-ary QAM ignal can be defined a

where Emin i the energy of the ignal with the lowe t amplitude, and ai and bi are a pair of independent integer cho en according to the location of the parti

cular ignal point. Note that M-ary QAM doe not have con tant energy per ymbo l, nor doe it have con tant di tance between po ible ymbol tate . It rea on that particular value of Si(t) will be detected with higher probability than other .

Figure: M-ary PSK power pectral den ity, for M=8, 16 (PSD for rectangular and r ai ed co ine filtered pul e are hown for fixed Rb). If rectangular pul e hape are a umed, the ignal Si(t) may be expande d in term of a pair of ba i function defined a The coordinate of the ith me age point are

Figure: Con tellation diagram of an ary QAM (M=16) ignal et.

It can be hown that the average probability of error in an AWGN channel for M-ary QAM, u ing coherent detection, can be approximated by [Hay94]

In term of the average ignal energy Ea , thi can be expre ed a [Zie92]

Where L= For example, for a 16-QA with ignal con tellation a the L by L matrix i

hown in figure

The power pectrum and bandwidth efficiency of QAM modulation i identic al to M-ary PSK modulation. In term of power efficiency, QAM i uperior to Mary PSK. Table li t the bandwidth and power efficiencie of a QAM ignal for v ariou value of M, a uming optimum rai ed co ine roll off filtering in AWGN. A with M-PSK, the table i optimi tic, and actual bit error probability for wir ele y tem mu t be determined by imulating the variou impairment of the ch annel and the pecific receiver implementation. Pilot tone or equalization mu t be u ed for QAM inn mobile y tem . Table: Bandwidth and power Efficiency of QAM [Zie92] M 4 16 B 1 2 Eb/No for BER=10-6 64 3 10.5 256 4 15 1024 5 18.5 4096 6 24

28

33.5

Flat-fading channel cau e a multiplicative (gain) variation in the tran mitted ignal S(t). Since low flat-fading channel change much lower than th e applied modulation, it can be a umed that the attenuation and pha e hift of the ignal i con tant over at lea t one ymbol interval. Therefore, the receiv ed ignal r(t) may be expre ed a

Depending on whether it i po ible to make an accurate e timate of the pha e (t), coherent or noncoherent matched filter detection may be employed at t he receiver. To evaluate the probability of error of any digital modulation cheme in a low flat-fading channel, one mu t average the probability of error of the pa rticular modulation in AWGN channel over the po ible range of ignal trength due to fading. In other word , the probability of error in AWGN channel i vie wed a a conditional error probability, where the condition i that i fixed. H ence, the probability of error in low flat-fading channel can be obtained by a veraging the error in AWGN channel over the fading probability den ity function . In doing o, the probability of error in a low flat-fading channel can be ev aluated a

where Pe(X) i the probability of error for an arbitrary modulation at a pecif ic value of ignal-to-noi e ratio X,X= 2 Eb/N0, and p(X) i the probability den it y function of X due to the fading channel. Eb and N0 are con tant that repre e nt the average energy per bit and noi e power den ity in a non-fading AWGN chann el and the random variable 2 i u ed to repre ent in tantaneou power value of t he fading channel, with re pect to the non-fading Eb/N0. It i convenient to a ume i one, for a unity gain fading channel. Then, p(X) can imply be viewed a the di tribution of the in tantaneou value of Eb/N0 in a fading channel, an

where (t) i the gain of the channel, (t) i additive Gau ian noi e.

(t) i the pha e hift of the channel, and n

Performance of Digital Modulation in Slow Flat-Fading Channel

5. Di cu

the performance of Digital Modulation in low-Fading Channel .

d Pe(X) can be een to be the conditional probability of bit error for a given value of the random Eb/N0 due to fading. For Rayleigh fading channel , the fading amplitude ha a Rayleigh di tri bution, o the fading power 2 and con equently X have a chi- quare di tribution w ith two degree of freedom. Therefore,

where i the average value of the ignal-to-noi e ratio. For e pond to the average Eb/N0 for fading channel.

note that

By u ing Equation (3) and the probability of error of a particular modul ation cheme in AWGN, the probability of error in a low flat-fading channel can be evaluated. It can be hown that for coherent binary PSK and coherent binary FSK. Equation (2) evaluate to [Ste87]

It can al o be hown that the average error probability of DPSK and orth ogonal noncoherent FSK in a low, flat, Rayleigh fading channel are given by

Figure illu trate how the BER for variou modulation change a a func tion of Eb/N0 in a Rayleigh flat-fading environment. The figure wa produced u ing imulation in tead of analy i , but agree clo ely with Equation (4) to (7) [Rap91b]. For large value of Eb/N0 (i.e., large value of X), the error probabili ty Equation may be implified a For GMSK, the expre ion for BER in the AWGN channel i given in Equatio n(10.a) which when evaluated in Equation (2) yield a Rayleigh fading BER of

A een from equation (7) to (11), for lower error rate all five modula tion technique exhibit an inver e algebraic relation between error rate and me an SNR. Thi i in contra t with the exponential relation hip between error rat e and SNR in an AWGN channel. According to the e re ult , it i een that opera ting at BER of 10-3 to 10-6 require roughly a 30 dB to 60 dB mean SNR. Thi i ignificantly larger than that required when operating over a non fading Gau ian noi e channel (20 dB to 50 dB more link i required). However, it can ea il y be hown that the poor error performance i due to the non-zero probability of very deep fade , when the by u ing efficient technique uch a diver ity or er ror control coding to totally avoid the probability of deep fade . Work by Yao [Ya092] demon trate how the analytical technique of Equatio n (2) may be applied to de ired ignal a well a interfering ignal which und ergo Rayleigh, Ricean, or log-normal fading.

UNIT IV SIGNAL PROCESSING IN WIRELESS SYSTEM PART A 1. Write the Diver ity concept.

corr

2. Define micro copic diver ity. Small- cale fading typically re ult in a prevent deep fade from occurring, mic ro copic diver ity technique can exploit the rapidly changing ignal. 3. Define pace Diver ity. By electing the be t ignal at all time , a receiver can mitigate mall- cale f ading effect thi i called antenna diver ity or pace diver ity. 4. Define Macro copic Diver ity. By electing a ba e tation which i not hadowed when other are, the mobile ca n improve ub tantially the average ignal-to-noi e ratio on the forward link. T hi i called macro copic diver ity. 5. What are cla ified for the pace Diver ity? (i)Selection diver ity (ii)Feedback diver ity (iii)Maximal ratio combining (iv)Equal gain diver ity

Channel code that are u ed to detect error are called error detection code , w hile code that can detect and correct error are called error correction code . 7.What i the principle of diver ity?

8.What i the difference between Micro diver ity and Macro diver ity? Diver ity technique that mitigate the effect of multipath fading are called mic ro diver ity

9.What are the different combining technique ?

In tran mitted diver ity there are multiple antenna and tran mit power

10.What i

tran mitted diver ity?

i) ii) iii) iv)

Selection combining technique Thre hold combining technique Maximal ratio combining technique Equal gain combining technique

Diver ity to mitigate the effect of hadowing from building and object i ed Macro diver ity

One of the be t technique to mitigate combining of independently fading ignal path e fact that independent ignal path have a low eep fade imultaneou ly. Thu the idea behind a over independent fading path

the effect of fading i diver ity . Diver ity combining exploit th probability of experiencing the d diver ity i to ent the ame dat

6. Define error detection code

& error correction code .

The diver ity concept can be explained imply. If one radio path undergoe p fade, another independent path may have a trong ignal.

a dee

call

i divided among the e antenna . Tran mitted diver ity i de irable in y tem w here more pace, power , proce ing capability i available on the tran mitted ide than receiver ide, A be t exemplified by cellular y tem . Tran mitted div er ity de ign depend on whether or not the complex channel gain i known to the tran mitter. When thi gain i known the y tem i quiet imilar to receiver di ver ity 11.What are different type of equalizer ?

Equalizer (MLSE) 12.What i Vocoder ?

Vocoder are a cla of peech coding y tem that analyze the voice ignal at the tran mitter, tran mit parameter derived from the analy i , and then ynt he ize the voice at the receiver u ing tho e parameter . 13. What are the different type of Vocoder ?

14. What i the Channel Vocoder ? The Channel vocoder wa the fir t among the analy i ynthe i y tem o f peech demon trated practically. Channel vocoder are frequency domain vocode r that determine the envelope of the peech ignal for a number of frequency ba nd and then ample, encode, and multiplex the e ample with the encoded output of the other filter . The ampling i done ynchronou ly every 10 m to 30 m . Along with the energy information about each band, the voiced / unvoiced deci ion, and the pitch frequency for voiced peech are al o tran mitted.

1. 2. 3. 4.

Channel Vocoder Formant Vocoder Cep trum Vocoder Voice-Excited Vocoder

i) ii)

Deci ion feedback Equalization(DFE) Maximum likelihood equence e timation

Further non linear equalizer i

cla ified in to two method

i) ii)

Zero-Forcing(ZF) Equalizer Minimum Mean-Square Error(MMSE) Equalizer

1. Linear Equalizer 2. Non linear Equalizer Linear Equalizer i divided into two categorie

uch a

There are two different type of equalizer

uch a

15. What i the Formant Vocoder ? The formant vocoder (Bay 73) i imilar in concept to the channel vocode r. Theoretically, the formant vocoder can operate at lower bit rate than the c hannel vocoder becau e it u e fewer control ignal . In tead of ending ample of the power pectrum envelope, the formant vocoder attempt to tran mit the p o ition of the peak (formant ) of the pectral envelope. Typically, a formant vocoder mu t be able to identify at lea t three formant for repre enting the peech ound , and it mu t al o control the inten itie of the formant . Formant vocoder can reproduce peech at bit rate lower than 1200 bit / . However, dut to difficultie in accurately computing the location of formant and formant tran ition from human peech, they have not been very ucce ful.

16. What i the Cep trum Vocoder ? The Cep trum vocoder eparate the excitation and vocal tract pectrum b y inver e Fourier tran forming of the log magnitude pectrum to produce the cep trum of the ignal. The low frequency coefficient in the cep trum corre pond t o the vocal tract pectral envelope, with the high frequency excitation coeffici ent forming a periodic pul e train at multiple of the ampling period. Linear filtering i performed to eparate the vocal tract cep tral coefficient from th e excitation coefficient . In thi receiver, the vocal tract cep tral coefficie nt are Fourier tran formed to produce the vocal tract impul e re pon e. By con volving thi impul e re pon e with a ynthetic excitation ignal (random noi e o r periodic pul e train), the original peech i recon tructed. 17. What i the Voice Excited Vocoder? Voice excited vocoder eliminate the need for pitch extraction and voici ng detection operation . Thi y tem u e a hybrid combination of PCM tran mi ion for the low frequency band of peech, combined with channel vocoding of high er frequency band . A pitch ignal i generated at the ynthe izer by rectifyin g, bandpa filtering, and clipping the ba eband ignal, thu creating a pectra lly flat ignal with energy at pitch harmonic . Voice excited vocoder have bee n de igned for operation at 7200 bit / to 9600 bit / , and their quality i typ ically uperior to that obtained by the traditional pitch excited vocoder . 18. A digital mobile communication y tem ha a forward channel frequency band r anging between 810 MHz to 826 MHz and a rever e channel band between 940 MHz to 956 MHz. A ume that 90 percent of the bandwidth i u ed by traffic channel. I t i required to upport at lea t 1150 imultaneou call u ing FDMA. The modul ation cheme employed ha a pectral efficiency of 1.68 bp /Hz. A uming that t he channel impairment nece itate the u e of rate 1/2 FEC code , find the upper bound on the tran mi ion bit rate that a peech coder u ed in thi y tem hou ld provide? Solution:

Total Bandwidth available for traffic channel = 0.9 x (810-826) = 14.4MHz. Number of imultaneou u er = 1150 Therefore, maximum channel bandwidth = 14.4/1150 MHz = 12.5 kHz Spectral Efficiency = 1.68 bp / Hz. Therefore, maximum channel data rate = 1.68 x 12500 bp = 21kbp . FEC coder rate = 0.5 Therefore, maximum net data rate = 21 x 0.5kbp = 10.5kbp Therefore, we need to de ign a peech coder with a data rate le than or equal to 10.5kbp .

19. The output of a peech coder ha bit which contribute to ignal quality wit h varying degree of importance. Encoding i done on block of ample of 20m d uration (260 bit of coder output). The fir t 50 of the encoded peech bit ( a y type 1) in each block are con idered to be the mo t ignificant and hence to p rotect them from channel error are appended with 10 CRC bit and convolutionall y encoded with a rate 1/2 FEC coder. The next 132 bit ( ay type 2) are appende d with 5 CRC bit and the la t 78 bit ( ay type 3) are not error protected. Co mpute the gro channel data rate achievable. Solution: Number of type 1 channel bit to be tran mitted every 20 m (50 + 10) x 2 = 120 bit Number of type 2 channel bit to be tran mitted every 20 m 132 + 5 = 137 bit Number of type 3 channel bit to be encoded = 78bit Total number of channel bit to be tran mitted every 20 m 120+137+78bit = 335bit Therefore, gro channel bit rate = 335/(20 x 10-3) = 16.75kbp

20. Explain the principle behind the zero forcing equalizer? (Nov/Dec-2007). In zero forcing equalizer, the equalizer coefficient cn are cho en to f orce the ample of the combined channel and equalizer impul e re pon e to zero at all but one of the NT paced ample point in the tapped delay filter.

PART B

Equalization technique can be ubdivided into two general categorie -li

Survey of Equalization Technique

1. Di cu

the different type of Equalization Tech.

near and nonlinear equalization. The e categorie are determined from how the o utput of an adaptive equalizer i u ed for ub equent control (feedback) of the equalizer. In general, the analog ignal d(t) i proce ed by the deci ion maki ng device in the receiver. The deci ion maker determine the value of the digi tal data bit being received and applie a licing or thre holding operation (a n on-linear operation) in order to determine the value of d(t) ( ee Figure). If d (t) i not u ed in the feedback path to adapt the equalizer, the equalization i linear. On the other hand, if d(t) i fed back to change the ub equent output of the equalizer, the equalization i nonlinear. Many filter tructure are u ed to implement linear and nonlinear equalizer . Further, for each tructure, there are numerou algorithm u ed to adapt the equalizer. Figure provide a ge neral categorization of the equalization technique according to the type , tru cture , and algorithm u ed. The mo t common equalizer tructure i a linear tran ver al equalizer (L TE). A linear tran ver al filter i made up of tapped delay line , with the tap ing paced a ymbol period (TS) apart, a hown in Figure. A uming that the d elay element have unity gain and delay TS, the tran fer function of a linear tr an ver al equalizer can be written a a function of the delay operator exp(-j TS) or z-1. The imple t LTE u e only feed forward tap , and the tran fer function of the equalizer filter i a polynomial in z-1. Thi filter ha many zeeroe b ut pole only at z=0, and i called a finite impul e re pon e (FIR) filter, or imply a tran ver al filter. If the equalizer ha both feed forward and feedback tap , it tran fer function i a rational function of z-1, and i called an inf inite impul e re pon e (IIR) filter with pole and zero . Figure how a tapped delay line filter with both feed forward and feedback tap . Since IIR filter t end to be un table when u ed in channel where the tronge t pul e arrive after an echo pul e (i.e., leading echoe ), they are rarely u ed.

Figure: Cla

ification of equalizer .

Figure: Ba ic linear tran ver al equalizer tructure.

Figure: Tapped delay line filter with both feed forward and feedback tap .

A linear equalizer can be implemented a an Fir filter, otherwi e known a the tran ver al filter. Thi type of equalizer i the imple t type availabl e. In uch an equalizer, the current and pa t value of the received ignal are linearly weighted by the filter coefficient and ummed to produce the output, a hown, Figure. If the delay and the tap gain are analog, the continuou out put of the equalizer i ampled at the ymbol rate and the ample are applied t o the deci ion device. The implementation i , however, u ually carried out in t he digital domain where the ample of the received ignal are tored in a hift regi ter. The output of thi tran ver al filter before a deci ion i made (thr e hold detection) i [Kor85]

where cn* repre ent the complex filter coefficient or tap weight , i the ou tput at time index k, yi i the input received ignal at time t0+iT, t0 i the e qualizer tarting time, and N=N1+N2+1 i the number of tap . The value N1 and N2 denote the number of tap u ed in the forward and rever e portion of equaliz er, re pectively. The minimum mean quared error E [ e (n) 2] that a linear tran v er al equalizer can achieve i [Pro89]

Figure: Structure of linear tran ver al equalizer. Figure: Structure of linear tran ver al equalizer

where F(ej T) i the frequency re pon e of the channel, and N0 i the noi e power pectral den ity.

The linear equalizer can al o be implemented a a lattice filter, who e tructure i hown forward and backward error ignal , fn(k) and bn(k) re pectiv ely, which are u ed a input to the tap multiplier and are u ed to calculate t he updated coefficient . Each tage of the lattice i the characterized by the following recur ive equation [Bin88].

where Kn(k) i the reflection coefficient for the nth tage of the lattice. The backward error ignal , bn, are then u ed a input to the tap weight , and the output of the equalizer i given by

Linear Equalizer

Figure: The tructure of a lattice equalizer [from [Pro91] IEEE]. Two main advantage of the lattice equalizer i it numerical tability a nd fa ter convergence. Al o, the unique tructure of the lattice filter allow the dynamic a ignment of the mo t effective length of the lattice equalizer. H ence, if the channel i not very time di per ive, only a fraction of the tage a re u ed. When the channel become more time di per ive, the length of the equal izer can be increa ed by the algorithm without topping the operation of the equ alizer. The tructure of a lattice equalizer, however, i more complicate than a linear tran ver al equalizer. Nonlinear Equalization Nonlinear equalizer are u ed in application where the channel di torti on i too evere for a linear equalizer to handle, and are commonplace in practi cal wirele y tem . Linear equalizer do not perform well on channel which h ave deep pectral null in the pa band. In an attempt to compen ate for the d i tortion, the linear equalizer place too much gain in the vicinity of the pec tral null, thereby enhancing the noi e pre ent in tho e frequencie . Three very effective nonlinear method have been developed which offer i mprovement over linear equalization technique and are u ed in mo t 2G and 3G y tem . The e are [Pro91]:

Figure: Deci ion feedback equalizer (DFE) Deci ion Feedback Equalization (DFE) The ba ic idea behind deci ion feedback equalization i that once an inf ormation ymbol ha been detected and decided upon, the ISI that it induce on f uture ymbol can be e timated and ubtracted out before detection of ub equent ymbol [Pro89]. The DFE can be realized in either the direct tran ver al form or a a lattice filter. The direct form i hown in Figure. It con i t of a feed forward filter (FFF) and a feedback filter (FBF). The FBF i driven by dec i ion on the output of the detector, and it coefficient can be adju ted to ca ncel the ISI on the current ymbol from pa t detected ymbol . The equalizer ha N1+N2+1 tap in the feed forward filter and N3 tap in the feedback filter, an d it output can be expre ed a :

where and yn are tap gain and the input , re pectively, to the forward filter , the feedback filter, and di(i<k) i the previou deci ion made on the detect

1. 2. 3.

Deci ion Feedback Equalization (DFE) Maximum Likelihood Symbol Detection Maximum Likelihood Sequence E timation (MLSE)

ed ignal. Thi i , once i obtained u ing Equation (7), dk i decided from i t. Then, dk along with previou deci ion dk-1, dk-1. are fed back into the equal izer, and i obtained u ing Equation (7). The minimum mean quared error a DF E can achieve i [Pro89] It can be hown that the minimum MSE for a DFE in Equation (8) i alway maller than that of and LTE in Equation (2) unle i a con tant (i.e., when adaptive equalization i not needed) [Pro89]. If there are null in a DFE h a ignificantly maller minimum MSE than an LTE. Therefore, an LTE i well beh aved when the channel pectrum i comparatively flat, but if the channel i eve rely di torted or exhibit null in the pectrum, the performance of an LTE dete riorate and the mean quared error of a DFE i much better than a LTE. Al o, a n LTE ha difficulty equalizing a non minimum pha e channel, where the tronge t energy arrive after the fir t arriving ignal component. Thu , a DFE i more appropriate for everely di torted wirele channel . The lattice implementation of the DFE i equivalent to a tran ver al DFE having a feed forward filter of length N1 and a feedback filter of length N2, w here N1>N2 Another form of DFE propo ed by Belfiore and park [Bel79] i called a pr edictive DFE, and i hown in Figure. It al o con i t of a feed forward filter (FFF) a in the conventional DFE. However, the feedback filter (FBF) i driven by an input equence formed by the difference of the output of the detector and the output of the feed forward filter. Hence, the FBF here i called a noi e p redictor becau e it predict the noi e and the re idual ISI contained in the ig nal at the FFF output and ubtract from it the detector output after ome feedb ack delay. The predictive DFE perform a well a the conventional DFE a the l imit in the number of tap in the FFF and the FBF approach infinity. The FBF in the predictive DFE can al o be realized a a lattice tructure [Zho90]. The RL S lattice algorithm can be u ed in thi ca e to yield fa t convergence. Maximum Likelihood Sequence E timation (MLSE) Equalizer The MSE-ba ed linear equalizer de cribed previou ly are optimum with re pect to the criterion of minimum probability of ymbol error when the channel d oe not introduce any amplitude di tortion. Yet thi i preci ely the condition in which an equalizer i needed for a mobile communication link. Thi limitat ion on MSE-ba ed equalizer led re earcher to inve tigate optimum or nearly opt imum nonlinear tructure . The e equalizer u e variou form of the cla ical maximum likelihood receiver tructure . The e equalizer u e variou form of t he cla ical maximum likelihood receiver tructure. U ing a channel impul e re pon e imulator within the algorithm the MLSE te t all po ible data equence (rather than decoding each received ymbol by it elf), and choo e the data equ ence will the maximum probability a the output. An MLSE u ually ha a large co mputational requirement, e pecially when the delay pread of the channel i larg e. U ing the MLSE a an equalizer wa fir t propo ed by Forney [For78] in which he et up a ba ic MLSE e timator tructure and implemented it with the Viterbi algorithm. Thi algorithm de cribed wa recognized to be a maximum likelihood equence e timator (MLSE) of the tate equence of a finite tate Markov proce ob erved in memory le noi e. It ha recently been implemented ucce fully f or equalizer in mobile radio channel .

Figure: Predictive deci ion feedback equalizer The MLSE can be viewed a a problem in e timating the tate of a di cret e-time finite tate machine, which in thi ca e happen to be the radio channel with coefficient fk, and with a channel tate which at any in tant of time i e timated by the receiver ba ed on the L mo t recent input ample . Thu , the ch annel ha ML tate , where M i the ize of the ymbol alphabet of the modulatio n. That i , an ML trelli i u ed by the receiver to model the channel over tim e. The Viterbi algorithm then track the ate of the channel by the path throu gh the trelli and given at tage k a rank ordering of the ML mo t probable eq uence terminating in the mo t recent L ymbol .

The block diagram of a MLSE receiver ba ed on the DFE i hown in Figure . The MLSE i optimal in the en e that it minimize the probability of a eque nce error. The MLSE require knowledge of the channel characteri tic in order to compute the metric for making deci ion . The MLSE al o require knowledge o f the tati tical di tribution of the noi e corrupting the ignal. Thu , the pr obability di tribution of the noi e determine the form of the metric for optimu m demodulation of the received ignal. Notice that the matched filter operate on the continuou time ignal, wherea the MLSE and channel e timator rely on di cretized (nonlinear) ample .

2. Explain Adaptive Equalization.

Figure: The tructure of a maximum likelihood adaptive matched filter.

equence e timator (MLSE) with an

Since an adaptive equalizer compen ate for an unknown and time-varying channel, it require a pecific algorithm to update the equalizer coefficient a nd track the channel variation . A wide range of algorithm exi t to adapt the filter coefficient . The development of adaptive algorithm i a complex undert aking, and it i beyond the cope of thi text to delve into great detail on how thi i done. Excellent reference exi t which treat algorithm development [Wi d85], [Hay86], [Pro91]. Thi ection de cribe ome practical i ue regarding equalizer algorithm de ign, and outline three of the ba ic algorithm for adapt ive equalization. Though the algorithm detailed in thi ection are derived fo rt the linear, tran ver al equalizer, they can be extended to other equalizer t ructure , including nonlinear equalizer . nclude.

Rate of convergence: Thi i defined a the number of iteration required for th e algorithm, in re pon e to tationary input , to converge clo e enough to the o ptimum olution. A fa t rate of convergence allow the algorithm to adapt rapid ly to a tationary environment of unknown tati tic . Furthermore, it enable t he algorithm to track tati tical variation when operating in a non tationary e nvironment. Mi adju tment: For an algorithm of intere t, thi parameter provide a quantitat ive mea ure of the amount by which the final value of the mean quare error, ave raged over an en emble of adaptive filter , deviate from the optimal minimum me an quare error.

In practice, the co t of the computing platform, the power budget, and t he radio propagation characteri tic dominate the choice of an equalizer tructu re and it algorithm. In portable radio application , battery drain at the ub criber unit i a paramount con ideration, a cu tomer talk time need to be maxi mized. Equalizer are implemented only if they can provide ufficient link impr ovement to ju tify the co t and power burden. The radio channel characteri tic and intended u e of the ub criber equ ipment i al o key. The peed of the mobile unit determine the channel fading rate and the Doppler pread, which i directly related to the coherence time of the channel. The choice of algorithm, and it corre ponding rate of convergence , depend on the channel data rate and coherence time. The maximum expected time delay pread of the channel dictate the numbe r of tap u ed in the equalizer de ign. An equalizer can only equalize over del ay interval le than or equal to the maximum delay within the filter tructure . For example, if each delay element in an equalizer offer a 10 micro econd de lay, and four delay element are u ed to provide a five tap equalizer, then the maximum delay pread that could be ucce fully equalized i 4x10 =40 . Tran mi i on with multipath delay pread in exce of 40 could not be equalized. Since the circuit complexity and proce ing time increa e with the number of tap and delay element , it i important to know the maximum number of delay element be

Numerical propertie : When an algorithm i implemented numerically, inaccuracie are produced due to round-off noi e and repre entation error in the computer. The e kind of error influence the tability of the algorithm.

Computational complexity: Thi complete iteration of the algorithm.

i the number of operation required to make one

The performance of an algorithm i

determined by variou factor

which i

Algorithm

for Adaptive Equalization.

fore electing an equalizer tructure and it algorithm. The effect of channel fading are di cu ed by Praxi [Pro91], with regard to the de ign of the US Dig ital Cellular equalizer. A tudy which con idered a number of equalizer for a wide range of channel condition wa conducted by Rappaport, et al. [Rap93a]. The cla ic equalizer algorithm are di cu ed below. The e include the zero forcing ZF) algorithm, the lea t mean quare (LMS) algorithm, and the rec ur ive lea t quare (RLS) algorithm. While the e algorithm are primitive for mo t of today wirele tandard , they offer fundamental in ight into algorithm de ign and operation.

In a zero forcing equalizer, the equalizer coefficient cn are cho en to force the ample of the combined channel and equalizer impul e re pon e to zer o at all but one of the NT paced ample point in the tapped delay line filter. By letting the number of coefficient increa e without bound, an infinite leng th equalizer with zero ISI at the output can be obtained. When each of the dela y element provide a time delay equal to the ymbol T, the frequency re pon e H eq (f) of the equalizer i periodic with a period equal to the ymbol rate 1/T. The combined re pon e of the channel with the equalizer mu t ati fy Nyqui t fi r t criterion.

Where Hch(f) i the folded frequency re pon e of the channel. Thu , an infinite length, zeri, ISI equalizer i imply an inver e filter which invert the folde d frequency re pon e of the channel. The infinite length equalizer i u ually i mplemented by a truncated length ver ion. The zero forcing algorithm wa developed by Lucky [Luc65] for wireline communication. The zero forcing equalizer ha the di advantage that the inver e filter may exce ively amplify noi e at frequencie where the folded channel p ectrum ha high attenuation. The ZF equalizer thu neglect the effect of noi e altogether, and i not often u ed for wirele link . However, it perform wel l for tatic channel with high SNR, uch a local wired telephone line . Lea t Mean Square Algorithm Amore robu t equalizer i the LMS equalizer which the criterion u ed i the minimization of the mean quare error (MSE) between the de ired equalizer ou tput and the actual equalizer output. U ing the notation developed, the LMS alg orithm can be readily under tood.

For a pecific channel condition, the prediction error ek i dependent o n the tap gain vector wN, o the MSE of an equalizer i a function of wN. Let t he co t function J(wN) denote the mean quared error a a function of tap gain v ector wN. Following the derivation in order to minimize the MSE, it i required to et the derivative of Equation (13) to zero

The LMS algorithm eek .

to minimize the mean quare error given in Equation (12)

To compute the mean quare error ain

at time in tant k, Equation i

Referring to Figure, the prediction error i

given by

quared to obt

Zero Forcing Algorithm

Simplifying Equation (13)

Equation (14) i a cla ic re ult, and i called the normal equation, i nce the error i minimized and i made orthogonal (normal) to the projection rel ated to the de ired ignal xk. When Equation (14) i ati fied, the MMSE of the equalizer i

To obtain the optimal tap gain vector the normal equation in (14) mu t be olved iteratively a the equalizer converge to an acceptable mall value o f opt. There are everal way to do thi , and many variant of the LMS algorithm have been built upon the olution of Equation (15). One obviou technique i t o calculate

However, inverting a matrix require O(N3) arithmetic operation [Joh82] . Other method uch a Gau ian elimination [Joh82] and Chole ky factorization [Bie77] require O(N2) operation per iteration. The advantage of the e method which directly olve Equation (16) i that only N ymbol input are required to olve the normal equation. Con equently, a long training equence i not nece ary.

In practice, the minimization of the MSE i carried out recur ively, and may be performed by u e of the tocha tic gradient algorithm introduced by Widr ow [Wid66]. Thi i more commonly called the lea t mean quare (LMS) algorithm. The LMS algorithm i the imple t equalization algorithm and require only 2N+ 1 operation per iteration. The filter weight are updated by the update equati on given below [Ale86]. Letting the variable n denote the equence of iteratio n . LMS i computed iteratively by

The LMS equalizer maximize the ignal to di tortion ratio at it output within the con traint of the equalizer filter length. If an input ignal ha a time di per ion characteri tic that i greater than the propagation delay thro ugh the equalizer, then the equalizer will be unable to reduce di tortion. The convergence rate of the LMS algorithm i low due to the fact that there i only one parameter, the tep ize , that control the adaptation rate. To prevent th e adaptation form becoming un table, the value of i cho en from

where i i the ith eigenvalue of the covariance matrix RNN. Since the tep iz e can be controlled by the total input power in order to avoid in tability in th e equalizer [Hay86]. Recur ive Lea t Square Algorithm

The convergence rate of the gradient-ba ed LMS algorithm i very pecially when the eigen value of the input covariance matrix RNN have a rge pread, i.e., max/ min 1. In order to achieve fa ter convergence, orithm which involve additional parameter are u ed. Fa ter converging hm are ba ed on a lea t quare approach, a oppo ed to the tati tical

low, e very la complex alg algorit approac

where the ub cript N dentate the number of delay tage in i the ch control the convergence rate and tability of the algorithm.

tep ize whi

h u ed in the LMS algorithm. That i , rapid convergence relie on error mea ure expre ed in term of a time average of the actual received ignal in tead of a tati tical average. The lead to the family of powerful, albeit complex, ada ptive ignal proce ing technique known a recur ive lea t quare (RLS), which ignificantly improve the convergence of adaptive equalizer

The lea t quare error ba ed on the time average i defined a [Hay86], [Pro91]

where yN(i) i the data input vector at time i, and wN(n) i the new tap gain ve ctor at time n. Therefore, e(i,n) i the error the new tap gain at time n to te t the old data at time i, and J(n) i the cumulative quared error of the new t ap gain on all the old data. The RLS olution require finding the tap gain vector of the equalizer w N(n) uch that the cumulative quared error J(n) i minimized. It u e all the previou data to te t the new tap gain . The parameter i a data weighting fact or that weight recent data more heavily in the computation , o that j(n) tend to forget the old data in a non tationary environment. If the channel i ta tionary, may be et be to one [Pro89]. To obtain the minimum of lea t quare error J(n), the gradient of J(n) i n Equation (19) i et to zero.

U ing Equation (20)-(22), it can be hown that [Pro89]

The matrix RNN(n) in Equation (24) i the determini tic correlation matr ix of input data of the equalizer yN(i), and in Equation (25) i the determini t ic cro -correlation vector between input of the equalizer yN(i) and the de ire d output d(i), where d(i)=x(i). To compute the equalizer weight vector u ing Equation (23), it i required to compute From the definition of RNN(n) in Equation (24), it i po ible to obtain a recur ive equation expre ing RNN(n) in term of RNN(n-1)

Since the three term in Equation (26) are all N by N matrice , a matrix inver e lemma [Bie77] can be u ed to derive a recur ive update for in term o f the previou inver e,

Ba ed on the e recur ive equation, the RLS minimization lead

to the following w

where

i the optimal tap gain vector of the RLS equalizer,

where i the weighting factor clo e to 1, but maller than 1, e* (i,n) i mplex conjugate of e (i,n), and the error e(i,n) i

the co

eight update equation :

2. Recur ively compute the following:

3. Explain different Diver ity Tech.

Diver ity i a powerful communication receiver technique that provide w irele link improvement a relatively low co t. Unlike equalization, diver ity require no training overhead ince a training equence i not required by the t ran mitter. Furthermore, there are a wide range of diver ity implementation , m any which are very practical and provide ignificant link improvement with littl e added co t. Diver ity exploit the random nature of radio propagation by finding ind ependent (or at lea t highly uncorrelated) ignal path for communication. In v irtually all application , diver ity deci ion are made by the receiver, and are unknown to the tran mitter. The diver ity concept can be explained imply. If one radio path underg oe a deep fade, another independent path may have a trong ignal. By having m ore than one path to elect from, both the in tantaneou and average SNR at the receiver may be improved, often by a much a 20 dB to 30 dB. There are two type of fading- mall- cale and large- cale fading. Small - cale fade are characterized by deep and rapid amplitude fluctuation which oc cur a the mobile move over di tance of ju t a ju t a few wavelength . The e fade are cau ed by multiple reflection from the urrounding in the vicinity o f the mobile. For narrowband ignal , mall- cale fading typically re ult in a Rayleigh fading di tribution of ignal trength over mall di tance . In order to prevent deep fade from occurring, micro copic diver ity technique can expl oit the rapidly changing ignal. For example, the mall- cale fading hown in F igure. reveal that if two antenna ere eparated by a fraction of a meter, one may receive a null while the other receive a trong. By electing the be t ig nal at all time , a receiver can mitigate mall- cale fading effect (thi i ca lled antenna diver ity or pace diver ity). Large- cale fading i cau ed by hadowing due to variation in both the terrain profile and the nature of the urrounding . In deeply hadowed conditio

Diver ity Technique

In Equation (33), i the weighting coefficient that can change the performance o f the equalizer. If a channel i tiem-invariant, can be et to one. U ually 0. 8< 1 i u ed. The value of ha no influence on the rate of convergence, but doe determine the tracking ability of the RLS equalizer . The maller the , the bet ter the tracking ability of the equalizer. However, if i too mall, the equali zer will be un table [Lin84]. Thi RLS algorithm de cribed above, called the Ka lman RLS algorithm, u e 2.5N2+4.5N arithmetic operation per iteration.

1. Initialize w(0)=k(0)=0, , where INN i arge po itive con tant.

an NxN identity matrix, and

The RLS algorithm may be

ummarized a follow : i a l

n , the received ignal trength at a mobile can drop well below that of free p ace. Large- cale fading wa hown to be log-normally di tributed with a tandard deviation of about 10 dB in urban environment . By electing a ba e tation wh ich i not hadowed when other are, mobile can improve ub tantially the averag e ignal-to-noi e ratio on the forward link. Thi i called macro copic diver i ty, ince the mobile i taking advantage of large eparation 9the macro y tem difference ) between the erving ba e tation . Macro copic diver ity i al o u eful at the ba e tation receiver. By u ing ba e tation antenna that are ufficiently eparated in pace, the ba e t ation i able to improve the rever e link by electing the antenna with the tro nge t ignal from the mobile.

Derivation of Selection Diver ity Improvement

where we a ume

PM( ) in Equation (38) i the probability of all branche failing to achieve an in tantaneou SNR= . If a ingle branch achieve SNR> , then the probability that SNR > for one or more branche i given by

Equation (5) i an expre ion for the probability of exceeding a thre ho ld when election diver ity i u ed [Jak71], and i plotted in Figure. To determine the average ignal-to-noi e ratio of the received ignal wh en diver ity i u ed, it i fir t nece ary to find the pdf of the fading ignal . For elction diver ity, the average SNR i found by fir t computing the deriv ative of the CDF PM( ) in order to find the pdf of , the in tantaneou SNR when M b

Now, the probability that all M independent diver ity branche receive hich are imultaneou ly le than ome pecific SNR thre hold i

ignal w

where i the mean SNR of each branch. The probability that a ingle branch ha an in tantaneou SNR le than ome thre hold i

i i

If each branch ha

an in tantaneou SNR= I, then from Equation, the pdf of

thwhile er ity. . Each ha the

Before di cu ing the many diver ity technique that are u ed, it i wor to quantitatively determine the advantage that can be achieved u ing div Con ider M independent Rayleigh fading channel available at a receiver channel i called a diver ity branch. Further, a ume that each branch ame average SNR given by

ranche are u ed. Proceeding along the e line ,

From Equation (8), it can be een that the average SNR in the branch whi ch i elected u ing election diver ity naturally increa e , ince it i alway guaranteed to be above the pecified thre hold. Thu , election diver ity offe r an average improvement in the link margin without requiring additional tran m itter power or ophi ticated receiver circuitry. The diver ity improvement can be directly related to the average bit error rate for variou modulation by u i ng the principle . Selection diver ity i ea y to implement becau e all that i needed i ide monitoring tation and an antenna witch at the receiver. However, it i no t an optimal diver ity technique becau e it doe not u e all of the po ible bra nche imultaneou ly. Maximal ratio combining u e each of the M branche in a co-pha ed and weighted manner uch that the highe t achievable SNR i available at the receiver at all time . Derivation of Maximal Ratio Combining Improvement In maximal ratio combining, the voltage ignal ri from each of the M di ver ity branche are co-pha ed to provide coherent voltage addition and are indi vidually weighted to provide optimal SNR. If each branch gain Gi, then re ultin g ignal envelope applied to the detector i

The value hown in Chapter al can be modeled having zero mean

for ii where ri i equal to r(t) a defined in Equation. A 5, the received ignal envelope for a fading mobile radio ign from two independent Gau ian random variable Tc and TS, each and equal variance 2. Thi i ,

Thu , the SNR out of the diver ity combiner in each branch.

imply the um of the SNR

U ing Chebychev inequality [Cou93],

which re ult

in an SNR applied to the detector, M, given by

M i maximized when Gi=ri/N, which lead

A uming that each branch ha the wer NT applied to the detector i ranch. Thu ,

ame average noi e power N, the total noi e po imply the weighted um of the noi e in each b

The following example illu trate

the advantage that diver ity provide .

to

where x= / . Note that i the average SNR for a ingle branch (when no diver ity i u ed). Equation(7) i evaluated to yield the average SNR improvement offered by election diver ity

Then, the mean SNR,

may be expre ed a

Hence M i a Chi- quare di tribution of 2M Gau ian random variable with variance 2/(2N)= /2, where i defined. The re ulting pdf for M can be hown to be

The probability di tribution for maximal ratio combining [Jak71]. It fo llow directly that the average SNR, i imply the um of the individual fro m each branch. In other word ,

The control algorithm for etting the gain and pha e for maximal rati o combining receiver are imilar to tho e required in equalizer and RAKE recei ver . Figure and Figure illu trate maximal ratio combining tructure . Maximal ratio combining, and can be applied to virtually any diver ity application, alt hough often at much greater co t and complexity than other diver ity technique .

Space diver ity, al o known a antenna diver ity, i one of the mo t pop ular form of diver ity u ed in wirele y tem . Conventional wirele y tem con i t of an elevated ba e tation antenna and a mobile antenna clo e to the g round. The exi tence of a direct path between the tran mitter and the receiver i not guaranteed and the po ibility of a number of catterer in the vicinity of the mobile ugge t a Rayleigh fading ignal. From thi model [Jak70], Jake deduced that the ignal received from patially eparated antenna on the mobil e would have e entially uncorrelated envelope for antenna eparation of one h alf wavelength or more.

The concept of antenna pace diver ity i al o u ed in ba e tation de i gn. At each cell ite, multiple ba e tation receiving antenna are u ed to pro vide diver ity reception. However, ince the important catterer are generally on the ground in the vicinity of the mobile, the ba e tation antenna mu t be paced con iderably far apart to achieve decor relation. Separation on the ord er of everal ten of wavelength are required at the ba e tation. Space diver ity can thu be u ed at either the mobile or ba e tation, or both. Figure ho w a general block diagram of a pace diver ity cheme [Cox83a].

1. 2. 3. 4.

Selection diver ity Feedback diver ity Maximal ratio combining Equal gain diver ity

Space diver ity reception method can be cla ified into four categorie [Jak71]:

Practical Space Diver ity Con ideration

The probability that

M i

le

than ome SNR thre hold

Figure: Generalized block diagram for pace diver ity.

Selection Diver ity Selection diver ity i the imple t diver ity technique analyzed. A bloc k diagram of thi method i imilar to that hown in Figure, where m demodulator are u ed to provide m diver ity branche who e gain are adju ted to provide t he ame average SNR for each branch. The receiver branch having the highe t in t antaneou SNR i connected to the demodulator. The antenna ignal them elve c ould be ampled and the be t one ent to a ingle demodulator. In practice, the branch with the large (S+N)/N i u ed, ince it i difficult to mea ure SNR al one. A practical election diver ity y tem cannot function on a truly in tanta neou ba i , but mu t be de igned o that the internal time con tant of the el ection circuitry are horter than the reciprocal of the ignal fading rate. Feedback or Scanning Diver ity Scanning diver ity i very imilar to election diver ity except that in tead of alway u ing the be t of M ignal , the M ignal are canned in a fixe d equence until one i found to be above a predetermined thre hold. Thi igna l i then received until it fall below thre hold and the canning proce i ag ain initiated. The re ulting fading tati tic are omewhat inferior to tho e o btained by the other method , but the advantage with thi method i that it i v ery imple to implement-only one receiver i required. A block diagram of thi method i hown in Figure.

Figure: Ba ic form of

canning diver ity.

Figure:

Maximal ratio combiner.

Maximal Ratio combining In thi method fir t propo ed by Kah [Kah54], the ignal from all of th e M branche are weighted according to their individual ignal voltage to noi e power ratio and then ummed. Figure how a block diagram of the technique. H ere, the individual ignal mu t be co-pha ed before being ummed (unlike elect ion diver ity) require and individual receiver and pha ing circuit for each ant enna element. Maximal ratio combining produce an output SNR equal to the um o f the individual SNR . Thu , it ha the advantage of producing an output with a n acceptable SNR even when none of the individual ignal are them elve accepta ble. Thi technique given the be t tati tical reduction of fading of any known diver ity combiner. Modern DSP technique and digital receiver are now making form of diver ity practical. Equal Gain Combining In certain ca e , it i not convenient to provide for the variable weigh ting capability required for true maximal ratio combining. In uch ca e , the b ranch weight are all et to unit, but the ignal from each branch are co-pha e d to provide equal gain combining diver ity. Thi allow the receiver to exploi t ignal that are imultaneou ly received on each branch. The po ibility of p roducing an acceptable ignal from a number of unacceptable input i till reta ined, and performance i only marginally inferior to maximal ratio combining and uperior to election diver ity. Polarization Diver ity At the ba e tation, pace diver ity i con iderably le practical than at the mobile becau e the narrow angle of incident field require large antenn a pacing [Vau90]. The comparatively high co t of u ing pace diver ity at the ba e tation prompt the con ideration of u ing orthogonal polarization to explo it polarization diver ity. While thi only provide two diver ity branche , it doe allow the antenna element to be co-located. In the early day of cellular radio, all ub criber unit were mounted i n vehicle and u ed vertical whip antenna . Today, however, over half of the u b criber unit are portable. Thi mean that mo t ub criber are no longer u i ng vertical polarization due to hand-tilting when the portable cellular phone i u ed. Thi recent phenomenon ha parked intere t in polarization diver ity at the ba e tation. Mea ured horizontal and vertical polarization path between a mobile and a ba e tation are reported to be uncorrelated by Lee and Yeh [Lee72]. The dec or relation for the ignal in each polarization i cau ed by multiple reflectio n in the channel between the mobile and ba e tation antenna . Chapter 4 howe d that the reflection coefficient for each polarization i different, which re u lt in different amplitude and pha e for each, or at lea t ome, of the reflec tion . After ufficient random reflection , the polarization tate of the igna l will be independent of the tran mitted polarization. In practice, however, th ere i ome dependence of the received polarization on the tran mitted polarizat ion.

Circular and linear polarized antenna have ltipath in ide building [Haw91], [Rap92a], [Ho94]. d, polarization diver ity wa found to dramatically pread without ignificantly decrea ing the received

been u ed to characterize mu When the path wa ob tructe reduce the multipath delay power.

While polarization diver ity ha been tudied in the pa t, it ha primar ily been u ed for fixed radio link which vary lowly in time. Line-of ight mi crowave link , for example, typically u e polarization diver ity to upport two imultaneou u er on the ame radio channel. Since the channel doe not chang e much in uch a link, there i little likelihood of cro polarization interfer ence. A portable u er proliferate, polarization diver ity i likely to become more important for improving link margin and capacity. An outline of a theoret ical model for the ba e tation polarization diver ity reception a ugge ted by Kozono [Koz85] i given below. Theoretical Model for Polarization Diver ity

It i a umed that the ignal i tran mitted from a mobile with vertical (or horizontal) polarization. It i received at the ba e tation by a polariza tion diver ity antenna with two branche . Figure how the theoretical model an d the y tem coordinate . A een in the figure, a polarization diver ity anten na i compo ed of two antenna element V1 and V2, which make a angle (polarizatio n angle) with the Y axi . A mobile tation i located in the direction of off e t angle from the main beam direction of the diver ity antenna a een in Figure( b). Some of the vertically polarized ignal tran mitted are converted to th e horizontal polarized ignal becau e of multipath propagation. The ignal arri ving at the ba e tation can be expre ed a

where x and y are ignal level which are received when =0, It i a umed that r 1 and r2 have independent Rayleigh di tribution , and 1 and 2 have independent uni form di tribution .

Figure: Theoretical model for ba e tation polarization diver ity ba ed on [Koz 85]: (a) x-y plane; (b) x-z plane.

Here. X i the cro polarization di crimination of the propagation path between a mobile and a ba e tation. The correlation coefficient i determined by three factor : polarizatio n angle, off et angle from the main beam direction of the diver ity antenna, and the cro polarization di crimination. The correlation coefficient generally b ecome higher a off et angle become larger. Al o, generally become lower a polarization angle increa e. Thi i becau e the horizontal polarization compon ent become larger a increa e .

Becau e antenna element V1 and V2 are polarized at to the vertical, the received ignal level i lower than that received by a vertically polarized ante nna. The average value of ignal lo L, relative to the received u ing polariz ation i given by L=a2/X+b2 (24)

The re ult of practical experiment carried out u ing polarization dive r ity [Koz85] how that polarization diver ity i a viable diver ity reception t echnique, and i exploited within wirele hand et a well a at ba e tation . Frequency Diver ity Frequency diver ity i implemented by tran mitting information on more t han one carrier frequency. The rationale behind thi technique i that frequenc ie eparated by more than the coherence bandwidth of the channel will be uncorr elated and will thu not experience the ame fade [Lem91]. Theoretically, if t he channel are uncorrelated, the probability of imultaneou fading will be the product of the individual fading probabilitie . Frequency diver ity i often employed in microwave line-of- ight link w hich carry everal channel in a frequency divi ion multiplex mode (FDM). Due t o tropo phere propagation and re ulting refraction, deep fading ometime occur . In practice, 1:N protection witching i provided by a radio licen ee, wherei n one frequency i nominally idle but i available on a tand-by ba i to provid e frequency diver ity witching for any on of the N other carrier (frequencie ) being u ed on the ame link, each carrying independent traffic. When diver ity i needed, the appropriate traffic i imply witched to the backup frequency. Thi technique ha the di advantage that it not only require pare bandwidth b ut al o require that there be a many receiver a there are channel u ed for the frequency diver ity. However, for critical traffic, the expen e may be ju t ified. New OFDM modulation and acce technique exploit frequency diver ity by providing imultaneou modulation ignal with error control coding acro a la rge bandwidth, o that if a particular frequency undergoe a fade, the compo ite

The correlation coefficient

can be written a

The received

ignal value at element

V1 and V2 cab be written a :

ignal will till be demodulated.

Time Diver ity Time diver ity repeatedly tran mit information at time pacing that exc eed the coherence time of the channel, o that multiple repetition of the igna l will be received with independent fading condition , thereby providing for div er ity. One modern implementation of time diver ity involve the u e of the RAK E receiver for pread pectrum CDMA, where the multipath channel provide redund ancy in the tran mitted me age. By demodulating everal replica of the tran m itted CDMA ignal, where each replica experience a particular multipath delay, the RAKE receiver i able to align the replica in time o that a better e timat e of the original ignal may be formed at the receiver. 4. Explain Rake Receiver. RAKE Receiver In CDMA pread pectrum y tem , the chip rate i typically much greater than the flat-fading bandwidth of the channel. Wherea conventional modulation technique require an equalizer to undo the inter ymbol interference between ad jacent ymbol , CDMA preading code are de igned to provide very low correlatio n between ucce ive chip . Thu , propagation delay pread in the radio channel merely provide multiple ver ion of the tran mitted ignal at the receiver. I f the e multipath component are delayed in time by more than a chip duration, t hey appear like uncorrelated noi e at a CDMA receiver, and equalization i not r equired. The pread pectrum proce ing gain make uncorrelated noi e negligibl e after di preading. However, ince there i u eful information in the multipath component , CDMA receiver may combine the time delayed ver ion of the original ignal tran mi ion in order to improve the ignal-to-noi e ratio at the receiver. A RAKE receiver doe ju t thi -it attempt to collect the time- hift ver ion of the or iginal ignal by providing a eparate correlation receiver for each of the multi path ignal . Each correlation receiver may be adju ted in time delay, o that a microproce or controller can cau e different correlation receiver to earch in different time window for ignificant multipath. The range of time delay t hat a particular correlate can earch i called a earch window. The RAKE recei ver, hown in Figure, i e entially a diver ity receiver de igned pecifically for CDMA, where the diver ity i provided by the fact that the multipath compone nt are practically uncorrelated from one another when their relative propagatio n delay exceed a chip period.

Figure: An M-branch (M-finger) RAKE receiver implementation. Each correlator de tect a time hifted ver ion of the original CDMA tran mi ion, and each finger of the RAKE correlate to a portion the ignal which i delayed by at la t one c hip in time from the other finger .

A RAKE receiver utilize multiple correlate to eparately detect the M tronge t multipath component . The output of each correlatro are then weighte d to provide a better e timate of the tran mitted ignal than i provided by a ingle component. Demodulation and bit deci ion are then ba ed on the weighted output of the M correlator . The ba ic idea of a RAKE receiver wa fir t propo ed by Price and Green [Pri58]. In outdoor environment , the delay between multipath component i u u ally large and, if the chip rate i properly elected, the low autocorrelation p ropertie of a CDMA preading equence can a ure that multipath component will appear nearly uncorrelated with each other. However, the RAKE receiver in IS-9 5 CDMA ha been found to perform poorly in indoor environment , which i to be e xpected ince the multipath delay pread in indoor channel ( 100 n ) are much ma ller than an IS-95 chip duration ( 800n ). In uch ca e , a RAKE will not work i nce multipath i unrecoverable, and Rayleigh flat-fading typically occur within a ingle chip period.

To explore the performance of a RAKE receiver, a ume M correlation are u ed in a CDMA receiver to capture the M tronge t multipath component . A wei ghting network i u ed to provide a linear combination of the correlator output for bit detection. Correlator 1 i ynchronized to the tronge t multipath m1. Multipath component m2 arrive 1 later than component m1 where 2- 1 i a umed to b e grater than a chip duration. The econd correlator i ynchronized to m2. It correlate trongly with m2, but ha low correlation with m1. Note that if onl y a ingle correlator i u ed in the receiver, once the output of the ingle cor relator i corrupted by fading, the receiver cannot correct the value. Bit deci ion ba ed on only a ingle correlation may produce a large bit error rate. In a RAKE receiver, if the output from one correlator i corrupted by fading, the other may not be, and the corrupted ignal may be di counted through the weight ing proce . Deci ion ba ed on the combination of the M eparate deci ion tat i tic offered by the RAKE provide a form of diver ity which can overcome fading and thereby improve CDMA reception. The M deci ion tati tic are weighted to form an overall deci ion tati tic a hown in Figure. The output of the M correlate are denoted a Z1, Z2, and ZM. They are weighted by 1, 2, and M, re pectively. The weighting coefficient are b ed on the power or the SNR from each correlator output. If the power or SNR i mall out of a particular correlator, it will be a igned a mall weight ing factor. Ju t a in the ca e of a maximal ratio combining diver ity cheme, the overall ignal Z i given by

The weighting coefficient . m, are normalized to the output ignal power of the correlator in uch a way that the coefficient um to unity,

A in the ca e of adaptive equalizer and diver ity combining, there are many way to generate the weighting coefficient . However, due to multiple acc e interference, RAKE finger with trong multipath amplitude will not nece a rily provide trong output after correlation. Choo ing weighting coefficient b a ed on the actual output of the correlator yield better RAKE performance 5. Explain the Vocoder . Vocoder are a cla of peech coding y tem that analyze the voice ig nal at the tran mitter, tran mit parameter derived from the analy i , and then ynthe ize the voice at the receiver u ing tho e parameter . All vocoder y tem attempt to model the peech generation proce a a dynamic y tem and try to

quantify certain phy ical con traint of the y tem. The e phy ical con traint are u ed to provide a par imoniou de cription of the peech ignal. Vocoder are, in general, much more complex than the waveform coder and achieve very hig h economy in tran mi ion bit rate. However, they are le robu t, and their pe rformance tend to be talker dependent. The mo t popular among the vocoding y tem i the linear predictive coder (LPC). The other vocoding cheme include t he channel vocoder, formant vocoder, cep trum vocoder and voice excited vocoder.

Figure how the traditional peech generation model that i the ba i o f all vocoding y tem [Fla 79]. The ound generating mechani m form the ourc e and i linearly eparated from the intelligence modulating vocal tract filter which form the y tem. The peech ignal i a umed to be of two type : voiced and unvoiced. Voiced ound (m, n, v pronunciation ) are a re ult of qua iperiodic vi bration of the vocal chord and unvoiced ound (f, , h pronunciation ) are fricativ e produced by turbulent air flow through a con triction. The parameter a oci ated with thi model are the voice pitch, the pole frequencie of the modulating filter, and the corre ponding amplitude parameter . The pitch frequency for mo t peaker i below 300 Hz, and extracting thi information from the ignal i very difficult. The pole frequencie corre pond to the re onant frequencie of the vocal tract and are often called the formant of the peech ignal. For adu lt peaker , the formant are centered around 500 Hz, 1500 Hz, 2500 Hz, and 3500 Hz. By meticulou ly adju ting the parameter of the peech generation model, g ood quality peech can be ynthe ized. (LPC).

Linear Predictive coder (LPC ) [Sch85a] belong to the time domain cla of vocoder . Thi cla of vocoder attempt to extract the ignificant featur e of peech from the time waveform. Though LPC coder are computationally inte n ive, they are by far the mo t popular among the cla of low bit rate vocoder . With LPC, it i po ible to tran mit good quality voice at 4.8 kbp and poore r quality voice at even lower rate . The linear predictive coding y tem model the vocal tract a an all pol e linear filter with a tran fer function de cribed by

Where G i a gain of the filter and z-1 repre ent a unit delay operation. The excitation to thi filter i either a pul e at the pitch frequency or random whi

6. Explain in detail about Linear Predictive Coder

te noi e depending on whether the peech egment i voiced or unvoiced. The coe fficient of the all pole filter are obtained in the time domain u ing linear pr ediction technique [Mak75]. The prediction principle u ed are imilar to tho e in ADPCM coder . However, in tead of tran mitting quantized value of the err or ignal repre enting the difference between the predicted and actual waveform, the LPC y tem tran mit only elected characteri tic of the error ignal. Th e parameter include the gain factor, pitch information, and the voiced / unvoic ed deci ion information, which allow approximation of the correct error ignal. At the receiver, the received information about the error ignal i u ed to det ermine the appropriate excitation for the ynthe i filter. That i , the error ignal i the excitation to the decoder. The ynthe i filter i de igned at th e receiver u ing the received predictor coefficient . In practice many LPC code r tran mit the filter coefficient which already repre ent the error ignal and can be directly ynthe ized by the receiver. Figure how a block diagram of a n LPC y tem [Jay86].

Where en i the prediction error (re idual). The predictor coefficient are cal culated to minimize the average energy E in the error ignal that repre ent the difference between the predicted and actual peech amplitude Where a0 = -1 typically, the error i computed for a time window of 10 m , which corre pond to a value of N = 80. To minimize E with re pect to am, it i requ ired to et the partial derivative equal to zero.

The inner ummation can be recognized a the correlation coefficient Crm and hence the above equation can be rewritten a

. After determining the correlation coefficient Crm, Equation (6) can be u ed to determine the predictor coefficient . Equation (6) i often expre ed i n matrix notation and the predictor coefficient calculated u ing matrix inver i on. A number of algorithm have been developed to peed up the calculation of p redictor coefficient . Normally, the predictor coefficient are not coded direc tly, a they would require 8 bit to 10 bit per coefficient for accurate repre entation [Del193]. The accuracy requirement are le ened by tran mitting the r eflection coefficient (a clo ely related parameter), which have a maller dynam ic range. The e reflection coefficient can be adequately repre ented by 6 bit per coefficient. Thu , for a 10th order predictor, the total number of bit a igned to the model parameter per frame i 72, which include 5 bit for a gai n parameter and 6 bit for the pitch period. If the parameter are e timated ev ery 15 m to 30 m , the re ulting bit rate i in the range of 2400 bp to 4800 b p . The coding of the reflection coefficient can be further improved by perform ing a nonlinear tran formation of the coefficient prior to coding. Thi nonlin ear tran formation reduce the en itivity of the reflection coefficient to qua ntization error . Thi i normally done through a log area ratio (LAR) tran for m which perform an inver e hyperbolic tangent mapping of the reflection coeffic ient , Rn (k)

Determination of Predictor ghted um of p pa t ample in the range of 10-15. U ten a a linear um of the

Coefficient - The linear predictive coder u e a wei to e timate the pre ent ample, where p i typically ing thi technique, the current ample n can be writ immediately preceding ample n-k

Variou LPC cheme differ in the way they recreate the error ignal (excitation ) at the receiver. Three alternative are hown in figure [Luc89]. The fir t o ne how the mo t popular mean . It u e two ource at the receiver, one of wh ite noi e and the other with a erie of pul e at the current pitch rate. The election of either of the e excitation method i ba ed on the voiced / unvoice d deci ion made at the tran mitter and communicated to the receiver along with t he other information. Thi technique require that the tran mitter extract pitc h frequency information which i often very difficult. Moreover, the pha e cohe rence between the harmonic component of the excitation pul e tend to produce a buzzy twang in the ynthe ized peech. The e problem are mitigated in the oth er two approache : Multipul e excited LPC and tocha tic or code excited LPC. Multipul e Excited LPC Atal howed [Ata86] that, no matter how well the pul e i po itioned, ex citation by a ingle pul e per pitch produce audible di tortion. Therefore, he ugge ted u ing more than one pul e, typically eight per period, and adju ting the individual pul e po ition and amplitude equentially to minimize a pectra lly weighted mean quare error. Thi technique i called the multipul e excited LPC (MPE LPC) and re ult in better peech quality, not only becau e the predic tion re idual i better approximated by everal pul e per pitch period, but al o becau e the multipul e algorithm doe not require pitch detection. The number of pul e u ed can be reduced, in particular for high pitched voice , by incorp orating a linear filter with a pitch loop in the ynthe izer. Code Excited LPC In thi method, the coder and decoder have a predetermined code book of tocha tic (zero mean white Gau ian) excitation ignal [Sch85b]. For each pe ech ignal, the tran mitter earche through it code book of tocha tic ignal for the one that give the be t perceptual match to the ound when u ed a an e xcitation to the LPC filter. The index of the code book where the be t match wa found i then tran mitted. The receiver u e thi index to the pick the corre ct excitation ignal for it ynthe izer filter. The code excited LPC (CELP) co der are extremely complex and can require more than 500 million multiply and ad d operation per econd. They can provide high quality even when the excitation i coded at only 0.25 bit per ample. The e coder can achieve tran mi ion b it rate a low a 4.8 kbp . Figure illu trate the procedure for electing the optimum excitation i gnal. The procedure i be t illu trated through an example. Con ider the codin g of a hort 5 m block of peech ignal. At a ampling frequency of 8 kHz, eac h block con i t of 40 peech ample . A bit rate of 1/4 bit per ample corre p ond to 10 bit per block. Therefore, there are 210 = 1024 po ible equence o

f length 40 for each block. Each member of the code book provide 40 ample of the excitation ignal with a caling factor that i changed every 5 m block. T he caled ample are pa ed equentially through two recur ive filter , which i ntroduce voice periodicity and adju t the pectral envelope. The regenerated p eech ample at the output of the econd filter are compared with ample of the original peech ignal to form a difference ignal. The difference ignal repr e ent the objective error in the regenerated peech ignal. Thi i further pr oce ed through a linear filter which amplifie the perceptually more important frequencie and attenuate the perceptually le important frequencie . Though computationally inten ive, advance in DSP and VLSI technology ha ve made real time implementation of CELP code po ible. The CDMA digital cellu lar tandard (IS 95) propo ed by QUALCOMM u e a variable rate CELP codec at 1.2 to 14.4 kbp . In 1995, QUALCOMM introduced QCELP13, a 13.4 kbp CELP coder tha t operate on a 14.4 kbp channel.

Re idual Excited LPC

The rationale behind the re idual excited LPC (RELP) i related to that of the DPCM technique in waveform coding [Del93]. In thi cla of LPC coder , after e timating the model parameter (LP coefficient or related parameter ) an d excitation parameter (voiced / unvoiced deci ion, pitch, and gain) from a pe ech frame, the peech i ynthe ized at the tran mitter and ubtracted from the original peech ignal to from a re idual ignal. The re idual ignal i quanti zed, coded, and tran mitted to the receiver along with the LPC model parameter . At the receiver, the re idual error ignal i added to the ignal generated u ing the model parameter to ynthe ize an approximation of the original peech ignal. The quality of the ynthe ized peech i improved due to the addition of the re idual error. Figure how a block diagram of a imple RELP codec.

7. Explain in detail about choo ing (on electing) peech code for mobile commu nication. Choo ing the right peech codec i an important tep in the de ign of a digital mobile communication y tem [Gow93]. Becau e of the limited bandwidth t hat i available, it i required to compre peech to maximize the number of u er on the y tem. A balance mu t be truck between the perceived quality of th e peech re ulting from thi compre ion and the overall y tem co t and capacit y. Other criterion that mu t be con idered include the end-to-end encoding dela y, the algorithmic complexity of the coder, the dc power requirement , compatibi lity with exi ting tandard , and the robu tne of the encoded peech to tran m i ion error .

A een in the mobile radio channel i a ho tile tran mi ion medium be et with problem uch with problem uch a fading, multipath, and interference. It i therefore important that the peech codec be robu t to tran mi ion error . Dep ending on the technique u ed, different peech coder how varying degree of imm unity to tran mi ion error . For example, under ame bit error rate condition , 40 kbp adaptive delta modulation (ADM) ound much better than 56 kbp log PC M [Ste93]. Thi doe not mean that decrea ing bit rate enhance the coder robu tne to tran mi ion error . On the contrary, a peech ignal are repre ente d by fewer and fewer bit , the information content per bit increa e and hence n eed to be more afely pre erved. Low bit rate vocoder type codec , which do par ametric modeling of the vocal tract and auditory mechani m , have ome bit carr ying critical information which if corrupted would lead to unacceptable di torti on. While tran mitting low bit rate encoded peech, it i imperative to determi ne the perceptual importance of each bit and group them according to their en i tivity to error . Depending on their perceptual ignificance, bit in each grou p are provided with different level of error protection through the u e of diff erent forward error correction (FEC) code . The choice of the peech coder will al o depend on the cell ize u ed. When the cell ize i ufficiently mall uch that high pectral efficiency i a chieved through frequency reu e, it may be ufficient to u e a imple high rate peech codec. In the cordle telephone y tem like CT2 and DECT, which u e ve ry mall cell (microcell ), 32 kbp ADPCM coder are u ed to achieve acceptable performance even without channel coding and equalization. Cellular y tem ope rating with much larger cell and poorer channel condition need to u e error co rrection coding, thereby requiring the peech codec to operate at lower bit rat e . In mobile atellite communication , the cell ize are very large and the a vailable bandwidth i very mall. In order to accommodate reali tic number of u er the peech rate mu t be of the order of 3 kbp , requiring the u e of vocode r technique (Ste93). The type of multiple acce technique u ed, being an important factor in determining the pectral efficiency of the y tem, trongly influence the choi ce of peech codec. The US digital TDMA cellular y tem (IS 136) increa ed the c apacity of the exi ting analog y tem (AMPS) threefold by u ing an 8 kbp VSELP peech codec. CDMA y tem , due to their innate interference rejection capabili tie and broader bandwidth availability, allow the u e of a low bit rate peech codec without regard to it robu tne to tran mi ion error . Tran mi ion err or can be corrected with powerful FEC code , the u e of which, in CDMA y tem ,

doe not effect band, width efficiency very ignificantly. The type of modulation employed al o ha con iderable impact on the choi ce of peech codec. For example, u ing bandwidth efficient modulation cheme c an lower the bit rate reduction requirement on the peech codec, and vice ver a . Table how a li ting of the type of peech codec u ed in variou digital mo bile communication y tem

Standard Service Type Speech Coder Type U ed Bit Rate (kbp ) GSM Cellular RPE LTP 9.6, 13 CD 900 Cellular SBC 16 USDC (IS 136) Cellular VSELP 8 IS 95 Cellular CELP 1.2, 2.4, 4.8, 9.6, 13.4, 14.4 IS 95 PCS PCS CELP 13.4, 14.4 PDC Cellular VSELP 4.5, 6.7, 11.2 CT2 Cordle ADPCM 32 DECT Cordle ADPCM 32 PHS Cordle ADPCM 32 DCS 1800 PCS RPE LTP 13 PACS PCS ADPCM 32

UNIT V ADVANCED TRANSCEIVERS SCHEMES PART A 1.What are the principle Spread pectrum technique ?

a. mation

The ignal occupie ignal.

a bandwidth much larger than i needed for the infor

Table: Speech Coder

U ed in Variou Fir t and Second Generation Wirele

Sy tem

b. The pread- pectrum modulation i done u ing a preading code,which i i ndependent of data in the ignal. c. De preading at the receiver i done by correlating the received ignal w ith a ynchronized copy of the preading code.

The cyclic prefix erve to eliminate ISI between the data block , becau e the fir t ample of the channel output affected by thi ISI can be di c arded without any lo relative to the original information equence.The benefit of adding a cyclic prefix come at a co t. 3.Con ider an OFDM y tem with total bandwidth B=1MHz a uming ==0.A single carrie r system would have sym ol time Ts=1/B=1 .The channel ha a maximum delay pread of Tm=5 , o with T =1 and Tm=5 there would be clearly evere ISI.A ume an OFDM y tem with MQAM modulation applied to each ubchannel.To keep the overhead mall ,the OFDM y tem u e N=128 ubcarrier to mitigate ISI.So Tn=NT =128 .The length of the cyclic prefix i et to =8>Tm/T to en ure no ISI between OFDM ymbol .For the e parameter ,find the ubchannel bandwidth,the total tran mi ion time a o ciated with each OFDM ymbol,the overhead of the cyclic prefix,and the data rate of the y tem a uming M=16. Solution : The ubchannel bandwidth Bn=1/Tn=7.812kHz, o Bn<<Bc=1/Tm=200kHz,e n ure negligible ISI. The total tran mi ion time for each OFDM ymbol i T=Tn+T =128+8=136 . The overhead a ociated with the cyclic prefix i 8/128 which i 62.5%. The y tem tran mit log216 =4bit per carrier every T econd , o the data rat e i 218*4/136*10^-6=3.76 Mbp ,which i lightly le than 4B a a re ult of the cyclic prefix overhead. 4.What are the principle of OFDM? The OFDM y tem effectively decompo e the wideband channel into a et of narrowband orthogonal ubchannel with a different QAM ymbol ent over e ach ubchannel.

5.What i the Capacity of Cellular CDMA? The capacity of CDMA y tem i interference limited, while it i bandwi dth limited in FDMA and TDMA. Therefore, any reduction in the interference will cau e a linear increa e in the capacity of CDMA. Put another way, in a CDMA y tem, the link performance for each u er increa e a the number of u er decrea e .

Example of econd generation wirele y tem include the Global Sy tem for Mobile (GSM), the TDMA and CDMA US digital tandard (the Telecommunication Indu try A ociation IS 136 and IS 95 tandard ), Second Generation Cordle elephone (CT2), the Briti h Standard for cordle telephony, the Per onal Acce Communication Sy tem (PACS) local loop tandard, and Digital European Cordle Telephone (DECT), which i the European tandard for cordle and office teleph ony.

6. What are the example

of econd Generation wirele

y tem ?

Precoding,Adaptive loading and coding acro pproache to mitigate the effect of flat fading acro

ubchannel are better a ubcarrier .

2. What i

cyclic prefix?

The mobile unit in the e network perform everal other function not p erformed by fir t generation ub criber unit , uch a received power reporting, adjacent ba e tation canning, data encoding, and encryption.

DECT i an example of a econd generation cordle telephone tandard wh ich allow each cordle phone to communicate with any of a number of ba e tati on , by automatically electing the ba e tation with the greate t ignal level. In DECT, the ba e tation have greater control in term of witching, ignali ng, and controlling handoff . In general, econd generation y tem have been d e igned to reduce the computational and witching burden at the ba e tation or MSC, while providing more flexibility in the channel allocation cheme o that y tem may be deployed rapidly and in a le coordinated manner.

The aim of third generation wirele network i to provide a ingle et of tandard that can meet a wide range of wirele application and provide un iver al acce throughout the world.

The term 3G Per onal Communication Sy tem (PCS) and 3G Per onal Communi cation Network (PCN) are u ed to imply emerging third generation wirele y tem for hand held device . Other name for PCS include Future Public Land Mobile Telecommunication Sy tem (FPLMTS) for worldwide u e which ha more recently bee n called International Mobile Telecommunication (IMT 2000), and Univer al Mobile Telecommunication Sy tem (UMTS) for advanced mobile per onal ervice in Europe . 11. What i meant by Bluetooth? Bluetooth i an open tandard that ha been embraced by over 1,000 manuf acturer of electronic appliance . It provide an ad-hoc approach for enabling variou device to communicate with one another within a nominal 10 meter range. Named after King Harald Bluetooth, the 10th century Viking who united Denmark and Norway, the Bluetooth tandard aim to unify the connectivity chore of appl iance within the per onal work pace of an individual.

12. Explain the GSM. Global Sy tem for Mobile (GSM) Global Sy tem for Mobile (GSM) i a econd generation cellular y tem t andard that wa developed to olve the fragmentation problem of the fir t cellu lar y tem in Europe. GSM wa the world fir t cellular y tem to pecify digit al modulation and network level architecture and ervice , and i the world mo t popular 2G technology. Before GSM, European countrie u ed different cellular tandard throughout the continent, and it wa not po ible for a cu tomer to u e a ingle ub criber unit throughout Europe. GSM wa originally developed to

10. What are the u e

of Third Generation Network ?

9. What i

the aim of Third generation wirele

network ?

8. How DECT i

u ed in Second Generation Wirele

Network ?

7. What are the u e of Second Generation Wirele

Network ?

erve a the pan European cellular ervice and promi ed a wide range of network ervice through the u e of ISDN. GSM ucce ha exceeded the expectation of virtually everyone, and it i now the world mo t popular tandard for new cellul ar radio and per onal communication equipment throughout the world. A of 2001 , there were over 350 million GSM ub criber world wide. The ta k of pecifying a common mobile communication y tem for Europe i n the 900 MHz band wa taken up in the mid 1980 by the GSM (Group pecial mobil e) committee which wa a working group of the CEPT. In 1992, GSM changed it na me to the Global Sy tem for Mobile Communication for marketing rea on [Mou92]. The etting of tandard for GSM i under the aegi of the European Technical Standard In titute (ETSI).

GSM wa fir t introduced into the European market in 1991. By the end o f 1993, everal non European countrie in South America, A ia, and Au tralia had adopted GSM and the technically equivalent off hoot, DCS 1800, which upport Per onal Communication Service (PCS) in the 1.8 GHz to 2.0 GHz radio band rece ntly created by government throughout the world.

GSM ervice follow ISDN guideline and are cla ified a either tele er vice or data ervice . Tele ervice include tandard mobile telephony and mobi le originated or ba e originated traffic. Data ervice include computer to com puter communication and packet witched traffic. U er ervice may be divided i nto three major categorie : Telephone Service , including emergency calling and fac imile. GSM al o upport Videotex and Teletex, though they are not integral part of the GSM tandard. Bearer Service or data ervice , which are limited to layer 1, 2, and 3 of the open y tem interconnection (OSI) reference model. Supported ervice include packet witched protocol and data rate from 300 bp to 9.6 kbp . Data may be tran mitted u ing either a tran parent mode (where GSM provide tandard channel coding for the u er data) or nontran parent mode (where GSM offer pecial codi ng efficiencie ba ed on the particular data interface. Supplementary ISDN ervice , are digital in nature, and include call diver ion, clo ed u er group , and caller identification, and are not available in analog m obile network . Supplementary ervice al o include the hort me aging ervice (SMS) which allow GSM ub criber and ba e tation to tran mit alphanumeric p age of limited length (160 7 bit ASCII character ) while imultaneou ly carryin g normal voice traffic. SMS al o provide cell broadca t, which allow GSM ba e tation to repetitively tran mit ASCII me age with a many a fifteen 93 cha racter tring in concatenated fa hion. SMS may be u ed for afety and advi ory application , uch a the broadca t of highway or weather information to all GS M ub criber within reception range. From the u er point of view, one of the mo t remarkable feature of GSM i the S ub criber Identity Module (SIM), which i a memory device that tore informatio n uch a the ub criber identification number, the network and countrie where

GSM Service and Feature

13. Explain the GSM Service

and Feature .

the ub criber i entitled to ervice, privacy key , and other u er pecific in formation. A ub criber u e the SIM with a four digit per onal ID number to ac tivate ervice from any GSM phone. SIM are available a mart card (credit ca rd ized card that may be in erted into any GSM phone) or plug in module , whic h are le convenient than the SIM card but are nonethele removable and porta ble. Without a SIM in talled, all GSM mobile are identical and nonoperational. It i the SIM that give GSM ub criber unit their identity. Sub criber may plug their SIM into any uitable terminal uch a a hotel phone, public phone, or any portable or mobile phone and are then able to have all incoming GSM call routed to that terminal and have all outgoing call billed to their home phone, no matter where they are in the world.

14. What i the example of a GSM Call? Example of a GSM Call To under tand how the variou traffic and control channel are u ed, con ider the ca e of a mobile call origination in GSM. Fir t, the ub criber unit mu t be ynchronized to a nearby ba e tation a it monitor the BCH. By receiv ing the FCCH, SCH, and BCCH me age , the ub criber would be locked on to the y tem and the appropriate BCH. To originate a call, the u er fir t dial the in tended digit combination and pre e the end button on the GSM phone. The mobile tran mit a bur t of RACH data, u ing the ame ARFCN a the ba e tation to whi ch it i locked. The ba e tation then re pond with an AGCH me age on the CCC H which a ign the mobile unit to a new channel for SDCCH connection. The ub criber unit, which i monitoring TS 0 of the BCH, would receive it ARFCN and TS a ignment from the AGCH and would immediately tune to the new ARFCH and TS. T hi new ARFCH and TS a ignment i phy ically the SDCCH (not the TCH). Once tun ed to the SDCCH, the ub criber unit fir t wait for the SACCH frame to be tran m itted (the wait would la t, at mo t, 26 frame or 120 m ), which inform the mob ile of any required timing advance and tran mitter power command. The ba e tat ion i able to determine the proper timing advance and ignal level from the mob ile earlier RACH tran mi ion and end the proper value over the SACCH for the mobile to proce . Upon receiving and proce ing the timing advance information in the SACCH, the ub criber i now able to tran mit normal bur t me age a r equired for peech traffic. The SDCCH end me age between the mobile unit an d the ba e tation, taking care of authentication and u er validation, while the PSTN connect the dialed party to the MSC, and the MSC witche the peech path to the erving ba e tation. After a few econd , the mobile unit i commanded by the ba e tation via the SDCCH to retune to a new ARFCN and new TS for the T CH a ignment. Once returned to the TCH, peech data i tran ferred on both the forward and rever e link , the call i ucce fully underway, and the SDCCH i vacated. When call are originated from the PSTN, the proce i quite imilar. The ba e tation broadca t a PCH me age during TS 0 within an appropriate fram e on the BCH. The mobile tation, locked on to that ame ARFCN, detect it pag e and replie with an RACH me age acknowledging receipt of the page. The ba e tation then u e the AGCH on the CCCH to a ign the mobile unit to a new phy ic

A Second remarkable feature of GSM i the on the air privacy which i ovided by the y tem. Unlike analog FM cellular phone y tem which can be read ily monitored, it i virtually impo ible to eave drop on a GSM radio tran mi i on. The privacy i made po ible by encrypting the digital bit tream ent by a GSM tran mitter, according to a pecific ecret cryptographic key that i known only to the cellular carrier. Thi key change with time for each u er. Every carrier and GSM equipment manufacturer mu t ign the Memorandum of Under tandin g (MoU) before developing GSM equipment or deploying a GSM y tem. The MoU i a n international agreement which allow the haring of cryptographic algorithm a nd other proprietary information between countrie and carrier .

pr

al channel for connection to the SDCCH and SACCH while the network and the ervi ng ba e tation are connected. Once the ub criber e tabli he timing advance a nd authentication on the SDCCH, the ba e tation i ue a new phy ical channel a ignment over the SDCCH and the TCH a ignment i made.

15. Explain the GSM peech order. Speech Coding - The GSM peech coder i ba ed on the Re idually Excited Linear P redictive Coder (RELP), which i enhanced by including a Long Term Predictor (LT P) [Hel89]. The coder provide 260 bit for each 20 m block of peech, which yield a bit rate of 13 kbp . Thi peech coder wa elected after exten ive u bjective evaluation of variou candidate coder available in the late 1980 . Pr ovi ion for incorporating half rate coder are included in the pecification . The GSM peech coder take advantage of the fact that in a normal conver ation, each per on peak on average for le than 40% of the time. By incorpo rating a voice activity detector (VAD) in the peech coder, GSM y tem operate in a di continuou tran mi ion mode (DTX) which provide a longer ub criber ba ttery life and reduce in tantaneou radio interference ince the GSM tran mitte r i not active during ilent period . A comfort noi e ub y tem (CNS) at the r eceiving end introduce a background acou tic noi e to compen ate for the annoyi ng witched muting which occur due to DTX. 16. What i DECT?

17. What are the Feature and Characteri tic of DECT?

DECT provide a cordle communication framework for high traffic den i ty, hort range telecommunication, and cover a broad range of application and environment . DECT offer excellent quality and ervice for voice and data app lication [Ow91]. The main function of DECT i to provide local mobility to po rtable u er in an in building Private Branch Exchange (PBX). The DECT tandard upport telepoint ervice , a well. DECT i configured around an open tanda rd (OSI) which make it po ible to interconnect wide area fixed or mobile netwo rk , uch a ISDN or GSM, to a portable ub criber population. DECT provide lo w power radio acce between portable part and fixed ba e tation at range of up to a few hundred meter . 18. Define WLL. In a telephone network wirele local loop i a generic term for an acce y tem that u e a wirele link to connect ub criber to their local exchan ge in place of conventional copper cable. The u age of wirele link horten t he con truction period and al o reduce in tallation & operating co t. for WLL?

The main implementation requirement are a) Quality b) Short con truction period

19. What are the implementation requirement

Feature

and Characteri tic

The Digital European Cordle Telephone (DECT) i a univer al cordle t elephone tandard developed by the European Telecommunication Standard In titu te (ETSI) [Och89], [Mul91]. It i the fir t pan European tandard for cordle telephone and wa finalized in July 1992.

Digital European Cordle

Telephone (DECT)

c) Low co t d) Ab ence of interference e) High maffic column 20. What i u e of Bluetooth? Blue tooth u e the 24 aHZ ISM band to form adhoc info connection betwe en computing device . Initially deigned a a method for eliminating the inter c onnect cable receded to connect lap to cell phone and per onal digital a i ta nt, Blue tooth i abro being propo ed a a method for e-commerce & local voice c ommunication . 12. If GSM u e a frame tructure where each frame con i t of eight time lot , and each time lot contain 156.25 bit , and data i tran mitted at 270.833 kbp in the channel, find (a) the time duration of a bit, (b) the time duration of a lot, (c) the time duration of a frame, and (d) how long mu t a u er occupying a ingle time lot wait between two ucce ive tran mi ion .

Solution: (a) The time duration of a bit, (b) The time duration of a lot, (c) The time duration of a frame, Tf = 8 x T lot = 4.615 m . (d) A u er ha to wait 4.615m , the arrival time of a new frame, for it nex t tran mi ion. 13. If a normal GSM time lot con i t of ix trailing bit , 8.25 guard bit , 26 training bit , and two traffic bur t of 58 bit of data, find the frame effici ency. Solution: A time lot ha 6 + 8.25 + 26 + 2(58) = 156.25 bit . A frame ha 8 x 156.25 = 1250 bit /frame The number of overhead bit per frame i given by

PART B

Second Generation wirele y tem employ digital modulation and advance d call proce ing capabilitie . Example of econd generation wirele y tem i nclude the Global Sy tem for Mobile (GSM), the TDMA and CDMA US digital tandard (the Telecommunication Indu try A ociation IS 136 and IS 95 tandard ), Seco nd Generation Cordle Telephone (CT2), the Briti h Standard for cordle teleph ony, the Per onal Acce Communication Sy tem (PACS) local loop tandard, and D igital European Cordle Telephone (DECT), which i the European tandard for co rdle and office telephony. There are many other econd generation y tem , a de cribed. Second Generation wirele network have introduced new network architecture th at have reduced the computational burden of the MSC. A hown in GSM introduced

Second Generation Wirele

Network

1. Explain in detail about Second Generation wirele

bOH = 8(6) + 8(8.25) + 8 (26) = 322 bit Thu , the frame efficiency

network .

the concept of a ba e tation controller (BSC), which i in erted between ever al ba e tation and the MSC. In PACS / WACS, the BSC i called a radio port con trol unit. Thi architectural change ha allowed the data interface between the ba e tation controller and the MSC to be tandardized thereby allowing carrier to u e different manufacturer for MSC and BSC component . Thi trend in tan dardization and interoperability i new to econd generation wirele network . Eventually, wirele network component , uch a the MSC and BSC, will be avail able a off the helf component , much like their wire line telephone counterpar t .

All econd generation y tem u e digital voice coding and digital modul ation. The y tem employ dedicated control channel (common channel ignaling) within the air interface for imultaneou ly exchanging voice and control inform ation between the ub criber, the ba e tation, and the MSC while a call i in p rogre . Second generation y tem al o provide dedicated voice and ignaling t runk between MSC , and between each MSC and the PSTN. In contra t to fir t generation y tem , which were de igned primarily f or voice, econd generation wirele network have been pecifically de igned to provide paging and other data ervice uch a fac imile and high data rate net work acce . The network controlling tructure i more di tributed in econd ge neration wirele y tem , ince mobile tation a ume greater control function . In econd generation wirele network , the handoff proce i mobile control led and i known a mobile a i ted handoff (MAHO). The mobile unit in the e ne twork perform everal other function not performed by fir t generation ub cri ber unit , uch a received power reporting, adjacent ba e tation canning, dat a encoding, and encryption. DECT i an example of a econd generation cordle telephone tandard wh ich allow each cordle phone to communicate with any of a number of ba e tati on , by automatically electing the ba e tation with the greate t ignal level. In DECT, the ba e tation have greater control in term of witching, ignali ng, and controlling handoff . In general, econd generation y tem have been d e igned to reduce the computational and witching burden at the ba e tation or MSC, while providing more flexibility in the channel allocation cheme o that y tem may be deployed rapidly and in a le coordinated manner.

A di cu ed in third generation wirele y tem will evolve from matur e econd generation y tem . The aim of third generation wirele network i to provide a ingle et of tandard that can meet a wide range of wirele applic ation and provide univer al acce throughout the world. In third generation wi rele y tem , the di tinction between cordle telephone and cellular teleph one will di appear, and a univer al per onal communicator (a per onal hand et) will provide acce to a variety of voice, data, and video communication ervice .

Third generation y tem will u e the Broadband Integrated Service Digi tal Network (B ISDN) to provide acce to information network , uch a the Inte rnet and other public and private databa e . Third generation network will car ry many type of information (voice, data, and video), will operate in varied re gion (den e or par ely populated region ), and will erve both tationary u er and vehicular u er traveling at high peed . Packet radio communication wil l likely be u ed to di tribute network control while providing a reliable inform ation tran fer. The term 3G Per onal Communication Sy tem (PCS) and 3G Per onal Communi

Third Generation Wirele

Network

2. Explain in detail about Third Generation Wirele

Network .

cation Network (PCN) are u ed to imply emerging third generation wirele y tem for hand held device . Other name for PCS include Future Public Land Mobile Telecommunication Sy tem (FPLMTS) for worldwide u e which ha more recently bee n called International Mobile Telecommunication (IMT 2000), and Univer al Mobile Telecommunication Sy tem (UMTS) for advanced mobile per onal ervice in Europe .

The following table ummarize the major cellular, cordle and per onal communication tandard throughout the world. In addition, the US mobile radio frequency pectrum allocation i hown in figure. Fir t generation analog cellular radio y tem throughout the world are compared in Table 1. Digital cordle tandard are compared in Table 2. The t hree mo t popular econd generation cellular radio tandard are li ted in Table 3. Table 4 li t three major econd generation cordle / PCS tandard. Table 1: Analog Cellular Sy tem Overview Standard Mobile Tx / Ba e Tx (MHz) Channel Spacing (kHz) f Channel Region AMPS 824-849/869-894 30 832 The America TACS 890-915/935-960 25 1000 Europe ETACS 872-905/917-950 25 1240 United Kingdom NMT 450 453-457.5/463-467.5 25 180 Europe NMT 900 890-915/935-960 12.5 1999 Europe

Number o

3. Explain in detail about third generation wirel ld.

Standard through out the wor

C-450 450-455.74/460-465.74 10 573 RTMS 450-455/460-465 25 200 Italy Radiocom 2000 192.5-199.5/200.5-207.5 215.5-233.5/207.5-215.5 165.2-168.4/169.8-173 414.8-418/424.8428 12.5 560 640 256 256 France NTT 925-940/870-885 915-918.5/860-863.5 922-925/867-870 25/6.25 6.25 6.25 600/2400 560 480 Japan JTACS/NTACS 915-925/860-870 898-901/843-846 918.5-922/863.5-867 25/12.5 25/12.5 12.5 400/800 120/240 280 Japan

Germany, Portugal

CT2 CT2+ DECT Region Europe Canada Europe Duplexing TDD TDD Frequency band (MHz) 964-968 0/ 1930-1990 or 1920-1930 Carrier pacing (kHz) 100 Number of carrier 40 Channel / carrier 1 Channel bit rate kbp 72 Modulation GFSK GFSK Speech coding 32 kb/ 32 kb/ Average hand et TX power (mW) Peak hand et TX power (mW) Frame duration (m ) 2

PHS Japan TDD 944-948

PACS US FDD or TDD 1880-1900

1728 300 10 77 12 4 1152 384 /4 DQPSK 32 kb/ 32 kb/ 5 10 10 250 10 5

300/300 400 or 32 8 or 4 384 /4 QPSK 10 25 80 100 2.5 or 2.0

Table 2: Digital Cordle

Air Interface Parameter Summary

1895-1918

1850-191

Table 3: Second Generation Digital Cellular Standard Summary GSM IS 136 PDC Year of Introduction 1990 1991 1993 Frequencie 890-915 MHz (R) 935-960 MHz (F) 824-849 MHz (R) 869-894 MHz (F) 810-830 & 1429-1453 MHz (R) 940-960 & 1477-1501 MHz (F) Frequencie 890-915 MHz (R) 935-960 MHz (F) 824-849 MHz (R) 869-894 MHz (F) 810-830 & 1429-1453 MHz (R) 940-960 & 1477-1501 MHz (F) Multiple Acce TDMA/FDMA/FDD TDMA/FDMA/FDD TDMA/FDMA/FDD Modulation GMSK (BT=0.3) /4 DQPSK /4 DQPSK Carrier Separation 200 kHz 30 kHz 25 kHz Channel Data Rate 270.833 kbp 48.6 kbp 42 kbp Number of Voice Channel 1000 2500 3000 Spectrum Efficiency 1.35 bp / Hz 1.62 bp / Hz 1.68 bp / Hz Speech Coding RELP-LTP @ 13 kbp VSELP @ 7.95 kbp VSELP @ 6.7 kbp Channel Coding CRC with r=1/2; L=5 Conv. 7bit CRC with r=1/2; L=6 Conv. CRC with Conv. Equalizer Adaptive Adaptive Adaptive Portable Tx. Power Max./Avg. 1W/125mW 600mW/200mW

PACS-UB DCS 1800 PHS Year of Introduction 1992 1993 1993 Frequencie 1920-1930 MHz 1710-1785 MHz(R) 1805-1880 MHz(F) 1895-1907 MHz Multiple Acce TDMA/FDMA/TDD TDMA/FDMA/FDD TDMA/FDMA/FDD Modulation /4 QPSK GMSK /4 QPSK Carrier Separation 300 kHz 200 kHz 300 kHz Data Rate 384 kbp 270.833 kbp 384 kbp Voice ch./RF. Ch. 4 8 4 Speech Coding ADPCM @ 32 kbp RELP-LTP @ 13 kbp ADPCM @ 32 kbp Channel Coding CRC Conv. r=1/2 CRC Receiver Coherent Coherent Coherent Portable Tx. Power Max./Avg. 200mW/20mW 1W/125 mW 80mW/10mW

Table 4: Unlicen ed / Short Range PCS Standard

Overview

The rapid growth of the Internet ha created a concurrent demand for bro adband Internet and computer acce from bu ine e and home throughout the wor ld. Particularly in developing nation where there i inadequate telecommunicat ion backbone infra tructure, there i a tremendou need for inexpen ive, reliab le, rapidly deployable broadband connectivity that can bring individual and ent erpri e into the information age. In fact, a voice over Internet protocol (Vo lP) become prevalent, it i quite conceivable that a ingle broadband Internet c onnection could omeday provide all of the needed telecommunication ervice , i ncluding telephone ervice, televi ion, radio, fax, and Internet, for a home or bu ine cu tomer. Fixed wirele equipment i extremely well uited for rapidly deploying a broadband connection in many in tance , and thi approach i teadily becoming more popular for providing la t mile broadband local loop acce , a well a for emergency or redundant point to point or point to multipoint private network . Unlike mobile cellular telephone y tem de cribed earlier, fixed wirele communication y tem are able to take advantage of the very well defined, ti me invariant nature of the propagation channel between the fixed tran mitter and fixed receiver. Furthermore, modern fixed wirele y tem are u ually a igne d microwave or millimeter radio frequencie in the 28 GHz band and higher, which i greater than ten time the carrier frequency of 3G terre trial cellular tele phone network . At the e higher frequencie , the wavelength are extremely mal l, which in turn allow very high gain directional antenna to be fabricated in mall phy ical form factor . At higher frequencie , too, more bandwidth can be ea ily u ed. High gain antenna have patial filter propertie that can reject multipath ignal that arrive from direction other than the de ired line of ig ht (LOS), and thi in turn upport the tran mi ion of very wide bandwidth ign al (on the order of ten or hundred of megabit per econd) without di tortion. Al o, ince the carrier frequencie of the e fixed wirele acce terminal a re o high, the radio channel behave much like an optical channel if you can e e an antenna, you can ucce fully communicate to it! Conver ely, if you cannot ee the antenna you will be unable to communicate with it. Thu , fixed wirele network at very high microwave frequencie are only viable where there are no ob truction , uch a in a relatively flat uburban or rural etting. Microwave wirele link can be u ed to create a wirele local loop (WL L) uch a the one hown in figure. The local loop can be thought of a the la t mile of the telecommunication network that re ide between the central office (C O) and the individual home and bu ine e in clo e proximity to the CO. In mo t developed countrie , copper or fiber optic cable already ha been in talled to re idence and bu ine e . However, in many developing nation , nation , cable i too expen ive or can take month or year to in tall. Wirele equipment, o n the other hand, can u ually be deployed in ju t a couple of hour . An additio nal benefit of WLL technology i that once the wirele equipment i paid for, t here are no additional co t for tran port between the CO and the cu tomer premi e equipment (CPE), wherea buried cable often mu t be lea ed from a ervice p rovider or utility company on a monthly ba i . It i po ible that WLL y tem could compete with copper wire ba ed Digital Sub criber Loop (DSL) technologie that are rapidly proliferating.

Government throughout the world have realized that WLL could greatly im prove the efficiency of their citizen while timulating competition that could

Figure: Example of the emerging application

and market for broadband

ervice

Wirele

Local Loop (WLL)

4. Explain in detail about Wirele

Local Loop (WLL).

Bluetooth and Per onal Area Network (PAN ) Given the revolutionary tride that wirele technology ha made during the pa t two decade , electronic manufacturer recently realized there i huge con umer appreciation for removing the wire. The ability to replace the cumber o me cord that connect device to one another ( uch a printer cable , headphone cable , wire that run from a per onal computer to a mou e) with an invi ible, l ow power hort range wirele connection would provide convenience and flexibili ty. Furthermore, wirele connectivity would allow the ability to move equipmen t throughout an office, and would allow collaborative communication between indi vidual , their appliance , and their environment. Bluetooth i an open tandard that ha been embraced by over 1,000 manufacturer of electronic appliance . It provide an ad-hoc approach for enabling variou device to communicate with one another within a nominal 10 meter range. Named after King Harald Bluetooth, the 10th century Viking who united Denmark and Norw ay, the Bluetooth tandard aim to unify the connectivity chore of appliance w ithin the per onal work pace of an individual. Bluetooth operate in the 2.4 GHz ISM Band (2400 2483.5 MHz) and u e a frequenc y hopping TDD cheme for each radio channel. Each Bluetooth radio channel ha a 1 MHz bandwidth and hop at a rate of approximately 1600 hop per econd. Tran mi ion are performed in 625 micro econd lot with a ingle packet tran mitte d over a ingle lot. For long data tran mi ion , particular u er may occupy multiple lot u ing the ame tran mi ion frequency, thu lowing the in tantan eou hopping rate to below 1600 hop / econd. The frequency hopping cheme of each Bluetooth u er i determined from a cyclic code of length 227 1, and each u er ha a channel ymbol rate of 1 Mbp u ing GFSK modulation. The tandard ha been de igned to upport operation in very high interference level and relie on a number of forward error control (FEC) coding and automatic repeat reque t ( ARQ) cheme to upport a raw channel bit error rate (BER) of about 10-3.

Different countrie have allocated variou channel for Bluetooth operat ion. In the US and mo t or Europe, the FHSS 2.4 GHz ISM band i available for B luetooth u e. A detailed li t of tate are defined in the Bluetooth tandard t o upport a wide range of application , appliance , and potential u e of the Pe r onal Area Network. Audio, text, data, and even video i contemplated in the B luetooth tandard. Figure provide a depiction of the Bluetooth concept where a gateway to the Internet via IEEE 802.11b i hown a a conceptual po ibility. The IEEE 802.15 tandard committee ha been formed to provide an intern ational forum for developing Bluetooth and other PAN that interconnect pocket P C , per onal digital a i tant (PDA ), cell phone , light project , and other a ppliance [Bra00]. With the rapid proliferation of wearable computer , uch a PDA , cell phone , mart card , and po ition location device , PAN may provide the connection to an entire new era of remote retrieval and monitoring of the wo rld around u .

5. Explain in detail about Bluetooth and per onal Area Network

lead to improved telecommunication

ervice . (PAN ).

6. Explain in detail about GSM. Global Sy tem for Mobile (GSM) Global Sy tem for Mobile (GSM) i a econd generation cellular y tem t andard that wa developed to olve the fragmentation problem of the fir t cellu lar y tem in Europe. GSM wa the world fir t cellular y tem to pecify digit al modulation and network level architecture and ervice , and i the world mo t popular 2G technology. Before GSM, European countrie u ed different cellular tandard throughout the continent, and it wa not po ible for a cu tomer to u e a ingle ub criber unit throughout Europe. GSM wa originally developed to erve a the pan European cellular ervice and promi ed a wide range of network ervice through the u e of ISDN. GSM ucce ha exceeded the expectation of virtually everyone, and it i now the world mo t popular tandard for new cellul ar radio and per onal communication equipment throughout the world. A of 2001 , there were over 350 million GSM ub crier world wide. The ta k of pecifying a common mobile communication y tem for Europe i n the 900 MHz band wa taken up in the mid 1980 by the GSM (Group pecial mobil e) committee which wa a working group of the CEPT. In 1992, GSM changed it na me to the Global Sy tem for Mobile Communication for marketing rea on [Mou92]. The etting of tandard for GSM i under the aegi of the European Technical Standard In titute (ETSI). GSM wa fir t introduced into the European market in 1991. By the end o f 1993, everal non European countrie in South America, A ia, and Au tralia had adopted GSM and the technically equivalent off hoot, DCS 1800, which upport Per onal Communication Service (PCS) in the 1.8 GHz to 2.0 GHz radio band rece ntly created by government throughout the world.

GSM ervice follow ISDN guideline and are cla ified a either tele er vice or data ervice . Tele ervice include tandard mobile telephony and mobi le originated or ba e originated traffic. Data ervice include computer to com puter communication and packet witched traffic. U er ervice may be divided i

GSM Service

and Feature

nto three major categorie : Telephone Service , including emergency calling and fac imile. GSM al o upport Videotex and Teletex, though they are not integral part of the GSM tandard. Bearer Service or data ervice , which are limited to layer 1, 2, and 3 of the open y tem interconnection (OSI) reference model. Supported ervice include packet witched protocol and data rate from 300 bp to 9.6 kbp . Data may be tran mitted u ing either a tran parent mode (where GSM provide tandard channel coding for the u er data) or nontran parent mode (where GSM offer pecial codi ng efficiencie ba ed on the particular data interface. Supplementary ISDN ervice , are digital in nature, and include call diver ion, clo ed u er group , and caller identification, and are not available in analog m obile network . Supplementary ervice al o include the hort me aging ervice (SMS) which allow GSM ub criber and ba e tation to tran mit alphanumeric p age of limited length (160 7 bit ASCII character ) while imultaneou ly carryin g normal voice traffic. SMS al o provide cell broadca t, which allow GSM ba e tation to repetitively tran mit ASCII me age with a many a fifteen 93 cha racter tring in concatenated fa hion. SMS may be u ed for afety and advi ory application , uch a the broadca t of highway or weather information to all GS M ub criber within reception range. From the u er point of view, one of the mo t remarkable feature of GSM i the S ub criber Identity Module (SIM), which i a memory device that tore informatio n uch a the ub criber identification number, the network and countrie where the ub criber i entitled to ervice, privacy key , and other u er pecific in formation. A ub criber u e the SIM with a four digit per onal ID number to ac tivate ervice from any GSM phone. SIM are available a mart card (credit ca rd ized card that may be in erted into any GSM phone) or plug in module , whic h are le convenient than the SIM card but are nonethele removable and porta ble. Without a SIM in talled, all GSM mobile are identical and nonoperational. It i the SIM that give GSM ub criber unit their identity. Sub criber may plug their SIM into any uitable terminal uch a a hotel phone, public phone, or any portable or mobile phone and are then able to have all incoming GSM call routed to that terminal and have all outgoing call billed to their home phone, no matter where they are in the world.

GSM Sy tem Architecture The GSM y tem architecture con i t of three major interconnected ub y tem that interact between them elve and with the u er through certain networ k interface . The ub y tem are the Ba e Station Sub y tem (BSS), Network and S witching Sub y tem (NSS), and the Operation Support Sub y tem (OSS). The Mobile Station (MS) i al o a ub y tem, but i u ually con idered to be part of the B SS for architecture purpo e . Equipment and ervice are de igned within GSM to upport one or more of the e pecific ub y tem .

The BSS, al o known a the radio ub y tem, provide and manage radio t ran mi ion path between the mobile tation and the Mobile Switching Center (M SC). The BSS al o manage the radio interface between the mobile tation and a ll other ub y tem of GSM. Each BSS con i t of many Ba e Station Controller

A Second remarkable feature of GSM i the on the air privacy which i ovided by the y tem. Unlike analog FM cellular phone y tem which can be read ily monitored, it i virtually impo ible to eave drop on a GSM radio tran mi i on. The privacy i made po ible by encrypting the digital bit tream ent by a GSM tran mitter, according to a pecific ecret cryptographic key that i known only to the cellular carrier. Thi key change with time for each u er. Every carrier and GSM equipment manufacturer mu t ign the Memorandum of Under tandin g (MoU) before developing GSM equipment or deploying a GSM y tem. The MoU i a n international agreement which allow the haring of cryptographic algorithm a nd other proprietary information between countrie and carrier .

pr

(BSC ) which connect the MS to the NSS via the MSC . The NSS manage the witch ing function of the y tem and allow the MSC to communicate with other networ k uch a the PSTN and ISDN. The OSS upport the operation and maintenance of GSM and allow y tem engineer to monitor, diagno e, and trouble hoot all a pe ct of the GSM and allow y tem engineer to monitor, diagno e, and trouble hoo t all a pect of the GSM y tem. Thi ub y tem interact with the other GSM u b y tem , and i provided olely for the taff of the GSM operating company whic h provide ervice facilitie for the network. Figure how the block diagram of the GSM y tem architecture. The Mobi le Station (MS ) communicate with the Ba e Station Sub y tem (BSS) over the rad io air interface. The BSS con i t of many BSC which connect to a ingle MSC, and each BSC typically control up to everal hundred Ba e Tran ceiver Station (BTS ). Some of the BTS may be co-located at the BSC, and other may be remote ly di tributed and phy ically connected to the BSC by microwave link or dedicate d lea ed line . Mobile handoff , (called handover , or HO, in the GSM pecifica tion) between two BTS under the control of the ame BSC are handled by the BSC, and not the MSC. Thi greatly reduce the witching burden of the MSC. A hown in figure, the interface which connect a BTS to a BSC i calle d the Abi interface. The Abi interface carrie traffice and maintenance data, and i pecified by GSM to be tandardized for all manufacturer . In practice, however, the Abi for each GSM ba e tation manufacturer ha ubtle difference , thereby forcing ervice provider to u e the ame manufacturer for the BTS and BSC equipment. The BSC are phy ically connected via dedicated / lea ed line or microw ave link to the MSC. The interface between a BSC and a MSC i called the A Inte rface, which i tandardized within GSM. The A interface u e an SS7 protocol c alled the Signaling Correction Control Part (SCCP) which upport communication between the MSC and the BSS, a well a network me age between the individual ub criber and the MSC. The A interface allow a ervice provider to u e ba e tation and witching equipment made by different manufacturer .

The NSS handle the witching of GSM call between external network and the BSC in the radio ub y tem and i al o re pon ible for managing and providing exte rnal acce to everal cu tomer databa e . The MSC i the central unit in the N SS and control the traffic among all of the BSC . In the NSS, there are three different databa e called the Home Location Regi ter (HLR), Vi itor Location Re gi ter (VLR), and the Authentication Center (AUC). The HLR i a databa e which contain ub criber information and location information for each u er who re id e in the ame city a the MSC. Each ub criber in a particular GSM market i a igned a unique International Mobile Sub criber Identity (IMSI), and thi numbe r i u ed to identify each home u er. The VLR i a databa e which temporarily tore the IMSI and cu tomer information for each roaming ub criber who i vi it ing the coverage area of a particular MSC. The VLR i linked between everal ad joining MSC in a particular market or geographic region and contain ub cripti on information of every vi iting u er in the area. Once a roaming mobile i log ged in the VLR, the MSC end the nece ary information to the vi iting ub crib er HLR o that call to the roaming mobile can be appropriately routed over the PSTN by the roaming u er HLR. The authentication Center i a trongly protected databa e which handle the authentication and encryption key for every ingle ub criber in the HLR and VLR. The Authentication Center contain a regi ter ca lled the Equipment Identity Regi ter (EIR) which identifie tolen or fraudulent ly altered phone that tran mit identity data that doe not match with informati on contained in either the HLR or VLR. The OSS upport one or everal Operation Maintenance Center (OMC) whic h are u ed to monitor and maintain the performance of each MS, BS, BSC, and MSC within a GSM y tem. The OSS ha three main function , which are 1) to maintain all telecommunication hardware and network operation with a particular market , 2) manage all charging and billing procedure , 3) manage all mobile equipment in the y tem. Within each GSM y tem, an OMC i dedicated to each of the e ta k and ha provi ion for adju ting all ba e tation parameter and billing proc edure , a well a for providing y tem operator with the ability to determine the performance and integrity of each piece of ub criber equipment in the y te m. GSM Radio Sub y tem GSM originally u ed two 25 MHz cellular band et a ide for all member c ountrie , but now it i u ed globally in many band . The 890-915 MHz band wa f or ub criber to ba e tran mi ion (rever e link), and the 935-960 MHz band wa for ba e to ub criber tran mi ion (forward link). GSM u e FDD and a combin ation of TDMA and FHMA cheme to provide multiple acce e to mobile u er . Th e available forward and rever e frequency band are divided into 200 kHz wide ch

annel called ARFCN (Ab olute Radio Frequency Channel Number ). The ARFCN deno te a forward and rever e channel pair which i eparated in frequency by 45 MHz and each channel i time hared between a many a eight ub criber u ing TDMA . Each of the eight ub criber u e the ame ARFCN and occupie a unique time lot (TS) per frame. Radio tran mi ion on both the forward and rever e li nk are made at a channel data rate of 270.833 kbp (1625.0/6.0 kbp ) u ing binar y BT = 0.3 GMSK modulation. Thu , the ignaling bit duration i 3.692 , and the effective channel tran mi ion rate per u er i 33.854 kbp (270.833 kbp /8 u er ). With GSM overhead (de cribed ub equently), u er data i actually ent at a maximum rate of 24.7 kbp . Each TS ha an equivalent time allocation of 156.25 channel bit , but of thi , 8.25 bit of guard time and ix total tart and top bit are provided to prevent overlap with adjacent time lot . Each TS ha a t ime duration of 576.92 a hown in figure, and a ingle GSM TDMA frame pan 4. 615 m . The total number of available channel within a 25 MHz bandwidth i 125 (a uming no guard band). Since each radio channel con i t of eight time lot , there are thu a total of 1000 traffic channel within GSM. In practical imp lementation , a guard band of 100 kHz i provided at the upper and lower end of the GSM pectrum, and only 124 channel are implemented. Table ummarize the G SM air interface. The combination of TS number and an ARFCN con titute a phy ical channel for both the forward and rever e link. Each phy ical channel in a GSM y tem c an be mapped into different logical channel at different time . That i , each pecific time lot or frame may be dedicated to either handling traffic data (u er data uch a peech, fac imile, or teletext data), ignaling data (required b y the internal working of the GSM y tem), or control channel data (from the MS C, ba e tation, or mobile u er). The GSM pecification define a wide variety of logical channel which can be u ed to link the phy ical layer with the data l ink layer of the GSM network. The e logical channel efficiently tran mit u er data while imultaneou ly providing control of the network on each ARFCN. GSM p rovide explicit a ignment of time lot and frame for pecific logical chann el , a de cribed below.

Table: GSM Air Interface Specification Summary

Parameter Specification Rever e Channel Frequency

890-915 MHz

Forward Channel Frequency 935-960 MHz ARFCN Number 0 to 124 and 975 to 1023 Tx/Rx Frequency Spacing Tx/Rx Time Slot Spacing 45 MHz 3 Time lot Modulation Data Rate 270.833333 kbp Frame Period 4.615 m U er Per Frame (Full Rate) 8 Time Slot Period 576.9 Bit Period 3.692 Modulation 0.3 GMSK ARFCN Channel Spacing 200 kHz Interleaving (max.delay) 40 m Voice Coder Bit Rate 13.4 kbp

There are two type of GSM logical channel , called traffic channel (TC H ) and control channel (CCH ) [Hod90]. Traffic channel carry digitally encod ed u er peech or u er data and have identical function and format on both the forward and rever e link. Control channel carry ignaling and ynchronizing c ommand between the ba e tation and the mobile tation. Certain type of contr ol channel are defined for ju t the forward or rever e link. There are ix dif ferent type of TCH provided for in GSM, and an even larger number of CCH , bot h of which are now de cribed. GSM Traffic Channel (TCH ) GSM traffic channel may be either full rate or half rate and may carry either digitized peech or u er data. When tran mitted a full-rate, u er data i contained within one TS per frame. When tran mitted a half rate, u er data i mapped onto the ame time lot, but i ent in alternate frame . That i , two half rate channel u er would hare the ame time lot, but would alternately tran mit during every other frame.

Full Rate TCH

Full Rate Speech Channel (TCH/FS) - The full rate peech channel carrie u er peech which i digitized at a raw data rate of 13 kbp . With GSM channel coding added to the digitized peech, the full rate peech channel carrie 22.8 kbp . Full Rate Data Channel for 9600 bp (TCH/F9.6) The full rate traffic data chann l carrie raw u er data which i ent at 9600 bp . With additional forward erro r correction coding applied by the GSM tandard, the 9600 bp data i ent at 22 .8 kbp .

The following full rate peech and data channel are

upported:

In the GSM tandard, TCH data may not be ent in TS 0 within a TDMA fram e on certain ARFCN which erve a the broadca t tation for each cell ( ince th i time lot i re erved for control channel bur t in mo t every frame, a de c ribed ub equently). Furthermore, frame of TCH data are broken up every thirte enth frame by either low a ociated control channel data (SACCH) or idle frame . Figure illu trate how the TCH data i tran mitted in con ecutive frame . Ea ch group of twenty ix con ecutive TDMA frame i called a multiframe (or peech multiframe, to di tingui h it from the control channel multiframe de cribed bel ow). For every twenty ix frame , the thirteenth and twenty ixth frame con i t of Slow A ociated Control Channel (SACCH) data, or the idle frame, re pective ly. The twenty ixth frame contain idle bit for the ca e when full rate TCH are u ed, and contain SACCH data when half rate TCH are u ed.

GSM Channel Type

Half Rate TCH

Half Rate Speech Channel (TCH/HS) The half rate peech channel ha been de igne to carry digitized peech which i ampled at a rate half that of the full rate channel. GSM anticipate the availability of peech coder which can digitize peech at about 6.5 kbp . With GSM channel coding added to the digitized peech , the half rate peech channel will carry 11.4 kbp . Half Rate Data Channel for 4800 bp (TCH/H4.8) The half rate traffic data chann l carrie raw u er data which i ent at 4800 bp . With additional forward erro r correction coding applied by the GSM tandard, the 4800 bp data i ent at 11 .4 kbp . Half Rate Data Channel for 2400 bp (TCH/H2.4) The half rate traffic data chann l carrie raw u er data which i ent at 2400 bp . With additional forward erro r correction coding applied by the GSM tandard, the 2400 bp data i ent at 11 .4 kbp .

There are three main control channel in the GSM y tem. The e are the broadca t channel (BCH), the common control channel (CCCH), and the dedicated co ntrol channel (DCCH). Each control Channel con i t of everal logical channel which are di tributed in time to provide the nece ary GSM control function . The BCH and CCCH forward control channel in GSM are implemented only on certain ARFCN channel and are allocated time lot in a very pecific manner. Specifically, the BCH and CCCH forward control channel are allocated only TS 0 and are broadca t only during certain frame within a repetitive fifty one frame equence (called the control channel multiframe) on tho e ARFCN which are de i gnated a broadca t channel . TS 1 through TS 7 carry regular TCH traffic, o t hat ARFCN which are de ignated a control channel are till able to carry full rate u er on even of the eight time lot . The GSM pecification define thirty four ARFCN a tandard broadca t c hannel . For each broadca t channel, frame 51 doe not contain any BCH / CCCH f orward channel data and i con idered to be an idle frame. However, the rever e channel CCCH i able to receive ub criber tran mi ion during TS 0 of any fra me (even the idle frame). On the other hand, DCCH data may be ent during any t ime lot and any frame, and entire frame are pecifically dedicated to certain D CCH tran mi ion . GSM control channel are now de cribed in detail. Broadca t Channel (BCH ) The broadca t channel operate on the forward link of a pecific ARFCN within each cell, and tran mit data only in the fir t time lo t (TS 0) of certain GSM frame . Unlike TCH which are duplex, BCH only u e the forward link. Ju t a the forward control channel (FCC) in AMPS i u ed a a b eacon for all nearby mobile to camp on to, the BCH erve a a TDMS beacon chan nel for any nearby mobile to identify and lock on to. The BCH provide ynchron ization for all mobile within the cell and i occa ionally monitored by mobile in neighboring cell o that received power and MAHO deci ion may be made by o ut-of-cell u er . Although BCH data i tran mitted in TS 0, the other even tim

GSM Control Channel

(CCH)

The following half rate peech and data channel are

upported:

l r p l r p

Full Rate Data Channel for 4800 bp (TCH/F4.8) The full rate traffic data chann carrie raw u er data which i ent at 4800 bp . With additional forward erro correction coding applied by the GSM tandard, the 4800 bp i ent at 22.8 kb . Full Rate Data Channel for 2400 bp (TCH/F2.4) The full rate traffic data chann carrie raw u er data which i ent at 2400 bp . With additional forward erro correction coding applied by the GSM tandard, the 2400 bp i ent at 22.8 kb .

e lot n a GSM frame for that ame ARFCN are available for TCH data, DCCH data, or are filled with dummy bur t . Furthermore, all eight time lot on all other ARFCN within the cell are available for TCH or DCCH data. The BCH i defined by three eparate channel which are give acce to TS 0 duri ng variou frame of the 51 frame equence. Figure illu trate how the BCH i a llocated frame . The three type of BCH are now de cribed.

a) Broadca t Control Channel (BCCH): The BCCH i a forward control channel that i u ed to broadca t information uch a cell and network identify, and op erating characteri tic of the cell (current control channel tructure, channel availability, and conge tion). The BCCH al o broadca t a li t of channel that are currently in u e within the cell. Frame 2 through frame 5 in a control mul tiframe (4 out of every 51 frame ) contain BCCH data. It hould be noted from f igure that TS 0 contain BCCH data during pecific frame , and contain other BC H channel (FCCH and SCH), common control channel (CCCH ), or an idle frame ( e nt every 51 t frame) during other pecific frame . b) Frequency Correction Channel (FCCH): The FCCH i a pecial data bur t wh ich occupie TS 0 for the very fir t GSM frame (frame 0) and i repeated every t en frame within a control channel multiframe. The FCCH allow each ub criber unit to ynchronize it internal frequency tandard (local o cillator) to the ex act frequency of the ba e tation. c) Synchronization Channel (SCH): SCH i broadca t in TS 0 of the frame imm ediately following the FCCH frame and i u ed to identify the erving ba e tati on while allowing each mobile to frame ynchronize with the ba e tation. The f rame number (FN), which range from 0 to 2,715,647, i ent with the ba e tatio n identity code (BSIC) during the SCH bur t. The BSIC i uniquely a igned to e ach BST in a GSM y tem. Since a mobile may be a far a 30 km away from a erv ing ba e tation, it i often nece ary to adju t the timing of a particular mob ile u er uch that the received ignal at the ba e tation i ynchronized with the ba e tation clock. The BS i ue coar e timing advancement command to the mobile tation over the SCH, a well. The SCH i tran mitted once every ten f

d) Common Control Channel (CCCH ): On the broadca t (BCH) ARFCN, the commo n control channel occupy TS 0 of every GSM frame that i not otherwi e u ed by the BCH or the idle frame. CCCH con i t of three different channel : the pagin g channel (PCH), which i a forward link channel, the random acce channel (RAC H) which i a rever e link channel, and the acce grant channel (AGCH), which i a forward link channel. A een in figure, CCCH are the mo t commonly u ed c ontrol channel a are u ed to page pecific ub criber , a ign ignaling chann el to pecific u er , and receive mobile reque t for ervice. The e channel are de cribed below: a) Paging Channel (PCH): The PCH provide paging ignal from the ba e tat ion to all mobile in the cell, and notifie a pecific mobile of an incoming ca ll which originate from the PSTN. The PCH tran mit the IMSI of the target ub criber, along with a reque t for acknowledgement from the mobile unit on the RA CH. Alternatively, the PCH may be u ed to provide cell broadca t ASCII text me age to all ub criber , a part of the SMS feature of GSM.

b) Random Acce Channel (RACH): The RACH i a rever e link channel u ed by a ub criber unit to acknowledge a page from the PCH, and i al o u ed by mobil e to originate a call. The RACH u e a lotted ALOHA acce cheme. All mobil e mu t reque t acce or re pond to a PCH alert within TS 0 of a GSM frame. At the BTS, every frame (even the idle frame) will accept RACH tran mi ion from mobile during TS 0. In e tabli hing ervice, the GSM ba e tation mu t re pond to the RACH tran mi ion by allocating a channel and a igning a tand alone de dicated control channel (SDCCH) for ignaling during a call. Thi connection i confirmed by the ba e tation over the AGCH. c) Acce Grant Channel (AGCH): The AGCH i u ed by the ba e tation to pro vide forward link communication to the mobile, and carrie data which in truct the mobile to operate in a particular phy ical channel (time lot and ARFCN) wit h a particular dedicated control channel. The AGCH i the final CCCH me age e nt by the ba e tation before a ub criber i moved off the control channel. Th e AGCH i u ed by the ba e tation to re pond to a RACH ent by a mobile tation in a previou CCCH frame. d) Dedicated Control Channel (DCCH ): There are three type of dedicated c ontrol channel in GSM, and, like traffic channel ( ee figure), they are bidire ctional and have the ame format and function on both the forward and rever e li nk . Like TCH , DCCH may exit in any time lot and on any ARFCN except TS 0 of the BCH ARFCN. The tand along dedicated control channel (SDCCH ) are u ed fo r providing ignaling ervice required by the u er . The Slow and Fa t A ocia ted Control Channel (SACCH and FACCH ) are u ed for upervi ory data tran mi ion between the mobile tation and the ba e tation during a call. a) Stand alone Dedicated Control Channel (SDCCH ): The SDCCH carrie igna ling data following the connection of the mobile with the ba e tation, and ju t before a TCH a ignment i i ued by the ba e tation. The SDCCH en ure that the mobile tation and the ba e tation remain connected while the ba e tation and MSC verify the ub criber unit and allocate re ource for the mobile. The S DCCH can be thought of a an intermediate and temporary channel which accept a newly completed call from the BCH and hold the traffic while waiting for the ba e tation to allocate a TCH channel. The SDCCH i u ed to end authentication and alert me age (but not peech) a the mobile ynchronize it elf with the f rame tructure and wait for a TCH. SDCCH may be a igned their own phy ical c hannel or may occupy TS 0 of the BCH if there i low demand for BCH or CCCH traf fic.

b) Slow A ociated Control Channel (SACCH): The SACCH i alway a ociated with a traffic channel or a SDCCH and map onto the ame phy ical channel. Thu

rame within the control channel multiframe, a

hown in figure.

, each ARFCH y tematically carrie SACCH data for all of it current u er . A in the USDC tandard, the SACCH carrie general information between the MS and BTS. On the forward link, the SACCH i u ed to end low but regularly changing control information to the mobile, uch a tran mit power level in truction an d pecific timing advance in truction for each u er on the ARFCN. The rever e SACCH carrie information about the received ignal trength and quality of the TCH, a well a BCH mea urement re ult from neighboring cell . The SACCH i tr an mitted during the thirteenth frame (and the twenty ixth frame when half rate traffic i u ed) of every peech / dedicated control channel multiframe, and wi thin thi frame, the eight time lot are dedicated to providing SACCH data to ea ch of the eight full rate (or ixteen half rate) u er on the ARFCN. c) Fa t A ociated Control Channel (FACCH ): FACCH carrie urgent me age , and contain e entially the ame type of information a the SDCCH. A FACCH i a igned whenever a SDCCH ha not been dedicated for a particular u er and the re i an urgent me age ( uch a a handoff reque t). The FACCH gain acce to a time lot by tealing frame from the traffic channel to which it i a igned. Thi i done by etting two pecial bit , called tealing bit , in a TCH forward channel bur t. If the tealing bit are et, the time lot i known to contain FACCH data, not a TCH, for that frame. Example of a GSM Call To under tand how the variou traffic and control channel are u ed, con ider the ca e of a mobile call origination in GSM. Fir t, the ub criber unit mu t be ynchronized to a nearby ba e tation a it monitor the BCH. By receiv ing the FCCH, SCH, and BCCH me age , the ub criber would be locked on to the y tem and the appropriate BCH. To originate a call, the u er fir t dial the in tended digit combination and pre e the end button on the GSM phone. The mobile tran mit a bur t of RACH data, u ing the ame ARFCN a the ba e tation to whi ch it i locked. The ba e tation then re pond with an AGCH me age on the CCC H which a ign the mobile unit to a new channel for SDCCH connection. The ub criber unit, which i monitoring TS 0 of the BCH, would receive it ARFCN and TS a ignment from the AGCH and would immediately tune to the new ARFCH and TS. T hi new ARFCH and TS a ignment i phy ically the SDCCH (not the TCH). Once tun ed to the SDCCH, the ub criber unit fir t wait for the SACCH frame to be tran m itted (the wait would la t, at mo t, 26 frame or 120 m ), which inform the mob ile of any required timing advance and tran mitter power command. The ba e tat ion i able to determine the proper timing advance and ignal level from the mob ile earlier RACH tran mi ion and end the proper value over the SACCH for the mobile to proce . Upon receiving and proce ing the timing advance information in the SACCH, the ub criber i now able to tran mit normal bur t me age a r equired for peech traffic. The SDCCH end me age between the mobile unit an d the ba e tation, taking care of authentication and u er validation, while the PSTN connect the dialed party to the MSC, and the MSC witche the peech path to the erving ba e tation. After a few econd , the mobile unit i commanded by the ba e tation via the SDCCH to retune to a new ARFCN and new TS for the T CH a ignment. Once returned to the TCH, peech data i tran ferred on both the forward and rever e link , the call i ucce fully underway, and the SDCCH i vacated. When call are originated from the PSTN, the proce i quite imilar. The ba e tation broadca t a PCH me age during TS 0 within an appropriate fram e on the BCH. The mobile tation, locked on to that ame ARFCN, detect it pag e and replie with an RACH me age acknowledging receipt of the page. The ba e tation then u e the AGCH on the CCCH to a ign the mobile unit to a new phy ic al channel for connection to the SDCCH and SACCH while the network and the ervi ng ba e tation are connected. Once the ub criber e tabli he timing advance a nd authentication on the SDCCH, the ba e tation i ue a new phy ical channel a ignment over the SDCCH and the TCH a ignment i made.

Frame Structure for GSM Each u er tran mit a bur t of data during the time lot a igned to it. The e data bur t may have one of five pecific format , a defined in GSM [Ho d90]. Figure illu trate the five type of data bur t u ed for variou control and traffic bur t . Normal bur t are u ed for TCH and DCCH tran mi ion on b oth the forward and rever e link. FCCH and SCH bur t are u ed in TS 0 of peci fic frame ( hown Figure (a)) to broadca t the frequency and time ynchronizatio n control me age on the forward link. The RACH bur t i u ed by all mobile t o acce ervice from any ba e tation, and the dummy bur t i u ed a filter in formation for unu ed time lot on the forward link. Figure illu trate the data tructure within a normal bur t. It con i t of 148 bit which are tran mitted a a rate of 270.833333 kbp (an unu ed guar d time of 8.25 bit i provided at the end of each bur t). Out of the total 148 bit per TS, 114 are information bearing bit which are tran mitted a two 57 b it equence clo e to the beginning and end of the bur t. The midamble con i t of a 26 bit training equence which allow the adaptive equalizer in the mobile or ba e tation receiver to analyze the radio channel characteri tic before de coding the u er data. On either ide of the midamble, there are control bit ca lled tealing flag . The e two flag are u ed to di tingui h whether the TS con tain voice (TCH) or control (FACCH) data, both which hare the ame phy ical ch annel. During a frame, a GSM ub criber unit u e one TS to tran mit, one TS to receive, and may u e the ix pare time lot to mea ure ignal trength on fiv e adjacent ba e tation a well a it own ba e tation.

Normal

RACH bur t

Dummy bur t

3 tart bit 58 mixed bit 26 training bit it 8.25 bit guard period Figure: Time lot data bur t in GSM

8 tart bit 3 top bit

41 bit of ynchronization 36 bit 68.25 bit extended guard period

of encrypted data

58 mixed bit

top b

3 tart bit 58 bit of encrypted data 26 training bit of encrypted data 3 top bit 8.25 bit guard period

58 bit

A hown in figure, there are eight time lot per TDMA frame, and the fr ame period i 4.615 m . A frame contain 8 x 156.25 = 1250 bit , although ome bit period are not u ed. The frame rate i 270.833 kbp / 1250 bit / frame, or 216.66 frame per econd. The 13th or 26th frame are not u ed for traffic, but for control purpo e . Each of the normal peech frame are grouped into larger tructure called multiframe which in turn are grouped into uperframe and hy per frame (hyper frame are not hown in figure). One multiframe contain 26 T DMA frame , and one uperframe contain 51 multiframe , or 1326 TDMA frame . A hyperframe contain 2048 uperframe , or 2,715,648 TDMA frame . A complete hype rframe i ent about every 3 hour , 28 minute , and 54 econd , and i important to GSM ince the encryption algorithm rely on the particular frame number, and ufficient ecurity can only be obtained by u ing a large number of frame a p rovided by the hyperframe.

Signal Proce ing in GSM

Speech Coding - The GSM peech coder i ba ed on the Re idually Excited Linear P redictive Coder (RELP), which i enhanced by including a Long Term Predictor (LT P) [Hel89]. The coder provide 260 bit for each 20 m block of peech, which yield a bit rate of 13 kbp . Thi peech coder wa elected after exten ive u bjective evaluation of variou candidate coder available in the late 1980 . Pr ovi ion for incorporating half rate coder are included in the pecification . The GSM peech coder take advantage of the fact that in a normal conver ation, each per on peak on average for le than 40% of the time. By incorpo rating a voice activity detector (VAD) in the peech coder, GSM y tem operate in a di continuou tran mi ion mode (DTX) which provide a longer ub criber ba ttery life and reduce in tantaneou radio interference ince the GSM tran mitte r i not active during ilent period . A comfort noi e ub y tem (CNS) at the r eceiving end introduce a background acou tic noi e to compen ate for the annoyi ng witched muting which occur due to DTX. TCH/FS, SACCH, and FACCH Channel Coding The output bit of the peech coder are ordered into group for error protection, ba ed upon their ignificance in contr ibuting to peech quality. Out of the total 260 bit in a frame, the mo t impor tant 50 bit , called type Ia bit , have 3 parity check (CRC) bit added to them. Thi facilitate the detection of non correctable error at the receiver. The next 132 bit along with the fir t 53 (50 type Ia bit + 3 parity bit ) are rec orded and appended by four trailing zero bit , thu providing a data block of 18 9 bit . Thi block i then encoded for error protection u ing a rate 1/2 convol utional encoder with con traint length K = 5, thu providing a equence of 378 b it . The lea t important 78 bit do not have any error protection and are conca tenated to the exi ting equence to form a block of 456 bit in a 20 m frame. The error protection coding cheme increa e the gro data rate of the GSM pee ch ignal, with channel coding, to 22.8 kbp . Thi error protection cheme a d

Figure illu trate all of the GSM operation

from tran mitter to receiver.

Figure how that the control multiframe pan 51 frame oppo ed to 26 frame 9120 m ) u ed by the traffic / dedicated ultiframe . Thi i done intentionally to en ure that any GSM er in the erving or adjacent cell) will be certain to receive tran mi ion from the BCH, no matter what particular frame or e u ing.

(235.365 m ), a control channel m ub criber (wheth the SCH and FCCH time lot they ar

e cribed i illu trated in figure.

Channel Coding for Data Channel The coding provided for GSM full rate data chan nel (TCH/F9.6) i ba ed on handling 60 bit of u er data at 5 m interval , in accordance with the modified CCITT V.110 modem tandard. A de cribed by Steele [Ste94], 240 bit of u er data are applied with four tailing bit to a half rat e punctured convolutional coder with con traint length K = 5. The re ulting 488 coded bit are reduced to 456 encoded data bit through puncturing (32 bit are not tran mitted), and the data i eparated into four 114 bit data bur t that are applied in an interleaved fa hion to con ecutive time lot . Channel Coding for Control Channel GSM Control Channel me age are def ined to be 184 bit long, and are encoded u ing a hortened binary cyclic fire c ode, followed by a half rate convolutional coder. The fire code u e the generator polynomial

Which produce 184 me age bit , followed by 40 parity bit four tail bit are a dded to clear the convolutional coder which follow , yielding a 228 bit data blo ck. Thi block i applied to a half rate K = 5 convolutional code (CC(2,1,5)) u ing the generator polynomial G0(x) = 1 + x3 + x4 and G1(x) = 1+ x + x3 + x4 (w hich are the ame polynomial u ed to code TCH type Ia data bit ). The re ultin g 456 encoded bit are interleaved onto eight con ecutive frame in the ame man ner a TCH peech data. Interleaving In order to minimize the effect of udden fade on the rece ived data, the total of 456 encoded bit within each 20 m peech frame or contr ol me age frame are broken into eight 57 bit ub block . The e eight ub block which make up a ingle peech frame are pread over eight con ecutive TCH time lot . (i.e., eight con ecutive frame for a pecific TS). If a bur t i lo t due to interference or fading, channel coding en ure that enough bit will ti ll be received correctly to allow the error correction to work. Each TCH time lot carrie two 57 bit block of data from two different 20 m (456 bit) peech (or control) egment . Figure illu trate exactly how the peech frame are dia gonally interleaved within the time lot . Note that TS 0 contain 57 bit of d ata from the 0th ub block of the nth peech coder frame (denoted a a in the figu re) and 57 bit of data from the 4th ub block of the (n-1) t peech coder fram e (denoted a b in the figure) Ciphering Ciphering modifie the content of the eight interleaved block through the u e of encryption technique known only to the particular mobile tation and ba e tran ceiver tation. Security i further enhanced by the fact t hat the encryption algorithm i changed from call to call. Two type of cipheri ng algorithm , called A3 and A5, are u ed in GSM to prevent unauthorized network acce and privacy for the radio tran mi ion re pectively. The A3 algorithm i u ed to authenticate each mobile by verifying the u er pa code within the SI M with the cryptographic key at the MSC. The A5 algorithm provide the crambli ng for the 114 coded data bit ent in each TS.

Frame Number 0a 5a 4b 1b 1a 6a 5b 2b 2a 7a 6b 3b 3a 7b 4a 0b

Bur t Formatting Bur t formatting add binary data to the ciphered block , in or der to help ynchronization and equalization of the received ignal. Modulation The modulation cheme u ed by GSM i 0.3 GMSK, where 0.3 de c ribe the 3 dB bandwidth of the Gau ian pul e haping filter with relation to t he bit rate (e.g., BT = 0.3). A de cribed in GMSK i a pecial type of digital FM modulation. Binary one and zero are repre ented in GSM by hifting the RF carrier by 67.708 kHz. The channel data rate of GSM i 270.833333 kbp , which i exactly four time the RF frequency hift. Thi minimize the bandwidth occu pied by the modulation pectrum and hence improve channel capacity. The MSK mo dulated ignal i pa ed through a Gau ian filter to mooth the rapid frequency tran ition which would otherwi e pread energy into adjacent channel .

Frequency Hopping Under normal condition , each data bur t belonging to a particular phy ical channel i tran mitted u ing the ame carrier frequency. However, if u er in a particular cell have evere multipath problem , the cell may be defined a a hopping cell by the network operator, in which ca e low fre quency hopping may be implemented to combat the multipath or interference effect in that cell. Frequency hopping i carried out on a frame by frame ba i , thu hopping occur at a maximum rate of 217.6 hop per econd. A many a 64 diff erent channel may be u ed before a hopping equence i repeated. Frequency hop ping i completely pecified by the ervice provider. Equalization Equalization i performed at the receiver with the help of the training equence tran mitted in the midamble of every time lot. The type of equalizer for GSM i not pecified and i left up to the manufacturer. Demodulation The portion of the tran mitted forward channel ignal which i of intere t to a particular u er i determined by the a igned TS and ARFCN. The appropriate TS in demodulated with the aid of ynchronization data provide d by the bur t formatting. After demodulation, the binary information i deciph ered, de-interleaved, channel decoded, and peech decoded.

7. Explain the GSM y tem Architecture. GSM Sy tem Architecture The GSM y tem architecture con i t of three major interconnected ub y tem that interact between them elve and with the u er through certain networ k interface . The ub y tem are the Ba e Station Sub y tem (BSS), Network and S witching Sub y tem (NSS), and the Operation Support Sub y tem (OSS). The Mobile Station (MS) i al o a ub y tem, but i u ually con idered to be part of the B SS for architecture purpo e . Equipment and ervice are de igned within GSM to upport one or more of the e pecific ub y tem .

The BSS, al o known a the radio ub y tem, provide and manage radio t ran mi ion path between the mobile tation and the Mobile Switching Center (M SC). The BSS al o manage the radio interface between the mobile tation and a ll other ub y tem of GSM. Each BSS con i t of many Ba e Station Controller (BSC ) which connect the MS to the NSS via the MSC . The NSS manage the witch ing function of the y tem and allow the MSC to communicate with other networ k uch a the PSTN and ISDN. The OSS upport the operation and maintenance of GSM and allow y tem engineer to monitor, diagno e, and trouble hoot all a pe ct of the GSM and allow y tem engineer to monitor, diagno e, and trouble hoo t all a pect of the GSM y tem. Thi ub y tem interact with the other GSM u b y tem , and i provided olely for the taff of the GSM operating company whic h provide ervice facilitie for the network. Figure how the block diagram of the GSM y tem architecture. The Mobi le Station (MS ) communicate with the Ba e Station Sub y tem (BSS) over the rad io air interface. The BSS con i t of many BSC which connect to a ingle MSC, and each BSC typically control up to everal hundred Ba e Tran ceiver Station (BTS ). Some of the BTS may be co-located at the BSC, and other may be remote ly di tributed and phy ically connected to the BSC by microwave link or dedicate d lea ed line . Mobile handoff , (called handover , or HO, in the GSM pecifica tion) between two BTS under the control of the ame BSC are handled by the BSC, and not the MSC. Thi greatly reduce the witching burden of the MSC.

A hown in figure, the interface which connect a BTS to a BSC i calle d the Abi interface. The Abi interface carrie traffice and maintenance data, and i pecified by GSM to be tandardized for all manufacturer . In practice, however, the Abi for each GSM ba e tation manufacturer ha ubtle difference , thereby forcing ervice provider to u e the ame manufacturer for the BTS and BSC equipment. The BSC are phy ically connected via dedicated / lea ed line or microw ave link to the MSC. The interface between a BSC and a MSC i called the A Inte rface, which i tandardized within GSM. The A interface u e an SS7 protocol c alled the Signaling Correction Control Part (SCCP) which upport communication between the MSC and the BSS, a well a network me age between the individual ub criber and the MSC. The A interface allow a ervice provider to u e ba e tation and witching equipment made by different manufacturer .

The NSS handle the witching of GSM call between external network and the BSC in the radio ub y tem and i al o re pon ible for managing and providing exte rnal acce to everal cu tomer databa e . The MSC i the central unit in the N SS and control the traffic among all of the BSC . In the NSS, there are three different databa e called the Home Location Regi ter (HLR), Vi itor Location Re gi ter (VLR), and the Authentication Center (AUC). The HLR i a databa e which contain ub criber information and location information for each u er who re id e in the ame city a the MSC. Each ub criber in a particular GSM market i a igned a unique International Mobile Sub criber Identity (IMSI), and thi numbe r i u ed to identify each home u er. The VLR i a databa e which temporarily tore the IMSI and cu tomer information for each roaming ub criber who i vi it ing the coverage area of a particular MSC. The VLR i linked between everal ad joining MSC in a particular market or geographic region and contain ub cripti on information of every vi iting u er in the area. Once a roaming mobile i log ged in the VLR, the MSC end the nece ary information to the vi iting ub crib er HLR o that call to the roaming mobile can be appropriately routed over the PSTN by the roaming u er HLR. The authentication Center i a trongly protected databa e which handle the authentication and encryption key for every ingle ub criber in the HLR and VLR. The Authentication Center contain a regi ter ca lled the Equipment Identity Regi ter (EIR) which identifie tolen or fraudulent ly altered phone that tran mit identity data that doe not match with informati on contained in either the HLR or VLR. The OSS upport one or everal Operation Maintenance Center (OMC) whic h are u ed to monitor and maintain the performance of each MS, BS, BSC, and MSC within a GSM y tem. The OSS ha three main function , which are 1) to maintain all telecommunication hardware and network operation with a particular market , 2) manage all charging and billing procedure , 3) manage all mobile equipment in the y tem. Within each GSM y tem, an OMC i dedicated to each of the e ta k and ha provi ion for adju ting all ba e tation parameter and billing proc edure , a well a for providing y tem operator with the ability to determine the performance and integrity of each piece of ub criber equipment in the y te m. 8. Explain the GSM Radio Sub y tem. GSM Radio Sub y tem GSM originally u ed two 25 MHz cellular band et a ide for all member c ountrie , but now it i u ed globally in many band . The 890-915 MHz band wa f or ub criber to ba e tran mi ion (rever e link), and the 935-960 MHz band wa for ba e to ub criber tran mi ion (forward link). GSM u e FDD and a combin ation of TDMA and FHMA cheme to provide multiple acce e to mobile u er . Th e available forward and rever e frequency band are divided into 200 kHz wide ch annel called ARFCN (Ab olute Radio Frequency Channel Number ). The ARFCN deno te a forward and rever e channel pair which i eparated in frequency by 45 MHz and each channel i time hared between a many a eight ub criber u ing TDMA .

Each of the eight ub criber u e the ame ARFCN and occupie a unique time lot (TS) per frame. Radio tran mi ion on both the forward and rever e li nk are made at a channel data rate of 270.833 kbp (1625.0/6.0 kbp ) u ing binar y BT = 0.3 GMSK modulation. Thu , the ignaling bit duration i 3.692 , and the effective channel tran mi ion rate per u er i 33.854 kbp (270.833 kbp /8 u er ). With GSM overhead (de cribed ub equently), u er data i actually ent at a maximum rate of 24.7 kbp . Each TS ha an equivalent time allocation of 156.25 channel bit , but of thi , 8.25 bit of guard time and ix total tart and top

bit are provided to prevent overlap with adjacent time lot . Each TS ha a t ime duration of 576.92 a hown in figure, and a ingle GSM TDMA frame pan 4. 615 m . The total number of available channel within a 25 MHz bandwidth i 125 (a uming no guard band). Since each radio channel con i t of eight time lot , there are thu a total of 1000 traffic channel within GSM. In practical imp lementation , a guard band of 100 kHz i provided at the upper and lower end of the GSM pectrum, and only 124 channel are implemented. Table ummarize the G SM air interface. The combination of TS number and an ARFCN con titute a phy ical channel for both the forward and rever e link. Each phy ical channel in a GSM y tem c an be mapped into different logical channel at different time . That i , each pecific time lot or frame may be dedicated to either handling traffic data (u er data uch a peech, fac imile, or teletext data), ignaling data (required b y the internal working of the GSM y tem), or control channel data (from the MS C, ba e tation, or mobile u er). The GSM pecification define a wide variety of logical channel which can be u ed to link the phy ical layer with the data l ink layer of the GSM network. The e logical channel efficiently tran mit u er data while imultaneou ly providing control of the network on each ARFCN. GSM p rovide explicit a ignment of time lot and frame for pecific logical chann el , a de cribed below.

Rate 270.833333 kbp 4.615 m (Full Rate) 8

Parameter Rever e Channel Forward Channel ARFCN Number Tx/Rx Frequency Tx/Rx Time Slot 3 Time lot Modulation Data Frame Period U er Per Frame

Specification Frequency 890-915 MHz Frequency 935-960 MHz 0 to 124 and 975 to 1023 Spacing Spacing 45 MHz

: GSM Air Interface Specification

Summary

9. Explain the GSM Channel Type .

There are two type of GSM logical channel , called traffic channel (TC H ) and control channel (CCH ) [Hod90]. Traffic channel carry digitally encod ed u er peech or u er data and have identical function and format on both the forward and rever e link. Control channel carry ignaling and ynchronizing c ommand between the ba e tation and the mobile tation. Certain type of contr ol channel are defined for ju t the forward or rever e link. There are ix dif ferent type of TCH provided for in GSM, and an even larger number of CCH , bot h of which are now de cribed. GSM Traffic Channel (TCH ) GSM traffic channel may be either full rate or half rate and may carry either digitized peech or u er data. When tran mitted a full-rate, u er data i contained within one TS per frame. When tran mitted a half rate, u er data i mapped onto the ame time lot, but i ent in alternate frame . That i , two half rate channel u er would hare the ame time lot, but would alternately tran mit during every other frame.

Full Rate TCH

chann

chann

chann

Full Rate Speech Channel (TCH/FS) - The full rate peech channel carrie u peech which i digitized at a raw data rate of 13 kbp . With GSM channel coding added to the digitized peech, the full rate peech channel carrie 22.8 kbp . Full Rate Data Channel for 9600 bp (TCH/F9.6) The full rate traffic data l carrie raw u er data which i ent at 9600 bp . With additional forward erro r correction coding applied by the GSM tandard, the 9600 bp data i ent at 22 .8 kbp . Full Rate Data Channel for 4800 bp (TCH/F4.8) The full rate traffic data l carrie raw u er data which i ent at 4800 bp . With additional forward erro r correction coding applied by the GSM tandard, the 4800 bp i ent at 22.8 kb p . Full Rate Data Channel for 2400 bp (TCH/F2.4) The full rate traffic data l carrie raw u er data which i ent at 2400 bp . With additional forward erro r correction coding applied by the GSM tandard, the 2400 bp i ent at 22.8 kb

The following full rate peech and data channel are

upported: er

In the GSM tandard, TCH data may not be ent in TS 0 within a TDMA fram e on certain ARFCN which erve a the broadca t tation for each cell ( ince th i time lot i re erved for control channel bur t in mo t every frame, a de c ribed ub equently). Furthermore, frame of TCH data are broken up every thirte enth frame by either low a ociated control channel data (SACCH) or idle frame . Figure illu trate how the TCH data i tran mitted in con ecutive frame . Ea ch group of twenty ix con ecutive TDMA frame i called a multiframe (or peech multiframe, to di tingui h it from the control channel multiframe de cribed bel ow). For every twenty ix frame , the thirteenth and twenty ixth frame con i t of Slow A ociated Control Channel (SACCH) data, or the idle frame, re pective ly. The twenty ixth frame contain idle bit for the ca e when full rate TCH are u ed, and contain SACCH data when half rate TCH are u ed.

GSM Channel Type

Time Slot Period 576.9 Bit Period 3.692 Modulation 0.3 GMSK ARFCN Channel Spacing 200 kHz Interleaving (max.delay) 40 m Voice Coder Bit Rate 13.4 kbp

p . Half Rate TCH

Half Rate Speech Channel (TCH/HS) The half rate peech channel ha been de igne to carry digitized peech which i ampled at a rate half that of the full rate channel. GSM anticipate the availability of peech coder which can digitize peech at about 6.5 kbp . With GSM channel coding added to the digitized peech , the half rate peech channel will carry 11.4 kbp . Half Rate Data Channel for 4800 bp (TCH/H4.8) The half rate traffic data chann l carrie raw u er data which i ent at 4800 bp . With additional forward erro r correction coding applied by the GSM tandard, the 4800 bp data i ent at 11 .4 kbp . Half Rate Data Channel for 2400 bp (TCH/H2.4) The half rate traffic data chann l carrie raw u er data which i ent at 2400 bp . With additional forward erro r correction coding applied by the GSM tandard, the 2400 bp data i ent at 11 .4 kbp .

There are three main control channel in the GSM y tem. The e are the broadca t channel (BCH), the common control channel (CCCH), and the dedicated co ntrol channel (DCCH). Each control Channel con i t of everal logical channel which are di tributed in time to provide the nece ary GSM control function . The BCH and CCCH forward control channel in GSM are implemented only on certain ARFCN channel and are allocated time lot in a very pecific manner. Specifically, the BCH and CCCH forward control channel are allocated only TS 0 and are broadca t only during certain frame within a repetitive fifty one frame equence (called the control channel multiframe) on tho e ARFCN which are de i gnated a broadca t channel . TS 1 through TS 7 carry regular TCH traffic, o t hat ARFCN which are de ignated a control channel are till able to carry full rate u er on even of the eight time lot . The GSM pecification define thirty four ARFCN a tandard broadca t c hannel . For each broadca t channel, frame 51 doe not contain any BCH / CCCH f orward channel data and i con idered to be an idle frame. However, the rever e channel CCCH i able to receive ub criber tran mi ion during TS 0 of any fra me (even the idle frame). On the other hand, DCCH data may be ent during any t ime lot and any frame, and entire frame are pecifically dedicated to certain D CCH tran mi ion . GSM control channel are now de cribed in detail. Broadca t Channel (BCH ) The broadca t channel operate on the forward link of a pecific ARFCN within each cell, and tran mit data only in the fir t time lo t (TS 0) of certain GSM frame . Unlike TCH which are duplex, BCH only u e the forward link. Ju t a the forward control channel (FCC) in AMPS i u ed a a b eacon for all nearby mobile to camp on to, the BCH erve a a TDMS beacon chan nel for any nearby mobile to identify and lock on to. The BCH provide ynchron ization for all mobile within the cell and i occa ionally monitored by mobile in neighboring cell o that received power and MAHO deci ion may be made by o ut-of-cell u er . Although BCH data i tran mitted in TS 0, the other even tim e lot in a GSM frame for that ame ARFCN are available for TCH data, DCCH data, or are filled with dummy bur t . Furthermore, all eight time lot on all other ARFCN within the cell are available for TCH or DCCH data. The BCH i defined by three eparate channel which are give acce to TS 0 duri ng variou frame of the 51 frame equence. Figure illu trate how the BCH i a llocated frame . The three type of BCH are now de cribed.

GSM Control Channel

(CCH)

The following half rate peech and data channel are

upported:

e) Broadca t Control Channel (BCCH): The BCCH i a forward control channel that i u ed to broadca t information uch a cell and network identify, and op erating characteri tic of the cell (current control channel tructure, channel availability, and conge tion). The BCCH al o broadca t a li t of channel that are currently in u e within the cell. Frame 2 through frame 5 in a control mul tiframe (4 out of every 51 frame ) contain BCCH data. It hould be noted from f igure that TS 0 contain BCCH data during pecific frame , and contain other BC H channel (FCCH and SCH), common control channel (CCCH ), or an idle frame ( e nt every 51 t frame) during other pecific frame . f) Frequency Correction Channel (FCCH): The FCCH i a pecial data bur t wh ich occupie TS 0 for the very fir t GSM frame (frame 0) and i repeated every t en frame within a control channel multiframe. The FCCH allow each ub criber unit to ynchronize it internal frequency tandard (local o cillator) to the ex act frequency of the ba e tation. g) Synchronization Channel (SCH): SCH i broadca t in TS 0 of the frame imm ediately following the FCCH frame and i u ed to identify the erving ba e tati on while allowing each mobile to frame ynchronize with the ba e tation. The f rame number (FN), which range from 0 to 2,715,647, i ent with the ba e tatio n identity code (BSIC) during the SCH bur t. The BSIC i uniquely a igned to e ach BST in a GSM y tem. Since a mobile may be a far a 30 km away from a erv ing ba e tation, it i often nece ary to adju t the timing of a particular mob ile u er uch that the received ignal at the ba e tation i ynchronized with the ba e tation clock. The BS i ue coar e timing advancement command to the mobile tation over the SCH, a well. The SCH i tran mitted once every ten f rame within the control channel multiframe, a hown in figure. Common Control Channel (CCCH ): On the broadca t (BCH) ARFCN, the common contro l channel occupy TS 0 of every GSM frame that i not otherwi e u ed by the BCH or the idle frame. CCCH con i t of three different channel : the paging channe l (PCH), which i a forward link channel, the random acce channel (RACH) which i a rever e link channel, and the acce grant channel (AGCH), which i a forw ard link channel. A een in figure, CCCH are the mo t commonly u ed control c hannel a are u ed to page pecific ub criber , a ign ignaling channel to pecific u er , and receive mobile reque t for ervice. The e channel are de c ribed below: e) Paging Channel (PCH): The PCH provide paging ignal from the ba e tat ion to all mobile in the cell, and notifie a pecific mobile of an incoming ca ll which originate from the PSTN. The PCH tran mit the IMSI of the target ub criber, along with a reque t for acknowledgement from the mobile unit on the RA CH. Alternatively, the PCH may be u ed to provide cell broadca t ASCII text me age to all ub criber , a part of the SMS feature of GSM. f) Random Acce Channel (RACH): The RACH i a rever e link channel u ed by a ub criber unit to acknowledge a page from the PCH, and i al o u ed by mobil e to originate a call. The RACH u e a lotted ALOHA acce cheme. All mobil e mu t reque t acce or re pond to a PCH alert within TS 0 of a GSM frame. At the BTS, every frame (even the idle frame) will accept RACH tran mi ion from mobile during TS 0. In e tabli hing ervice, the GSM ba e tation mu t re pond to the RACH tran mi ion by allocating a channel and a igning a tand alone de dicated control channel (SDCCH) for ignaling during a call. Thi connection i confirmed by the ba e tation over the AGCH. g) Acce Grant Channel (AGCH): The AGCH i u ed by the ba e tation to pro vide forward link communication to the mobile, and carrie data which in truct the mobile to operate in a particular phy ical channel (time lot and ARFCN) wit h a particular dedicated control channel. The AGCH i the final CCCH me age e nt by the ba e tation before a ub criber i moved off the control channel. Th e AGCH i u ed by the ba e tation to re pond to a RACH ent by a mobile tation in a previou CCCH frame.

Dedicated Control Channel (DCCH ): There are three type of dedicated control c hannel in GSM, and, like traffic channel ( ee figure), they are bidirectional and have the ame format and function on both the forward and rever e link . Li ke TCH , DCCH may exit in any time lot and on any ARFCN except TS 0 of the BCH ARFCN. The tand along dedicated control channel (SDCCH ) are u ed for provid ing ignaling ervice required by the u er . The Slow and Fa t A ociated Cont rol Channel (SACCH and FACCH ) are u ed for upervi ory data tran mi ion bet ween the mobile tation and the ba e tation during a call. d) Stand alone Dedicated Control Channel (SDCCH ): The SDCCH carrie igna ling data following the connection of the mobile with the ba e tation, and ju t before a TCH a ignment i i ued by the ba e tation. The SDCCH en ure that the mobile tation and the ba e tation remain connected while the ba e tation and MSC verify the ub criber unit and allocate re ource for the mobile. The S DCCH can be thought of a an intermediate and temporary channel which accept a newly completed call from the BCH and hold the traffic while waiting for the ba e tation to allocate a TCH channel. The SDCCH i u ed to end authentication and alert me age (but not peech) a the mobile ynchronize it elf with the f rame tructure and wait for a TCH. SDCCH may be a igned their own phy ical c hannel or may occupy TS 0 of the BCH if there i low demand for BCH or CCCH traf fic. e) Slow A ociated Control Channel (SACCH): The SACCH i alway a ociated with a traffic channel or a SDCCH and map onto the ame phy ical channel. Thu , each ARFCH y tematically carrie SACCH data for all of it current u er . A in the USDC tandard, the SACCH carrie general information between the MS and BTS. On the forward link, the SACCH i u ed to end low but regularly changing control information to the mobile, uch a tran mit power level in truction an d pecific timing advance in truction for each u er on the ARFCN. The rever e SACCH carrie information about the received ignal trength and quality of the TCH, a well a BCH mea urement re ult from neighboring cell . The SACCH i tr an mitted during the thirteenth frame (and the twenty ixth frame when half rate traffic i u ed) of every peech / dedicated control channel multiframe, and wi thin thi frame, the eight time lot are dedicated to providing SACCH data to ea ch of the eight full rate (or ixteen half rate) u er on the ARFCN. f) Fa t A ociated Control Channel (FACCH ): FACCH carrie urgent me age , and contain e entially the ame type of information a the SDCCH. A FACCH i a igned whenever a SDCCH ha not been dedicated for a particular u er and the re i an urgent me age ( uch a a handoff reque t). The FACCH gain acce to a time lot by tealing frame from the traffic channel to which it i a igned. Thi i done by etting two pecial bit , called tealing bit , in a TCH forward channel bur t. If the tealing bit are et, the time lot i known to contain FACCH data, not a TCH, for that frame. 10. Explain the Frame Structure for GSM. Frame Structure for GSM Each u er tran mit a bur t of data during the time lot a igned to it. The e data bur t may have one of five pecific format , a defined in GSM [Ho d90]. Figure illu trate the five type of data bur t u ed for variou control and traffic bur t . Normal bur t are u ed for TCH and DCCH tran mi ion on b oth the forward and rever e link. FCCH and SCH bur t are u ed in TS 0 of peci fic frame ( hown Figure (a)) to broadca t the frequency and time ynchronizatio n control me age on the forward link. The RACH bur t i u ed by all mobile t o acce ervice from any ba e tation, and the dummy bur t i u ed a filter in formation for unu ed time lot on the forward link. Figure illu trate the data tructure within a normal bur t. It con i t of 148 bit which are tran mitted a a rate of 270.833333 kbp (an unu ed guar

d time of 8.25 bit i provided at the end of each bur t). Out of the total 148 bit per TS, 114 are information bearing bit which are tran mitted a two 57 b it equence clo e to the beginning and end of the bur t. The midamble con i t of a 26 bit training equence which allow the adaptive equalizer in the mobile or ba e tation receiver to analyze the radio channel characteri tic before de coding the u er data. On either ide of the midamble, there are control bit ca lled tealing flag . The e two flag are u ed to di tingui h whether the TS con tain voice (TCH) or control (FACCH) data, both which hare the ame phy ical ch annel. During a frame, a GSM ub criber unit u e one TS to tran mit, one TS to receive, and may u e the ix pare time lot to mea ure ignal trength on fiv e adjacent ba e tation a well a it own ba e tation.

Normal

FCCH bur t

SCH bur t

RACH bur t

Dummy bur t

Figure: Time lot data bur t in GSM

3 tart bit 58 mixed bit 26 training bit it 8.25 bit guard period

8 tart bit 3 top bit

41 bit of ynchronization 36 bit 68.25 bit extended guard period

of encrypted data

58 mixed bit

top b

3 tart bit 39 bit of encrypted data 64 bit of training of encrypted data 3 top bit 8.25 bit guard period

3 tart bit period

142 fixed bit

of all zeroe

top bit

8.25 bit guard

39 bit

3 tart bit 58 bit of encrypted data 26 training bit of encrypted data 3 top bit 8.25 bit guard period

58 bit

A hown in figure, there are eight time lot per TDMA frame, and the frame peri od i 4.615 m . A frame contain 8 x 156.25 = 1250 bit , although ome bit peri od are not u ed. The frame rate i 270.833 kbp / 1250 bit / frame, or 216.66 frame per econd. The 13th or 26th frame are not u ed for traffic, but for con trol purpo e . Each of the normal peech frame are grouped into larger tructu re called multiframe which in turn are grouped into uperframe and hyper fram e (hyper frame are not hown in figure). One multiframe contain 26 TDMA fram e , and one uperframe contain 51 multiframe , or 1326 TDMA frame . A hyperfra me contain 2048 uperframe , or 2,715,648 TDMA frame . A complete hyperframe i ent about every 3 hour , 28 minute , and 54 econd , and i important to GSM ince the encryption algorithm rely on the particular frame number, and uffici ent ecurity can only be obtained by u ing a large number of frame a provided by the hyperframe.

11. Explain the Signal Proce ing in GSM (OR) Explain all of the GSM operation from tran mitter to receiver. Signal Proce ing in GSM

Speech Coding - The GSM peech coder i ba ed on the Re idually Excited Linear P redictive Coder (RELP), which i enhanced by including a Long Term Predictor (LT P) [Hel89]. The coder provide 260 bit for each 20 m block of peech, which yield a bit rate of 13 kbp . Thi peech coder wa elected after exten ive u bjective evaluation of variou candidate coder available in the late 1980 . Pr ovi ion for incorporating half rate coder are included in the pecification . The GSM peech coder take advantage of the fact that in a normal conver ation, each per on peak on average for le than 40% of the time. By incorpo rating a voice activity detector (VAD) in the peech coder, GSM y tem operate in a di continuou tran mi ion mode (DTX) which provide a longer ub criber ba ttery life and reduce in tantaneou radio interference ince the GSM tran mitte r i not active during ilent period . A comfort noi e ub y tem (CNS) at the r eceiving end introduce a background acou tic noi e to compen ate for the annoyi

Figure illu trate all of the GSM operation

from tran mitter to receiver.

Figure how that the control multiframe pan 51 frame oppo ed to 26 frame 9120 m ) u ed by the traffic / dedicated ultiframe . Thi i done intentionally to en ure that any GSM er in the erving or adjacent cell) will be certain to receive tran mi ion from the BCH, no matter what particular frame or e u ing.

(235.365 m ), a control channel m ub criber (wheth the SCH and FCCH time lot they ar

ng witched muting which occur due to DTX. TCH/FS, SACCH, and FACCH Channel Coding The output bit of the peech coder are ordered into group for error protection, ba ed upon their ignificance in contr ibuting to peech quality. Out of the total 260 bit in a frame, the mo t impor tant 50 bit , called type Ia bit , have 3 parity check (CRC) bit added to them. Thi facilitate the detection of non correctable error at the receiver. The next 132 bit along with the fir t 53 (50 type Ia bit + 3 parity bit ) are rec orded and appended by four trailing zero bit , thu providing a data block of 18 9 bit . Thi block i then encoded for error protection u ing a rate 1/2 convol utional encoder with con traint length K = 5, thu providing a equence of 378 b it . The lea t important 78 bit do not have any error protection and are conca tenated to the exi ting equence to form a block of 456 bit in a 20 m frame. The error protection coding cheme increa e the gro data rate of the GSM pee ch ignal, with channel coding, to 22.8 kbp . Thi error protection cheme a d e cribed i illu trated in figure.

Channel Coding for Data Channel The coding provided for GSM full rate data chan nel (TCH/F9.6) i ba ed on handling 60 bit of u er data at 5 m interval , in accordance with the modified CCITT V.110 modem tandard. A de cribed by Steele [Ste94], 240 bit of u er data are applied with four tailing bit to a half rat e punctured convolutional coder with con traint length K = 5. The re ulting 488 coded bit are reduced to 456 encoded data bit through puncturing (32 bit are not tran mitted), and the data i eparated into four 114 bit data bur t that are applied in an interleaved fa hion to con ecutive time lot . Channel Coding for Control Channel GSM Control Channel me age are def ined to be 184 bit long, and are encoded u ing a hortened binary cyclic fire c ode, followed by a half rate convolutional coder. The fire code u e the generator polynomial

Which produce 184 me age bit , followed by 40 parity bit four tail bit are a dded to clear the convolutional coder which follow , yielding a 228 bit data blo ck. Thi block i applied to a half rate K = 5 convolutional code (CC(2,1,5)) u ing the generator polynomial G0(x) = 1 + x3 + x4 and G1(x) = 1+ x + x3 + x4 (w hich are the ame polynomial u ed to code TCH type Ia data bit ). The re ultin g 456 encoded bit are interleaved onto eight con ecutive frame in the ame man ner a TCH peech data. Interleaving In order to minimize the effect of udden fade on the rece ived data, the total of 456 encoded bit within each 20 m peech frame or contr ol me age frame are broken into eight 57 bit ub block . The e eight ub block which make up a ingle peech frame are pread over eight con ecutive TCH time lot . (i.e., eight con ecutive frame for a pecific TS). If a bur t i lo t due to interference or fading, channel coding en ure that enough bit will ti ll be received correctly to allow the error correction to work. Each TCH time lot carrie two 57 bit block of data from two different 20 m (456 bit) peech (or control) egment . Figure illu trate exactly how the peech frame are dia gonally interleaved within the time lot . Note that TS 0 contain 57 bit of d ata from the 0th ub block of the nth peech coder frame (denoted a a in the figu re) and 57 bit of data from the 4th ub block of the (n-1) t peech coder fram e (denoted a b in the figure) Ciphering Ciphering modifie the content of the eight interleaved block through the u e of encryption technique known only to the particular mobile tation and ba e tran ceiver tation. Security i further enhanced by the fact t hat the encryption algorithm i changed from call to call. Two type of cipheri ng algorithm , called A3 and A5, are u ed in GSM to prevent unauthorized network acce and privacy for the radio tran mi ion re pectively. The A3 algorithm i u ed to authenticate each mobile by verifying the u er pa code within the SI M with the cryptographic key at the MSC. The A5 algorithm provide the crambli ng for the 114 coded data bit ent in each TS.

Frame Number

0a 5a

4b 1b

1a 6a

5b 2b

2a 7a

6b 3b

3a

7b

4a

0b

Bur t Formatting Bur t formatting add binary data to the ciphered block , in or der to help ynchronization and equalization of the received ignal. Modulation The modulation cheme u ed by GSM i 0.3 GMSK, where 0.3 de cribe th e 3 dB bandwidth of the Gau ian pul e haping filter with relation to the bit r ate (e.g., BT = 0.3). A de cribed in GMSK i a pecial type of digital FM modu lation. Binary one and zero are repre ented in GSM by hifting the RF carrier by 67.708 kHz. The channel data rate of GSM i 270.833333 kbp , which i exact ly four time the RF frequency hift. Thi minimize the bandwidth occupied by the modulation pectrum and hence improve channel capacity. The MSK modulated ignal i pa ed through a Gau ian filter to mooth the rapid frequency tran it ion which would otherwi e pread energy into adjacent channel .

Frequency Hopping Under normal condition , each data bur t belonging to a partic ular phy ical channel i tran mitted u ing the ame carrier frequency. However, if u er in a particular cell have evere multipath problem , the cell may be d efined a a hopping cell by the network operator, in which ca e low frequency h opping may be implemented to combat the multipath or interference effect in tha t cell. Frequency hopping i carried out on a frame by frame ba i , thu hoppin g occur at a maximum rate of 217.6 hop per econd. A many a 64 different ch annel may be u ed before a hopping equence i repeated. Frequency hopping i completely pecified by the ervice provider. Equalization Equalization i performed at the receiver with the help of the trai ning equence tran mitted in the midamble of every time lot. The type of equa lizer for GSM i not pecified and i left up to the manufacturer. Demodulation The portion of the tran mitted forward channel ignal which i of i ntere t to a particular u er i determined by the a igned TS and ARFCN. The ap propriate TS in demodulated with the aid of ynchronization data provided by the bur t formatting. After demodulation, the binary information i deciphered, de -interleaved, channel decoded, and peech decoded. 12. Explain in detail about CDMA Digital Cellular Standard (IS 95). CDMA Digital Cellular Standard (IS 95) Code Divi ion Multiple Acce (CDMA) offer many advantage over TDMA an d FDMA. A US digital cellular y tem ba ed on CDMA which promi ed increa ed cap acity [Gil91] wa tandardized a Interim Standard 95 (IS 95) by the US Telecomm unication Indu try A ociation (TIA) [TIA93]. Like IS 136, the IS 95 y tem wa de igned to be compatible with the exi ting US analog cellular y tem (AMPS) f requency band, hence mobile and ba e tation can be economically produced for dual mode operation. Pilot production, CDMA/AMPS, dual mode phone were made av ailable by Qualcomm in 1994, and a of 2001, there are over 80 million CDMA ub criber worldwide. IS 95 allow each u er within a cell to u e the ame radio channel, and u er in adjacent cell al o u e the ame radio channel, ince thi i a direct equence pread pectrum CDMA y tem. CDMA completely eliminate the need for f requency planning within a market. To facilitate graceful tran ition from AMPS to CDMA, each IS 95 channel occupie 1.25 MHz of pectrum on each one-way link, or 10% of the available cellular pectrum for a US cellular provider (recall, th e US cellular y tem i allocated 25 MHz and each ervice provider receive half the pectrum or 12.5 MHz). In practice, AMPS carrier mu t prove a 270 kHz gua

rd band (typically 9 AMPS channel ) on each ide of the pectrum dedicated for I S 95. IS 95 i fully compatible with the IS 41 networking tandard. Unlike other cellular tandard , the u er data rate (but not the channel chip rate) change in real time, depending on the voice activity and requiremen t in the network. Al o, IS 95 u e a different modulation and preading techni que for the forward and rever e link . On the forward link, the ba e tation i multaneou ly tran mit the u er data for all mobile in the cell by u ing a diff erent preading equence for each mobile. A pilot code i al o tran mitted imu ltaneou ly and at a higher power level, thereby allowing all mobile to u e cohe rent carrier detection while e timating the channel condition . On the rever e l ink, all mobile re pond in an a ynchronou fa hion and have ideally a con tant ignal level due to power control applied by the ba e tation. The peech coder u ed in the IS 95 y tem i the Qualcomm 9600 bp Code Excited Linear Predictive (QCELP) coder. The original implementation of thi vo coder detect voice activity, and reduce the data rate to 1200 bp during ilen t period . Intermediate u er data rate of 2400, 4800, and 9600 bp are al o u ed for pecial purpo e . A 14,400 bp coder which u e 134 kbp of peech data (QCELP 13) wa introduced by Qualcomm in 1995. 4. Explain GSM Code . The original peech coder u ed in the pan European digital cellular tan dard GSM goe by a rather grandio e name of regular pul e excited long term pred iction (RPE LTP) codec. Thi codec ha a net bit rate of 13 kbp and wa cho en after conducting exhau tive ubjective te t [Col89] on variou competing codec . More recent GSM upgrade have improved upon the original codec pecification .

The RPE LTP codec [Var88] combine the advantage of the earlier French propo ed ba eband RELP codec with tho e of the multipul e excited long term prediction ( MPE LTP) codec propo ed by Germany. The advantage of the ba eband RELP codec i that it provide good quality peech at low complexity. The peech quality of a RELP codec i however, limited due to the tonal noi e introduced by the proce of high frequency regeneration and by the bit error introduced during tran mi ion. The MPE LTP technique, on the other hand, produce excellent peech qual ity at high complexity and i not much affected by bit error in the channel. B y modifying the RELP codec to incorporate certain feature of the MPE LTP codec, the net bit rate wa reduced from 14.77 kbp to 13.0 kbp without lo of quali ty. The mo t important modification wa the addition of a long term prediction loop. The GSM codec i relatively complex and power hungry. Figure how bloc k diagram of the peech encoder [Ste94]. The encoder i compri ed of four major proce ing block . The peech equence i fir t pre empha ized, ordered into egment of 20 m duration, and then Hamming windowed. Thi i followed by hort term prediction (STP) filtering analy i where the logarithmic area ratio (LAR ) of the reflection coefficient rn(k) (eight in number) are computed. The eig ht LAR parameter have different dynamic range and probability di tribution fun ction , and hence all of them are not encoded with the ame number of bit for t ran mi ion. The LAR parameter are al o decoded by the LPC inver e filter o a to minimize the error en.

LTP analy i which involve finding the pitch period pn and gain factor gn i then carried out uch that the LTP re idual rn i minimized. To minimize rn, pitch extraction i done by the LTP by determing that value of delay, D, whi ch maximize the cro correlation between the current STP error ample, en, and a previou error ample en-D. The extracted pitch pn and gain gn are tran mitt ed and encoded at a rate of 3.6 kbp . The LTP re idual, rn, i weighted and dec

Figure how a block diagram of the GSM peech decoder [Ste94]. It con i t of four block which perform operation complementary to tho e of the encod er. The received excitation parameter are RPF decoded and pa ed to the LTP y nthe i filter which u e the pitch and gain parameter to ynthe ize the long-te rm ignal. Short term ynthe i i carried out u ing the received reflection co efficient to recreate the original peech ignal.

Every 260 bit of the coder output (i.e., 20m block of peech) are ordered, de pending on their importance, into group of 50, 132, and 78 bit each. The bit in the fir t group are very important bit called type Ia bit . The next 132 b it are important bit called Ib bit , and the la t 78 bit are called type II b it . Since type Ia bit are the one which effect peech quality the mo t they have error detection CRC bit added. Both Ia and Ib bit are convolutionally en coded for forward error correction. The lea t ignificant type II bit have no error correction or detection. 10. Explain in detail about capacity of Cellular CDMA.

ompo ed into three candidate excitation equence ce are identified, and the one with the highe t the LTP re idual. The pul e in the excitation highe t amplitude, quantized, and tran mitted at

. The energie of the e equen energy i elected to repre ent equence are normalized to the a rate of 9.6 kbp .

Capacity of Cellular CDMA The capacity of CDMA y tem i interference limited, while it i bandwi dth limited in FDMA and TDMA. Therefore, any reduction in the interference will cau e a linear increa e in the capacity of CDMA. Put another way, in a CDMA y tem, the link performance for each u er increa e a the number of u er decrea e . A traightforward way to reduce interference i to u e multi ectorized ant enna , which re ult in patial i olation of u er . The directional antenna re ceive ignal from only a fraction of the current u er , thu leading to the red uction of interference. Another way of increa ing CDMA capacity i to operate i n a di continuou tran mi ion mode (DTX), where advantage i taken of the inter mittent nature of peech. In DTX, the tran mitter i turned off during the peri od of ilence in peech. It ha been ob erved that voice ignal have a duty f actor of about 3/8 in landline network [Bra68], and 1/2 for mobile y tem , whe re background noi e and vibration can trigger voice activity detector . Thu , t he average capacity of a CDMA y tem can be increa ed y a factor inver ely propo rtional to the duty factor. While TDMA and FDMA reu e frequencie depending on the i olation between cell provided by the path lo in terre trial radio propa gation, CDMA can reu e the entire pectrum for all cell , and thi re ult in an increa e of capacity by a large percentage over the normal frequency reu e facto r. A number of recent text provide detail on CDMA capacity and y tem de ign [Lib99], [Kim00], [Gar00], [Mo101]. For evaluating the capacity of a CDMA y tem, fir t con ider a ingle ce ll y tem [Gil91]. The cellular network con i t of a large number of mobile u er communicating with a ba e tation. (In a multiple cell y tem, all the ba e tation are interconnected y the mobile witching center). The cell- ite tran mitter con i t of a linear combiner which add the pread ignal of the indiv idual u er and al o u e a weighting factor for each ignal for forward link po wer control purpo e . For a ingle cell y tem under con ideration, the e weigh ting factor can be a umed to be equal. A pilot ignal i al o included in the cell ite tran mitter and i u ed by each mobile to et it own power control f or the rever e link. For a ingle cell y tem with power control, all the igna l on the rever e channel are received at the ame power level at the ba e tati on.

Let the number of u er be N. Then, each demodulator at the cell ite r eceive a compo ite waveform containing the de ired ignal of power S and (N-1) interfering u er , each of which ha power, S. Thu , the ignal to noi e ration i [Gil91]. (1) In addition to SNR, bit energy to noi e ratio i an important parameter in communication y tem . It i obtained by dividing the ignal power by the ba eband information bit rate, R, and the interference power by the total RF bandw idth, W. The SNR at the ba e tation receiver can be repre ented in term of Eb / N0 given by (2) Equation (2) doe not take into account the background thermal noi e, , i n the pread bandwidth. To take thi noi e into con ideration, Eb / N0 can be r epre ented a (3)

The number of u er that can acce

the

y tem i

thu given a

(4)

In order to achieve an increa e in capacity, the interference due to oth er u er hould be reduced. Thi can be done by decrea ing the denominator of E quation (1) or (2). The fir t technique for reducing interference in antenna ectorization. A an example, a cell ite with three antenna , each having a bea m width of 120 , ha interference N0 which i one third of the interference receive d by an omnidirectional antenna. Thi increa e the capacity by a factor of thr ee, ince three time a many u er may now be erved within a ector while matc hing the performance of the omnidirectional antenna y tem. Looking at it anoth er way, the ame number of u er in an omnidirectional cell may now be erved in 1/3rd the area. The econd technique involve the monitoring of voice activity uch that each tran mitter i witched off during period of no voice activity. Voice activity i denoted by a factor , and the interference term in Equation ( 2) become (NS 1) , where N i the number of u er per ector. With the u e of t he e two technique , the new average value of Eb / N0 within a ector i given a (5) When the number of u er i larger and the y tem i interference limite d rather than noi e limited, the number of u er can be hown to be (6) If the voice activity factor i a umed to have a value of 3/8, and thre e ector per cell ite are u ed, Equation (6) demon trate that the SNR increa e by a factor of eight, which lead to an eight fold increa e in the number of u er compared to an omnidirectional antenna y tem with no voice activity detec tion. CDMA Power Control In CDMA, the y tem capacity i maximized if each mobile tran mitter power level i controlled o that it ignal arrive at the cell ite wi th the minimum required ignal to interference ratio [Sal91]. If the ignal pow er of all mobile tran mitter within an area covered by a cell ite are control led, then the total ignal power received at the cell ite from all mobile will be equal to the average received power time the number of mobile operating in the region of coverage. A tradeoff mu t be made if a mobile ignal arrive at the cell ite with a ignal that i too weak, and often the weak u er will be dr opped. If the received power from a mobile u er i too great, the performance o f thi mobile unit will be acceptable, but it will add unde ired interference to all other u er in the cell. 11. Explain the Capacity of CDMA with Multiple Cell . In actual CDMA cellular y tem that employ eparate forward and rever e link , neighboring cell hare the ame frequency, and each ba e tation contro l the tran mit power of each of it own in cell u er . However, a particular b a e tation i unable to control the power of u er in neighboring cell , and th e e u er add to the noi e floor and decrea e capacity on the rever e link of th e particular cell of intere t. Figure illu trate an example of how u er in ad jacent cell may be di tributed over the coverage area. The tran mit power of each out-of-cell u er will add to the in cell interference (where u er are unde r power control) at the ba e tation receiver. The amount of out-of-cell interf erence determine the frequency reu e factor, f, of a CDMA cellular y tem. Ide ally, each cell hare the ame frequency and the maximum po ible value of f (f

Where W/R i called the proce ing gain. The background noi e determine ll radiu for a given tran mitter power.

the ce

=1) i achieved. In practice, however, the out-of-cell interference reduce f ignificantly. In contra t to CDMA y tem which u e the ame frequency for each cell, narrowband FDMA / FDD y tem typically reu e channel every even cell , in which ca e f i imply 1/7.

The frequency reu e factor for a CDMA y tem on the rever e link can be defined a [Rap92b] (1)

F = f x 100 %

In equation (1), N0 i the total interference power received from the N1 in cell u er , Ui i the number of u er in the ith adjacent cell, and Nai i the average interference power for a u er located in the ith adjacent cell. Wi thin the cell of intere t, the de ired u er will have the ame received power a the N 1 unde ired in cell u er when power control i employed, and the average received power from u er in an adjacent cell can be found by (3) Where Nij i the power received at the ba e tation of intere t from the jth u er in the ith cell. Each adjacent cell may have a different number of u er , and each out-of-cell u er will offer a different level of interference depe nding on it exact tran mitter power and location relative to the ba e tation o f intere t. The variance of Nai can be computed u ing tandard tati tical tech nique for a particular cell. An analy i de cribed by Liberti and Rappaport and Mil tein [Lib94b] [Ra p92b] [Lib99] u e a recur ive geometric technique to determine how the propagat ion path lo impact the frequency reu e of a CDMA y tem by con idering the in terference from both in cell and out of cell u er . The geometric technique, ca lled the concentric circle cellular geometry, con ider all cell to have equal geographic area and pecifie the cell of intere t to be a circular cell, which i located in the center of all urrounding cell . Interfering cell are wedge haped and are arranged in layer around the center cell of intere t. Figure ill u trate the concentric circle geometry for ingle layer of adjacent cell .

Let the center cell of intere t have radiu R, and a ume that there i

And the frequency reu e efficiency, F, i defined a

(2)

ome clo e-in di tance d0 uch that all u er in the center cell are located no clo er than d0 meter to the center ba e tation and that all u er in the cell of intere t are located a di tance d from the ba e tation of intere t uch that d0 d R. Then, a fir t layer of adjacent interfering cell i found on R d 3R, a econd layer i located on 3R d 5R, and the ith interfering layer i located on (2i 1) R d (2i + 1)R. In each urrounding layer, there are Mi adjacent cell , where i denote the layer number. If d0 i a umed to be much le than R, then the area A of the center cell of intere t i A = R2 - d20 R2 (4)

Within the fir t urrounding layer of cell , let A1 denote the entire ar ea of the region. If each cell in the fir t urrounding layer i to have the a me area A, then there hould be M1 wedge haped cell that each pan a particula r angle 1. Neglecting d0, it i clear that for the fir t urrounding layer. A1 = (3R)2 - R2 = M1A (5) (6)

1 = 2 /M1

Solving Equation (5) and (6), M1 = 8 and 1 = 45 . In other word , there are eight wedge haped cell in the fir t urrounding layer, each cell covering a pan of 45 . By recur ion, it can be hown that for the econd, and all ub equent urro unding layer , the area of the ith urrounding layer i related to the number of cell within the layer by Ai = MiA = iM1 A = i8A i i = 1/i = /4i 1 (7) (8)

The concentric circle geometry i appealing becau e, once the area of a cell of intere t i pecified; it become ea y to pecify an entire y tem of u rrounding cell that each occupy the ame coverage area. Furthermore, ince eac h adjacent cell ha the ame geometry and angular pan a any other cell within the urrounding layer, and ince each cell within a particular layer ha the am e radial geometry to the center cell of intere t, it become po ible to con ide

r the interference effect of ju t a ingle cell within the urrounding layer. One can imply multiply the effect of a ingle cell by the number of cell with in the urrounding layer.

If K i the u er den ity (i.e., the number of u er per unit area), then the total number of u er within the center cell i given by U = KA. If it i a umed that all cell have the ame number of u er , then in the fir t urround ing layer there will al o be KA u er . Weighting factor can be u ed to break u p the di tribution of adjacent cell u er between the inner and outer ector . In the fir t urrounding layer, the inner and outer ector of each cell have ar ea given by A1in / M1 =[ ( (2R)2 And A1 out / M1 = [( (3R)2 - (2r)2)[/8 = 5A / 8 (10) R2)/8 = 3A / 8 (9)

For each fir t layer cell to po e U = KA u er , weighting factor for the u er den ity within the inner (W1in) and outer (W1out) ector may be appli ed uch that U = KA = (KW1inA1in) / M1 + (KW1outA1out) / M1 And U = KA = KA [3/8W1in + 5 / 8 W1out) (12) U ing Equation (12), it can be een that if W1in = 1 and W1out = 1, then 3/8 of the u er will be in the inner ector and 5 / 8 of the u er will be in the outer ector. Thi can be though of a an optimi tic condition (or upper bo und) for frequency reu e, ince not even half of the interferer are le than 2 R away from the ba e tation, and 5 / 8 of the u er are farther than 2R away an d will offer maller level of interference to the center cell. However, if Win = 4/3 and W1out = 4/5, then half of the u er in the fir t layer cell will be clo er than 2R and half will be farther than 2R from the center ba e tation (th i corre pond to the ca e of hexagonal cell , where half of the u er would be clo er to the ba e tation of intere t, and half would be farther away from the ba e tation of intere t). For a wor t ca e interference cenario, all of the U u er in each of the fir t layer cell would be located in the inner ector (th ereby providing more interference to the center cell due to maller di tance and (11)

It i often u eful to con ider the adjacent cell interference effect fo r variou di tribution of u er within the interfering cell . Thi allow one t o determine wor t ca e frequency reu e and provide flexibility for determining a range of po ible value of f for variou u er di tribution . The concentric circle geometry allow the interfering layer to be broken into two ublayer , a n inner ublayer which i on (2i 1) R d 2iR d0, and the outer ublayer which i on 2iR + d0 d (2i + 1) R. Thi partitioning i hown in figure (a de cribed u b equently, there i a mall forbidden zone around the ba e tation of the adjac ent cell ). The partitioning of layer provide two ector within each wedge haped cell in a given layer: the inner ector (which contain a maller fraction of the area of the cell) and the outer ector (which contain a greater fractio n of the area of the cell). Since each cell contain the ame area a the cente r cell, it i clear that for a uniform di tribution of u er over an adjacent ce ll, the inner ector will contain fewer u er than the outer ector, and thi wi ll certainly impact the interference power received at the center cell ba e tat ion. To account for a wide range of u er di tribution in the interfering layer , weighting factor are u ed to redi tribute u er in the inner and outer ecto r of an adjacent cell.

Weighting Factor

maller path lo ). The weighting factor account for thi wor t ca e by etti ng W1in = 8/3 and W1out = 0. It i left a an exerci e to determine appropriate weighting factor for the econd and ub equent layer cell .

The area of the center and wedge haped cell in the concentric circle g eometry exceed the area of a traditional hexagonal cell de cribed. Wherea a h exagonal cell occupie an area of Ahex = = 2.598R2, the cell in the concentri c circle geometry each po e and area of R2. Thu , the center cell occupie an area which i 1.21 time a large a a traditional hexagonal cell and the fir t layer of eight wedge haped cell hown in figure occupie the area of 9.666 he xagonal cell . Since frequency reu e i ba ed on the relative amount of interfe rence from neighboring cell , a long a the number of u er and the coverage ar ea can be accurately repre ented and caled, the particular geometry doe not dr amatically impact capacity prediction [Lib99], [Lib94b], [Rap92b]. Thu , the c oncentric circle geometry offer analytical advantage for analyzing the effect of co-channel cell . U ing Concentric Circle Geometry to Find CDMA Capacity To find the capacity of a multicell CDMA y tem, the concentric circle g eometry can be u ed in conjunction with a propagation path lo model to determi ne interference from adjacent cell u er . Then, u ing Equation (1), the frequen cy reu e factor can be found. Note that the in-cell interference power N0 i i mply given by (13) Where P0 i the power received from any one of the U u er in the center cell ( ince all u er are a umed to be under power control and thu provide th e ame received power at the ba e tation receiver). In general, it i practica l to a ume that any adjacent cell will al o contain U u er and will receive p ower P0 from each of it own in cell u er . In the adjacent cell , each ub cri ber i under power control within it own cell, and i a di tance d from it own ba e tation. Since propagation path lo law are ba ed on all di tance great er than d0, a mall forbidden zone having width 2d0 i a umed to exi t in all urrounding ring . The forbidden zone i a mall annulu in each layer which i a umed not to contain u er , o that any propagation path lo model may be u e d in analy i without having d < d0. It i ea y to how that when d0 < R, the m all forbidden zone occupie negligible area and provide virtually the ame inte rference re ult a for the ca e when the forbidden zone i a circle of radiu d 0 around each adjacent cell.

(14) (15)

A light approximation for d i cent cell u er to it own ba e tation. e computed for inner and outer ector u l. U ing the law of co ine , it can be yer.

made when computing the power of the adja Figure illu trate how the di tance ar er within a fir t layer interfering cel hown that within any cell in the ith la

Figure: Geometry for computing di tance between adjacent u er and the center ba e tation of intere t. By breaking the adjacent cell into inner and outer ec tor , the di tribution of u er within neighboring cell may be varied. Then, u ing d and d, the interference power P0, i, j at the center cell f rom the jth u er in the ith interfering cell can be given by

(16) Where n i the propagation path lo exponent de cribed in d i a functio n of a given by equation (14) and (15). The two factor on the right ide of e quation (16) repre ent the actual tran mitter power radiated by the jth ub crib er in the ith cell, multiplied by the propagation path lo from that ub criber to the center ba e tation receiver. By evaluating equation (16) for each u er in an adjacent cell, it i po ible to compute Nai u ing equation (3) and then apply equation (1) to determine f. Simulation which con idered variou u er weighting factor , a well a path lo exponent of n = 2, 3, and 4, and varying cell ize were carried out by Rappaport and Mil tein [Rap92b]. Table indicate typical re ult , which how that f can range between 0.316 and 0.707, depending on the path lo exponent a nd the di tribution of u er . Thu , while a ingle cell CDMA y tem offer ide al frequency reu e (f=1), the actual frequency reu e i a trong function of u e r di tribution and path lo . Table: Frequency Reu e Factor for Rever e Channel of CDMA Cellular Sy tem, a a Function of n for Two Sy tem Implementation (from [Rap92b] IEEE)

W1 W2 W1 W2 W1 W2 2 2 2 10 10 10

= = = = = =

3.0 0.0 1.38 0.78 1.0 1.0 2 3 4 2 3 4

Hex Upper bound 0.316 0.408 0.479 0.308 0.396 0.462 0.425 0.558 0.646 0.419 0.550 0.634 0.462 0.613 0.707 0.455 0.603 0.695

d (km) n

Frequency reu e efficiency Lower bound