You are on page 1of 6

Head Exposure to Cellular Telephones: A System-Level Study

Hayat Abdulla and Renny E. Badra Departamento de Electrnica y Circuitos Universidad Simon Bolivar Caracas, Venezuela habdulla@usb.ve, renny@usb.ve
AbstractThis work is aimed at quantifying the effects of a number of system-level parameters and conditions on the amount of non-ionizing radiation dissipated in the head of cell phone users in the 850 MHz band. The three major cellular technologies of today, namely, GSM, CDMA2000 1X and UMTS, are evaluated. Five randomly chosen commercial products from each technology are investigated in terms of their Specific Absorption Rate (SAR) and maximum transmit power, as reported by the US Federal Communications Commission (FCC). A statistical system model for urban and suburban cells is used to calculate, with the help of the uplink power budget and the corresponding power control mechanisms, the statistics of the phone transmit power via Monte Carlo simulations. A linear model relating SAR and transmitted power is then used to obtain SAR statistics over the entire cell. Results show some minor differences among the technologies considered, and also between urban and suburban environments, but more importantly, they show a strong correlation between SAR levels and the percentage of cell area that is covered, which in turn depends on key system design parameters such as cell density, cell tower height, and the use of certain RF techniques. Keywords-component; Specific Absorption Rate; Non-ionizing radiation; GSM; CDMA2000 1X; UMTS.

strongly enforced the compliance of standards for maximum level of emissions by cell phone manufacturers. However, little attention has been paid to the potential benefits derived from actions and measures that can be implemented by cellular operators in order to reduce the level of emissions on the head of cell phone users, which is precisely the subject of this effort. This work is aimed at quantifying the influence of system design parameters and strategies on the level of head radiation on cellular voice services of three major cellular technologies. In this article, Sect. II establishes a key relationship between transmit power and head emissions. Sect. III describes the radio propagation model for the cellular channel, while Sect. IV discusses the cellular uplink budget and Sect. VI presents the overall simulation strategy. Sect. VII contains the main results, and Sect. VIII states the conclusions from the study. II. SAR AND TRANSMIT POWER The Specific Absorption Rate (SAR) is the most common measure of the level of exposure of organic tissue to nonionizing electromagnetic radiation [1]. It is defined as the amount of power dissipated per unit mass, and its units are watts per kilogram (W/kg). The effect of non-ionizing radiation over organic tissue is heat. Experimentally, SAR is obtained by taking an average of the power dissipated in heat over a certain tissue mass (typically 1 or 10 grams). Human dummies called phantoms filled with liquids emulating the electrical properties of the body part under study are used to perform SAR measurements, following tightly established procedures. In the US, the Federal Communication Commission (FCC) sets the parameters and procedures for SAR measurements, and also publishes the results for all commercially available cell phones. Head SAR measurements are given a PASS grade if they fall below the 1.6 W/kg. limit set by the American National Standards Institute (ANSI) for head radiation over general population users [2,3]. SAR measurements for every model of cell phone are obtained at maximum transmit power, and are tested under a number of different scenarios and conditions. For a given product to be authorized for commercial use, all of its SAR results are required to pass [3]. Heat produced from electromagnetic radiation is the result of electric currents induced in the organic tissue by the electric

I. INTRODUCTION Public concern regarding the potential harmful effects of electromagnetic emissions from cell phones on the head of voice service users has triggered in the past two decades a large number of research efforts, with results that can be qualified as inconclusive. While some studies indicate no relationship between cell phone use and the incidence of brain tumors, some others suggest there is a slight to moderate statistical correlation between the two. It can be stated that the question of potential health risks associated to using cell phones close to the head is still open. Some of these studies have identified three factors as potentially influential with regard to the possible health risks associated to cell phone use: the level of radiation, the duration of the exposure and the age of the user. While the two latter factors are entirely governed by consumer habits, the first one (level of emissions) can be effectively controlled by manufacturers and service providers. Regulatory agencies have

978-1-4244-7173-7/10/$26.00 2010 IEEE

and magnetic fields resulting from the close presence of the cell phone antenna. Thus, there is a direct relationship between the power emitted by the device and the heat that it produces in the neighboring tissue, as measured by its SAR. Furthermore, it can be stated that such relationship is strictly linear, that is

SAR = Pt K SAR ,

(1)

the mobile transmitter. It is accurately described by a random variable (RV) with a log-normal probability distribution [4], that is, the distribution of the RV expressed in dB is normal. In this study, the mean and standard deviation of such RV have been set to 0 dB and 8 dB, respectively, in agreement with widely used models [5].
TABLE. 1. PARAMETERS OF THE W-I PROPAGATION MODEL Parameter Frequency Cell Radius Mobile station antenna height Base station antenna height Angle between LOS path and street Average building height Inter-building distance Street width 24 m 15 m 30 m 30 m 90 9m 15 m Urban 1-2 km Suburban 2.5 5 km 1.5 m 20 m

where SAR is the device measured SAR under a certain set of conditions, <Pt> is the device average transmit power at the time of the measurement, and KSAR is a proportionality constant that depends on the test conditions and the device itself. Once an experimental SAR measurement and its corresponding transmit power are available, the parameter KSAR for that product under such conditions is obtained from (1). As the phone moves through the cell, its transmit power changes as a result of the corresponding power control mechanisms. However, the linear model in (1) makes it possible to obtain an estimate of the SAR for that phone in any location in the cell, once the transmit power for such location is calculated via system-level modeling and simulation. III. PROPAGATION MODEL The uplink cellular propagation model employed in all simulations accounts for signal power loss incurred between the transmit and receive antennas, that is, excluding all cellular system equipment. It is comprised of five terms,

850 MHz

LP = LD + LSh + LS + LPn + LB

(2)

where LD is the distance-dependent propagation loss, LSh is the shadowing factor, LS is the small-scale fading component, LPn is the building/car penetration loss and LB is the power loss due to interaction with the users body. All five terms in (1) are given in dB. A brief description of each of them is presented next. The distance-dependent propagation loss LD is estimated using the Walfish-Ikegami (W-I) model proposed by COST231 [4]. Such model considers two propagation scenarios: lineof-sight (LOS) and obstructed or non-line-of-sight (NLOS). The LOS model is essentially an adjusted free-space propagation model, while the NLOS model considers, in addition to free-space losses, those caused by energy dispersion and diffraction from buildings and obstructions. The W-I model has been found to be rather accurate in medium to large urban cells, for distances between 20 and 5000 meters. Table I shows the parameters employed in this work for the distancedependent W-I model. Before describing the signal fading and loss components of the channel model, it should be remarked that loss (as well as fading) margins cannot be directly used in this type of study, which is not concerned with estimating the maximum allowable propagation loss, as usual in regular link budget analysis; instead, this study requires the statistical estimation of the actual propagation loss, which in turn necessitates the use of statistical models and Monte Carlo trials, as will be explained. The shadowing factor LSh accounts for large-scale variations of signal strength due to obstructions located in the vicinity of

The small-scale fading component LS is aimed at modeling signal variations caused by multipath propagation over short distances. The amplitude of this fading factor is modeled by a Rice RV in LOS siuations, and a Rayleigh RV in NLOS scenarios [4]. In the case of the Rice RV, the ratio between the amplitude of the specular component and the RV standard deviation is set to 2. The mean of both Rice and Rayleigh RVs are set to 1, so that the amplitude of the faded signal could be obtained by simply multiplying the incoming signal amplitud by the corresponding RV. The small-scale fading component LS is obtained by taking 20 log () of the corresponding RV. It should be noted that both LSh and LS are RVs that may take on either possitive or negative values (in dB), which implies that such shadowing factors may actually represent a gain instead of a loss. Such gain (or loss) must be interpreted as relative to the average power of the signal for that particular path distance. The building penetration power loss has been reported to be about 10 dB, with additional losses as a function of the distance to the nearest window [6]. On the other hand, if the user is inside a vehicle, the mean penetration loss is estimated to be around 8 dB [7]. Accordingly, in this work the penetration loss (building or vehicle) is modeled as a uniform RV taking values between 0 and 12 dB in NLOS situations, and is set to zero for LOS trajectories. As seen, three of the terms in the proposed propagation model (2) require the determination of the type of propagation trajectory (LOS/NLOS) incurred by the signal. A two-tier statistical model for this binary RV has been used in this study, which is in agreement with previous models [5]. It postulates that as mobile units move closer to the cell tower, the probability of having a LOS propagation path increases. Accordingly, the probability of NLOS for distances above a certain threshold is set to one, and then it decreases linearly as the path distance drops below such threshold, becoming zero in the vicinity of the cell tower. The threshold distance is set to half the cell radius in all simulations, which results in 75% of NLOS locations in all cells, regardless of their size.

Finally, the RF power loss caused by electromagnetic interaction with the human body of the user LB is set to a constant value of 3 dB [8].

handover (SHO) in progress. In both CDMA systems the effect of antenna diversity gain is included in the estimated (Eb/No)req, so it is not explicitly considered as a part of the link budget itself. BS antenna gain is obtained by measuring the angle of arrival of the direct path between MS and BS, and reading the corresponding gain from the horizontal radiation pattern of the selected antenna. Antenna data has been extracted from a commercial product in the 850 MHz band [9], and is the same for all three technologies. About one-third of the mobiles in any given CDMA cell are typically in SHO situation. The maximum value of SHO gain has been estimated to be in the range of 2 to 4 dB, for both 1X and UMTS, although its exact value depends on numerous factors [8, 10]. The variability of this parameter has made it difficult for researchers to come up with a statistical model for its behavior. In this study, an empirical SHO gain probability distribution obtained for the 1X system is employed [11], which is given in Table II. The behavior of the random variable within the limits of each SHO gain interval in Table II is assumed to be uniform. From this distribution, the average SHO gain is found to be about 1.3 dB. For simplicity, in this study SHO gain is applied only to mobiles that would otherwise be out of coverage, which is a conservative scenario, since CDMA mobiles will sometimes go to SHO even when coverage is guaranteed by a single cell connection.
TABLE II. DISTRIBUTION FOR SHO GAIN [11].

IV. LINK BUDGET ANALYSIS Link budget analysis seeks to determine the minimum instantaneous transmit power Pmin that would be required in order to close the uplink, that is, the RF link from the Mobile Station (MS) to the Base Station (BS). The link budget equation expressed in dBm and dB is

Pmin = Rsen + L p + Li Gi ,
i i

(3)

where Rsen is the receiver sensitivity, Lp is the propagation loss defined in (2), Li is the i-th system loss factor and Gi is the i-th system gain factor. Receiver sensitivity is calculated using

Rsen = 174

E dBm + 10log( R) + NF + b , (4) Hz N o req

where R is the compressed voice information bit rate in bits per second, NF is the receiver noise figure in dB, excluding cables and connectors, and (Eb/No)req is the minimum required information bit energy to noise ratio, also in dB. The information bit rate corresponds to the uncoded digital voice stream (9.6 kbps for 1X and 12.2 kbps for UMTS). Equation (4) can be equivalently written as:

Interval of SHO Gain values 0 dB < SHO gain < 1 dB 1 dB < SHO gain < 2 dB 2 dB < SHO gain < 3 dB 3 dB < SHO gain < 4 dB 4 dB < SHO gain < 5 dB

Probability (%) 60.2 13.6 17.3 8.0 0.9

Rsen

dBm = 174 + 10log( B) + NF + SNRreq , (5) Hz

where B is the signal bandwidth in Hz and SNRreq is the required signal-to-noise ratio in dB. For convenience, (5) is often used instead of (4) for systems based in Time-Division Multiple Access (TDMA) such as the GSM system in our study, which has a nominal signal bandwidth of 200 kHz. The system loss factors Li considered in (3) for the GSM system are the other-cell co-channel interference factor and the base station cable losses. In the CDMA 1X and UMTS systems the loss factors are the cell load factor (a measure of co-channel interference, [8]), the base station cable losses, and the otherchannel power allocation factor, 10log(), where

The full list of parameters used in (3), as employed in the simulations, are given in Table III for each of the technologies under study.
TABLE III. LINK BUDGET PARAMETERS.

Parameter

GSM N/A 9.0 dB


-107.0 dBm

CDMA1X 3.0 dB N/A 5.0 dB


-126.6 dBm

UMTS 5.0 dB N/A


-123.1 dBm

(Eb/No)req SNRreq
Receiver Noise Figure Receiver Sensitivity BS Antenna Type BS Antenna Max Gain BS Cable losses Co-channel interference Uplink Load Factor

Pother Ptotal

(6)

with Pother being the power allocated by the MS transmitter to overhead code channels different from the code channel carrying the voice bit stream, and Ptotal being the total MS transmit power. The gain factors Gi considered in (3) for the GSM system are the base station antenna gain, and the BS antenna diversity gain. For the CDMA 1X and the UMTS systems the gain factors included in (3) are the base station antenna gain and the soft handover gain, which is defined as the amount of transmit power that the mobile station is able to save due to soft

Sectorized, 90 beamwidth 15.1 dBi 2 dB 2.0 dB N/A N/A 3.0 dB N/A 1.5 dB 0 dB Included in (Eb/No)req N/A 1.8 dB 3.0 dB (50% load)

10log() [eq. (6)]


MS ant. gain + cable loss BS ant. diversity gain

V. MONTE CARLO SIMULATIONS AND POWER CONTROL The methodology employed in this work establishes that each investigated cell is assumed to be hexagonal, with the cell radius as one of the controlled parameters. The spatial distribution of the cell locations to be simulated is determined by a rectangular grid that is scaled in space so that there is an approximate number of 26,000 simulation points within each cell, regardless of its radius. Simulation locations which are closer than 20 meters from the cell center are discarded, following restrictions of the W-I propagation model. Ten Monte Carlo trials are performed for each cell location, and their results are individually recorded for statistical purposes. This simulation set-up results in approximately 260,000 Monte Carlo trials uniformly spread over each tested cell. Once the minimum required MS instantaneous transmit power Pmin for each of the Monte Carlo trials is obtained using the statistical models presented in Sections III and IV, the corresponding MS transmit power Pt is determined using the power control algorithms that are inherent to each technology under study. Power control is accomplished in all three technologies by means of a closed-loop mechanism [4]. While the power control algorithm for both CDMA systems (1X and UMTS) is based on fast base station feedback of channel conditions (in the order of one-thousand commands per second), the power control in GSM, on the other hand, is relatively slow and subject to significant transient delays. In order to employ a common set of conditions for all three systems, voice users are assumed to be static or low-speed users (pedestrians), so that any transient aspects of the behavior of the respective power control algorithm are assumed to have died out, and therefore are not considered. Table IV presents the parameters associated to the power control mechanisms of the three technologies under study, as obtained from the standards and the technical data sheet of the phone models considered in the study. For pedestrian users, the effect of the closed-loop power control algorithms can be assumed to simply adjust the actual transmit power Pt of the cell phone to the power step that is immediately above the minimum required MS instantaneous transmit power Pmin. When Pmin, including SHO gain if applicable, exceeds the maximum transmit power given in Table IV, the cell phone is considered to be out of coverage, and the corresponding Monte Carlo trial is excluded from the calculation of power and SAR statistics.
TABLE IV. POWER CONTROL PARAMETERS.

In the GSM system, v is also affected by the transmission of silence descriptor frames and the Slow Associated Control Channel (SACCH). In both CDMA systems, v is affected by the overhead code channels (pilot, etc). Additionally, in GSM systems the average transmit power is further reduced by the TDMA duty cycle factor, which accounts for the fact that only one out of eight time slots is actually used by the mobile station transmitter. Thus, the TDMA duty cycle factor in GSM reduces the average transmit power by1/8, or 9.0 dB.
TABLE V. VOICE ACTIVITY FACTORS EMPLOYED IN THE STUDY.

Voice Activity Factor v Percent

GSM 70%

1X 67%

UMTS 67%

VI. RESULTS AND ANALYSIS Five cell phone models for each of the technologies under study were randomly selected. All of the telephones are designed to operate in the cellular (850 MHz) band, although some may also operate in other bands. Two parameters were extracted from the SAR test reports published by the FCC [12]: the maximum SAR for 850 MHz measurements, and the average transmit power at which such SAR value was obtained. The proportionality constant KSAR is then obtained for each model using (1). Parameters for the fifteen randomly chosen phone models are shown in Table VI. In the case of GSM phones, average power has been corrected to account for the TDMA duty cycle factor.
TABLE VI. CELL PHONE PARAMETERS.

GSM SAR
[w/kg]

Model1 0.840 23.97 3.368 Model1 1.160 23.30 5.426 Model1 1.06 23.00 5.313

Model2 0.776 23.22 3.698 Model2 0.721 24.87 2.349 Model2 0.876 22.05 5.464

Model3 0.660 22.92 3.409 Model3 1.090 25.07 3.392 Model3 1.00 23.70 4.266

Model4 1.24 22.97 6.259 Model4 1.130 25.20 3.413 Model4 1.08 23.60 4.714

Model5 0.476 20.06 5.912 Model5 0.728 25.64 1.988 Model5 0.706 22.94 3.588

Average 0.798 22.42 4.529 Average 0.996 24.82 3.314 Average 0.944 23.06 4.669

<Pt>[dBm] KSAR [kg ] 1X SAR


[w/kg]
-1

<Pt>[dBm] KSAR [kg ] UMTS SAR


[w/kg]
-1

Parameter Maximum Power Minimum Power Granularity

GSM 33 dBm 5 dBm 2 dB

1X 25 dBm -50 dBm 1 dB

UMTS 23 dBm -50 dBm 1 dB

<Pt>[dBm] KSAR [kg ]


-1

Finally, in order to obtain the statistics of the SAR using (1), the average transmit power <Pt> needs to be calculated from the instantaneous transmit power Pt available from each Monte Carlo trial. The average transmit power in all three technologies is reduced by the so-called voice activity factor v, which can be defined as the ratio of average transmit power to peak transmit power. Table V shows the assumed values for v.

Although the random sample size is relatively small, it appears as if CDMA1X phones do better in terms of KSAR than GSM or UMTS phones; that is, on the average, the transmission power in 1X phones translates into less of SAR than in other technologies (about 25% less). Nevertheless, larger sample of models should be used in order to corroborate this statement.

Monte Carlo simulations were divided into two sets. In the first set of tests, cell sizes are adjusted in order to provide the same quality of coverage (90% of cell area are is covered) for all three technologies. In the second set of trials, all cell radii are fixed (2 km. for urban cells and 5 km. for suburban cells). Two performance indicators are defined: the average SAR and the fraction of simulation points with SAR higher than 0.2 w/Kg, which is one-tenth of the limit set by the International Commission for Non-Ionizing Radiation Protection (ICNIRP) for head radiation over general population users [1]. This fraction is called F and is expressed as a percentage of all simulation points within coverage for each cell. Monte Carlo trials that result in the cell phone being out of coverage are excluded from both indicators. Tables VII and VIII display the results for the first set of simulations (cell area coverage fixed to 90%). In these tests, urban cell radii were set to 1150, 1750 and 1260 meters, and suburban cell radii were set to 2625 m, 3980 m and 2860 meters (for GSM, 1X and UMTS, respectively).
TABLE VII. RESULTS FOR FIXED 90% AREA COVERAGE, URBAN CELLS.

It can be seen from Tables VII and VIII that average SAR levels in all cases hover around 0.01 watt/kg, which is about 5% of the maximum ICNIRP limit of 2.0 watt/kg. It is also observed that for the same phone, both the average SAR and the F fraction of phones over 0.2 watt/Kg. are almost identical for both types of environment (urban, suburban), given that coverage is the same (90%). It can also be seen that, on the average, all three technologies perform similarly in terms of the two selected indicators, with a slight advantage for UMTS, which exhibits a lower average SAR and F by about 10%. Tables IX and X show the results for the second set of simulations, carried out for cell radius fixed to 2 km in urban cells and 5 km in suburban cells. In these tests, urban cell coverage was found to be 70%, 86% and 75%, and suburban cell coverage was found to be 66%, 84% and 70% (for GSM, 1X and UMTS, respectively). All of these coverage reliability values are significantly lower than the value used in the first set of simulations (90%)
TABLE IX. RESULTS FOR FIXED 2 KM CELL RADIUS, URBAN CELLS.

GSM Average SAR


[w/kg]

Model1 0.073

Model2 0.080

Model3 0.074

Model4 0.136

Model5 0.128

Average 0.098

GSM Average SAR


[w/kg]

Model1 0.131

Model2 0.143

Model3 0.132

Model4 0.243

Model5 0.229

Average 0.176

F [%] 1X Average SAR


[w/kg]

13 Model1 0.125

13 Model2 0.068

13 Model3 0.098

19 Model4 0.124

19 Model5 0.072

15.4 Average 0.098

F [%] 1X Average SAR


[w/kg]

26 Model1 0.160

26 Model2 0.081

26 Model3 `0.117

34 Model4 0.138

34 Model5 0.080

29.2 Average 0.115

F [%] UMTS Average SAR


[w/kg]

19 Model1 0.098

11 Model2 0.101

16 Model3 0.079

19 Model4 0.087

13 Model5 0.066

15.6 Average 0.086

F [%] UMTS Average SAR


[w/kg]

25 Model1 0.162

13 Model2 0.166

20 Model3 0.130

21 Model4 0.144

15 Model5 0.109

18.8 Average 0.142

F [%]

16

16

13

13

11

13.8

F [%]

29

29

24

24

20

25.2

TABLE VIII. RESULTS FOR FIXED 90% AREA COVERAGE, SUBURBAN CELLS.

TABLE X. RESULTS FOR FIXED 5 KM CELL RADIUS, SUBURBAN CELLS.

GSM Average SAR


[w/kg]

Model1 0.073

Model2 0.080

Model3 0.073

Model4 0.135

Model5 0.127

Average 0.097

GSM Average SAR


[w/kg]

Model1 0.141

Model2 0.155

Model3 0.143

Model4 0.262

Model5 0.248

Average 0.190

F [%] 1X Average SAR


[w/kg]

13 Model1 0.125

13 Model2 0.068

13 Model3 0.098

18 Model4 0.125

18 Model5 0.072

15.0 Average 0.098

F [%] 1X Average SAR


[w/kg]

28 Model1 0.176

28 Model2 0.090

28 Model3 0.130

37 Model4 0.153

37 Model5 0.089

31.6 Average 0.127

F [%] UMTS Average SAR


[w/kg]

19 Model1 0.097

11 Model2 0.099

16 Model3 0.078

19 Model4 0.087

13 Model5 0.065

15.6 Average 0.085

F [%] UMTS Average SAR


[w/kg]

28 Model1 0.174

15 Model2 0.179

22 Model3 0.140

24 Model4 0.155

17 Model5 0.118

21.1 Average 0.153

F [%]

16

16

13

13

11

13.8

F [%]

31

31

27

27

22

27.6

From Tables IX and X it can be observed that that average SAR and F values are substantially higher than those obtained for 90% cell area coverage. Actually, it can be seen that the relative increase in these two indicators is proportional to the decrease in coverage reliability. At a constant cell radius, CDMA1X is the technology with lower head exposure levels. UMTS SAR levels are about 20% higher than those of 1X, and GSM levels are about 50% higher than those of 1X. This is consistent with the coverage levels obtained for each case, with 1X being the system with higher cell area coverage, and GSM being the system with lower coverage, as mentioned before. One additional indicator that is often used to measure performance in terms of transmit power is the fraction of simulation points in which the cell phone uses its maximum power. This is not directly a measure of exposure, which is quantified by SAR rather that transmit power, but it is certainly related. Table XI shows such results for all three technologies (results for all five cell phone models in each technology have been averaged).
TABLE XI. FRACTION OF MONTE CARLO TRIALS IN WHICH CELL PHONES TRANSMIT AT MAXIMUM POWER.

levels and lower head emissions. This means that the implementation by cellular operators of system design strategies and techniques aimed at enhancing the uplink coverage, such as increased cell density, higher cell towers, the use of cell tower amplifiers, cell antenna diversity, microcells and active repeaters, tend to have a possitive impact in the average emission levels on the head of cellular voice users. All other system factors evaluated seem to have a lesser impact on the level of emissions. For instance, there are no major differences between the results for urban and suburban cells, as long as the cell coverage reliability levels are similar. When comparing technologies, UMTS shows a slight advantage in terms of average emissions at constant coverage reliability, whereas CDMA2000 1X displays the lower average SAR at constant cell radius. When comparing different cell phone models, all other factors kept constant, one can expect significant differences in average SAR levels: up to almost a 2:1 ratio within the same technology. This fact emphasizes how important it is for consumers to review the manufacturers specifications regarding SAR, which are often available only through the Internet. REFERENCES
[1] International Commission for Non-Ionizing Radiation Protection (ICNIRP), Guidelines for Limiting Exposure to Time-Varying Electric, Magnetic, and Electromagnetic Fields (up to 300 Ghz) Health Phys., vol. 74, no. 4, 1998. [2] American National Standards Institute (ANSI), "Safety Levels with Respect to Human Exposure to Radio Frequency Electromagnetic Fields, 3 kHz to 300 GHz," ANSI/IEEE C95.1-1992 (previously issued as IEEE C95.1-1991). Copyright 1992 by the Institute of Electrical and Electronics Engineers, Inc. (IEEE), New York. [3] Federal Communications Commission (FCC), Office of Engineering and Technology, Evaluating Compliance with FCC Guidelines for Human Exposure to Radiofrequency Electromagnetic Fields. OET Bulletin 65, 1st Edition, 1997. [4] Rappaport, T. Wireless Communications. Principles and Practice. 2/E. Prentice. 2001. [5] ETSI Rec TR 125 996. UMTS Spatial Channel Model for MIMO simulations. Versin 9. 2010. [6] H. Okamoto, K. Kitao, and S. Ichitsubo. Outdoor-to-Indoor Propagation Loss Prediction in 800-MHz to 8-GHz Band for an Urban Area. IEEE Trans. Veh. Techn., Vol. 58, No. 3, March 2009, pp 105967. [7] Kostanic, I. Hall, C. McCarthy, J. Measurements of the vehicle penetration loss characteristics at 800 MHz, 48th IEEE Vehicular Technology Conference, May 1998, pp. 1-4. [8] [12]. Holma H. and Toskala A., W-CDMA for UMTS: Radio Access for Third Generation Mobile Communications. John Willey and Sons, 1/E, 2000. [9] Amphenol Antel. Product Catalogs. Antenna Catalog. www.antelinc.com. Consulted on 10/10/2007. [10] J. S. Lee, L. E. Miller. CDMA Systems Engineering Handbook, Artech House, 1E, 1998. [11] Wang C. C., Huang J. F., Propagation Path Loss Characterization for an 870 MHz Cellular CDMA System in Taipei City. Int. Symp. Microwave Antenna, Propagation and EMC Technologies for Wireless Communications, 2007. [12] Federal Communications Commission (FCC), Office of Engineering and Technology, Equipment Authorization. ://fjallfoss.fcc.gov/oetcf/eas/reports/. Consulted on 23/02/2008.

Fixed 90% coverage Urban GSM 1X UMTS 4% 3% 3% Suburban 4% 3% 3%

Fixed cell radius Urban (2 km) 8% 4% 6% Suburban (5 km) 9% 4% 7%

From Table XI, it is clear that cells with the same area coverage exhibit approximately the same fraction of mobiles at maximum power, regardless of the type of environment (urban or suburban). At fixed radius, however, 1X cell phones tend to transmit at maximum power less frequently than their UMTS and GSM counterparts, which is consistent with the results presented previously. VII. CONCLUSIONS In this work, the problem of quantifying the level of head emissions from cellular phones operating under the three major technologies, as a function of system parameters, has ben analyzed. In order to achieve this goal, a novel methodology that is based on linearly relating transmit power to SAR, and employs experimental test results reported for a number of commercial devices, has been designed and implemented. It can be concluded from the results found that the SAR levels of commercial phones, as they are averaged throughout all possible locations in typical urban and suburban cells, depend strongly on the cell area coverage reliability provided by the system. As the percentage of cell area coverage decreases, that is, as the reverse link budget gets tighter, phones in increasingly more cell locations are driven by the power control mechanisms to higher transmit power levels, thus increasing their SAR levels. In other words, improving the reverse link budget design margins translates not only into higher coverage reliability but also into lower transmit power

You might also like