You are on page 1of 12

Peer-to-Peer Netw. Appl.

DOI 10.1007/s12083-014-0308-9

A channel-hopping scheme for continuous rendezvous and data


delivery in cognitive radio network
Yen-Wen Chen & Po-Yin Liao & Ying-Cheng Wang

Received: 17 April 2014 / Accepted: 19 August 2014


# Springer Science+Business Media New York 2014

Abstract The purpose of cognitive radio (CR) technology is Keywords Cognitive radio . Rendezvous . Dynamic
to improve spectrum utilization. In recent years, dynamic spectrum access . Peer to peer
spectrum access (DSA) has become one of critical researches
for achieving improved transmission performance in CR. A
rendezvous algorithm facilitates a connection between a pair 1 Introduction
of secondary users (SUs) to achieve peer to peer transmission
in CR environment. Most researchers have adopted the ren- The recent lack of radio resources has primarily been caused
dezvous degree and the time to rendezvous (TTR) as key by the rapid growth of mobile services. Generally, unlicensed
factors to design rendezvous algorithms. However, the occur- radio bands can be freely utilized; however, transmission
rence of a primary user (PU) can shield off the rendezvous quality cannot be guaranteed because of poor management.
even when both SUs occupy the same channel. A complete In licensed bands, spectra are divided into segments and
peer to peer transmission process in cognitive radio network deployed using various wireless technologies to achieve ap-
(CRN) provides not only rendezvous for SUs but also channel propriate management [1, 2]. Although licensed band users
access for data transmission under priority usage of PU. This can exclusively utilize the allocated radio resources, it intro-
paper proposes an algorithm to achieve shift-based channel duces low utilization of the spectrum. According to the statis-
hopping for continuous rendezvous (SCHCR) to guarantee tics of the FCC, spectrum usage is between 15 % and 85 %
that each SU can rendezvous continuously with its receiver from the time and space domain [3, 4]. The unutilized spec-
within the upper bound of the expected TTR without a cen- trum is regarded as spectrum hole. Consequently, cognitive
tralized control mechanism. This continuity property facili- radio (CR) technology is proposed to enhance the spectrum
tates rendezvous, particularly when PUs exist. Based on the utilization of licensed spectra. The CR end device has the
rendezvous algorithm, this paper further proposes a threshold capability of sensing spectrum holes and it can use them under
controlled skip (TCS) scheme and two policies for the occur- the constraint of not affecting licensed users; however, CR end
rence of PU to resolve channel contention and data transmis- devices are autonomously operated with loose management.
sion problems. Simulation results indicated that the proposed Contention and collision among end devices injure valuable
schemes achieve shorter rendezvous time, particularly when radio resources. In recent years, dynamic spectrum access
the load of PU is high, and facilitate superior transmission (DSA) has become a critical problem in achieving superior
performance. transmission performance in CR networks (CRNs). In CRNs,
two types of user exist: primary users (PUs) and secondary
users (SUs). Traditionally, if a spectrum has already been
Y.<W. Chen (*) : P.<Y. Liao : Y.<C. Wang
deployed for licensed users, license-exempt users are unable
Department of Communication Engineering, National Central
University, Jhongli, Taiwan, Republic of China to use the spectrum even if the PU is inactive. However, in
e-mail: ywchen@ce.ncu.edu.tw CRNs, an SU can temporarily use an unused spectrum to
P.<Y. Liao transmit data by scanning for and detecting the absence of
e-mail: dreamilylife@gmail.com PU in that spectrum channel. One of the research directions of
Y.<C. Wang CRNs is facilitating peer-to-peer communication through ef-
e-mail: eric3724135@gmail.com fective use of spectrum holes. Theoretically, an n-channel
Peer-to-Peer Netw. Appl.

CRN can provide up to n pairs of simultaneous peer-to-peer environments. Section 3 presents the system architecture and
communication if no PU is active in the spectrum; however, proposes the SCHCR algorithm for consecutive rendezvous.
both sender and receiver SUs can communicate with each The TCS-based contention mechanism and two policies are
other only when they reside in the same channel and no also provided to accomplish the data transmission of the SU
interference is caused by PUs or other SUs. Therefore, the rendezvous pair. The experimental results are illustrated in
rendezvous algorithm is necessary for a sender and receiver to Section 4 and discussions are provided. Finally, Section 5
meet at the same channel before transmitting data. The per- offers a conclusion.
formance of the rendezvous algorithm is typically evaluated
based on its time to rendezvous (TTR) and the number of
possible rendezvous channels, which is also regarded as the 2 Related works
channel overlapping degree. The algorithm should minimize
TTR and provide a guaranteed rendezvous time. Ideally, the Generally, rendezvous approaches can be classified into four
rendezvous of any pair of secondary users is spread across all types: dedicated-control, single-control, multiple-control, and
available channels. no-control channels [5]. The dedicated control channel be-
Rendezvous approaches are categorized into two branches: longs to the aided rendezvous system, and requires a central-
aided and unaided rendezvous systems [5]. In aided rendez- ized control mechanism to achieve rendezvous. As men-
vous system, the SU rendezvous procedure is managed by a tioned, aided systems suffer certain drawbacks. For unaided
centralized controller, such as a fusion center (FC) or base rendezvous schemes, the adaptive multiple rendezvous con-
station (BS). In the unaided rendezvous approach, the SU can trol channel (AMRCC) for CR ad hoc networks was proposed
either use a common control channel (CCC) to exchange [6]. The authors designed an adaptive hopping sequence to
information to achieve rendezvous [68] or perform its own avoid PU interference. Their hopping sequence was based on
process to achieve blind rendezvous without the assistance of a ranking table, which was predicted using the periodical
CCC. The CCC-based approach is conceptually similar to the sensing experience of SUs. When a pair of SUs, a sender
aided system except that it can arrange multiple control chan- SU and a receiver SU, achieve rendezvous, they exchange
nels to avoid the single failure point problem. However, the information, including a SYNC packet and a ranking table to
problem of control channel contention and signaling overhead synchronize in the following hops. The sender SU and receiv-
should be carefully considered. The blind rendezvous ap- er SU can compute a common hopping sequence through the
proach does not require a control channel to accomplish exchanged information. Although the AMRCC provides a
rendezvous; thus, it has no overhead in the rendezvous proce- low-overhead rendezvous procedure, its performance primar-
dure. Heuristic mechanisms are necessary to achieve the ren- ily depends on the historical statistics of PU; furthermore, it
dezvous of an SU pair; however, guaranteeing a rendezvous ignores the collision issue in its analysis, which can be im-
within a limited time is difficult. Because PUs have higher practical, particularly when the behavior of PU is irregular. In
priority to use a channel than SUs do, the appearance of a PU [7], the authors proposed a distributed coordination protocol
blocks the rendezvous of an SU pair even when both SUs are for establishing a common control channel (DCP-CCC) to
in the same channel. Furthermore, a PU can interrupt ad hoc construct a common channel cluster tree for the management
communication even when both SUs have already achieved a of clusters in large scale network topology. However, in [8],
rendezvous in the same channel and started data transmission. the cluster size using the same CCC was not limited; therefore,
Thus a complete CR-based ad hoc communication mecha- the overall system might require an additional common con-
nism should not only provide a rendezvous scheme but also a trol signal overhead when the loading of the primary system is
channel access scheme for data transmission. The appearance heavy. In the other unaided rendezvous scheme, the blind
of PU is also an unavoidable problem for both schemes. This rendezvous scheme, no specific control channel is required,
paper proposes a novel algorithm to achieve shift-based chan- no control overhead is incurred, and no Achilles heel exists to
nel hopping for establishing continuous rendezvous (SCHCR) jeopardize the operation of the rendezvous procedure. How-
for a sender SU to rendezvous with a receiver SU. The ever, a sender SU requires an implicit method to locate the
proposed SCHCR algorithm facilitates SU rendezvous in all receiver SU. The concept of channel hopping is widely
channels in consecutive time slots, which greatly minimizes adopted in blind rendezvous algorithms and the design of a
the variance of TTR except when all channels are occupied by hopping sequence is a critical factor that affects TTR and the
PUs, compared with existing schemes. In addition, this paper overlapping degree.
also proposes a threshold controlled skip (TCS) contention In [9], the authors proposed a cooperation framework for
algorithm for achieving effective transmission after asynchronous channel hopping, however, it might introduce a
rendezvous. high cooperative signaling overhead when cooperative neigh-
The remainder of this paper is organized as follows. Sec- bors increase. In [10], the authors proposed an array-based
tion 2 introduces related studies of rendezvous in CR quorum system and analyzed its property with random
Peer-to-Peer Netw. Appl.

rendezvous time. The probability of successfully achieving 3 The system architecture and proposed
blind rendezvous is low. In [11], the authors expressed that target-sequence-based rendezvous scheme
specific rules are critical to guarantee the rendezvous of a pair
of SUs within a time limit for practical applications. In [12], The proposed peer-to-peer CRN framework [18] consists of
the authors proposed a general message passing based frame- two procedures: the SCHCR rendezvous procedure and the
work, which maintains and exchanges rendezvous informa- the TCS-based data transmission procedure of SU after a
tion among SUs, to accelerate rendezvous process. Further- rendezvous. A brief overview of the system architecture is
more, the authors proposed an efficient greedy channel presented in Fig. 2. PUs and SUs coexist in a licensed band
switching algorithm for efficient rendezvous of both single- wireless network. A PU can utilize the channel if it must send
hop and multi-hop networks. The role-based parallel sequence data to the base station, and an SU can only seek a rendezvous
(RPS) assumed that SUs can be equipped with more than one with its receiver by hopping for spectrum holes among chan-
wireless transceiver to improve the rendezvous perfor- nels. SUs transmit data until a PU appears. Each channel can
mance by using multiple radios [13]. In [14], the au- be utilized by only one PU at a time.
thors proposed a novel short-sequence-based (SSB) ap- The objective of the proposed SCHCR scheme is to provide a
proach and presented an analysis of the benefits of short channel-hopping-sequence-based rendezvous mechanism in a
periodic channel hopping sequences in the rendezvous distributed environment without a fusion center or common
process. The sequence in SSB is based on discrete segment control channel. The proposed scheme assumes that the sender
which is arranged from the set of all available channels. The SU can obtain the channel-hopping sequence of the receiver SU
SUs start the rendezvous process from the one extreme to the through its hashing function on the unique ID, such as the IP
other extreme. Once reaching the end of the segment, SUs hop address or MAC address. This assumption is reasonable because
to the opposite path. the sender SU requires the IP or MAC address of the receiver to
The sequence-based rendezvous scheme was proposed to transmit data to a specific receiver. Figure 3 presents the SU
guarantee rendezvous within a given time limit [15]. In their operation of the proposed framework. Normally, each SU has its
approach, the SU seeks its receiver through channel hopping. own channel-hopping sequence Peri, and the sender SU switches
The order of the channel-hopping sequence is known to the its hopping sequence to Seqr to achieve rendezvous with the
SU in advance. To achieve rendezvous in an asynchronous receiver for data transmission. Both sender and receiver have the
environment, the SU inserts a permutation channel into the same hopping sequence for data transmission after rendezvous.
channel sequence of each hopping round. The example pre- The proposed scheme focuses on the rendezvous and data trans-
sented in Fig. 1 indicates that a CRN comprises four channels mission problems of pairs of SUs caused by PU interference and
and the permutation and channel order are {2, 1, 3, 4}. Thus, channel contention. The SCHCR rendezvous procedure and the
the arrangement of the original sequence is {2, 1, 3, 4, 2, 1, 3, TCS-based data transmission scheme are described as follows.
4, , 2, 1, 3, 4}. After the insertion of the permutation
channel, the hopping sequence of the sender SU becomes 3.1 Shift-based channel hopping for continuous rendezvous
{2, 2, 1, 3, 4, 1, 2, 1, 3, 4, 3, 2, 1, 3, 4, 4, 2, 1, 3, 4}. Although scheme
the sequence-based approach proposed in [15] can
achieve rendezvous, it only considers rendezvous for To provide asynchronous rendezvous within a limited time,
two SUs; the contention of multiple SUs and interfer- the shift sequence concept was adopted for channel hopping.
ence from PUs are not considered. These two problems
(i.e., channel contention and the appearance of PU) play
critical roles in transmission performance. These two
problems were considered in the present study and the
proposed scheme provides a complete peer-to-peer transmis-
sion process, which includes the rendezvous state and the data
transmission state, in a CRN.

Fig. 1 sequence-based rendezvous approach [15] Fig. 2 System Architecture


Peer-to-Peer Netw. Appl.

is {3, 2, 5, 1, 4} as the original permutation to generate the


hopping sequence of the sender, Seqs, as indicated by (2).

Seqs Per0r ; Per1r ; ; Perkr ; PerrN 1 ; k0; N 1
2

where Per0r is the original permutation of the receiver SU r,


which is {3, 2, 5, 1, 4} in the aforementioned example, and
Perkr denotes the k-times left-rotated sequence of the original
permutation sequence Per0r . Because Seqs consists of the
original permutation sequence and the N1 left-rotated se-
quence, the length of Seqs is N2. The detailed sequence gen-
eration of Seqs is illustrated in Algorithm 1 and an example of
the channel rotation with the aforementioned original permu-
tation is provided in Fig. 4. By using the generated sequence
Fig. 3 Operation States of SU
in Algorithm 1, the basic SCHCR rendezvous procedure is
illustrated in Algorithm 2.
Algorithm 1: Sequence Generation
Each SU was assumed to have a permutation sequence of all
01: sender create a new sequence by re-
available channels. For example, if channels available to the
ceivers permutation
SU are {1, 2, 3, 4, 5}, the permutation of the SU might be {1,
02: define shift(i,permutation) : left
2, 3, 4, 5}, {1, 4, 5, 2, 3}, or any other permutation sequence
shift the permutation by i
of the available channels. A CRN was assumed to comprise N
03: Let N is total channel numbers, and Seq-
channels, and all SUs were assumed to have access to these N
new is the new sequence of sender
channels for peer-to-peer communication. Let Peri be the
04: for i=0 to N-1
permutation sequence generated by the unique characteristic
05: Seq s add permutation of shift(i,
of SU i. Because the permutation sequence has no duplicate
permutation)
channel, the length of each permutation sequence is equal to
06: end for
N, as indicated by (1).
Algorithm 2: SCHCR procedure
Peri fch0 ; ch1 ; ; chi ; ; chN1 g 1 01: When sender needs to rendezvous to
receiver
The channel-hopping sequence is obtained by permuting 02: Sender create a new sequence through
the channel numbers through hashing of the IP address or Sequence Generation in Algorithm 1
MAC address of the SU. Although two SUs can have the same 04: Let N is the length of Seqs, and sender
channel-hopping sequence after hashing, it does not affect the start to channel hopping with Seqs
proposed scheme. Normally, each SU hops its own channel 05: for i=0 to N-1
sequence cyclically. When an SU sends data to another SU, 06: if Seqs[i] is the same as receivers
the sender SU changes the channel-hopping sequence pro- channel in the same time slot
posed by SCHCR to rendezvous with its receiver within 07: break for loop and rendezvous success
limited time. The proposed hopping sequence of the sender 08: end for
SU used to seek the receiver SU r is defined as Seqs. Because
the sender must know either the IP address or the MAC
address of the receiver before data transmission in a peer-to-
peer environment, the sender obtains the hopping sequence of
the receiver. The sender SU cascades N rounds of the re-
ceivers hopping sequence, however, with sequence rotation
for each round. For example, if there are five available chan-
nels, identified as 1, 2, 3, 4, and 5, in the CRN, and the
channel-hopping sequences of the sender SU and the receiver
SU are {3, 1, 2, 4, 5} and {3, 2, 5, 1, 4}, respectively, after
hashing. When the sender attempts to rendezvous with the
receiver, the sender adopts the receivers permutation, which Fig. 4 The sequence generation mechanism of sender
Peer-to-Peer Netw. Appl.

Fig. 5 The rendezvous examples


of the SCHCR with delay 1 and
delay 2 cases

The SCHCR approach also addresses the asynchronous hopping sequence generated by the proposed SCHCR scheme
problem and enables continuous rendezvous between the can guarantee the rendezvous happened within N2 time slots
sender and receiver. This characteristic can increase the prob- in consecutive N different channels.
ability of rendezvous and shorten the rendezvous time, partic- Based on the regular rule of all delay schemes, the average
ularly when PU interference is considered because the pro- time to rendezvous is obtained as
posed scheme allows both the sender and receiver to achieve .
  XN 1
rendezvous in consecutive time slots in distinct channels; E TTRavg D0
D  N 1 D 1
therefore, the rendezvous probability increases. Because the
f0  N 1  N N 1  N N g
sender can switch to Seqs to seek the receiver at any time, a . N 2 N
delay can occur between Seqs and Perr. However, the pro- D 1 1
2
posed Seqs can rendezvous with cyclic Perr consecutively. As
3
indicated by the examples presented in Fig. 5, if the delays of
the sender and receiver are 1 and 2, respectively, and the upper The average time to rendezvous based on various numbers
sequences are Seqs, both sequences can achieve consecutive of channels can be estimated using (3). Another condition,
rendezvous in each channel. degree, defined as TTRdegree which refers to the TTR which
In addition, the SCHCR scheme can guarantee the upper the sender SU and receiver SU can rendezvous consecutively
bound of the TTR under the asynchronous caused by various at how many different available channels. Let K denote the
delays and degrees. Let TTRdegree
delay be the time to rendezvous. degree of rendezvous channel and K {1, 2, 3, , N}; time to
Regarding the condition of delays, TTRdelay depends on the rendezvous with degree K can thereby be derived, which
asynchronous condition. Let D be the delay between the Per0r means that the sender and the receiver achieve rendezvous
of Seqs and the Perr of receiver r when the sender starts the through distinct K channels. For example, if K=3, the sender
rendezvous procedure, where D {0, 1, 2, , N-1}. For and receiver must achieve rendezvous through three distinct
example, if there are N channels and D=0, the TTR0 =0 channels in consecutive time slots. Thus, the average time to
N+1, thus the sender and receiver can rendezvous at the first rendezvous with degree K can be derived as follows:
time slot in the current period without any additional time slots
by asynchronous delay. If D=1 and TTR1 =1N+1, the send-    
E TTRdegree E TTRavg K1
er and receiver must use N time slots to achieve rendezvous.  2 
N N N 2 N
Similarly, if D =N1, the TTR is TTRN1 =(N1)N+1. The 1 K1 K
2 2
SCHCR scheme guarantees sender and receiver to rendezvous
to each other within N2 time slots. The maximum TTR 4
(MTTR) occurs when D is N1, consequently, MTTR =
(N1)N+1. Thus both SUs will start to rendezvous from The characteristic of the proposed approach can be ob-
the (N2 (N1)) -th slot to the N2 -th slot continuously, if there served in (4). In the proposed approach, the average TTR to
is no interference from PU. Lemma 1 provides the proof of achieve the maximum degree is the average TTR in addition
guaranteed rendezvous in the following. to all available channels (i.e., E[TTRdegree]=E[TTRavg]+K).
Lemma 1: The sender hopping sequence generated by This advantage is ascribed to continuous rendezvous for a
SCHCR guarantees to rendezvous with the receiver hopping
sequence within N2 time slots by degree N.
Proof: The sender hopping sequence generated for the
receiver r by SCHCR is the sequence of all possible rotated
sequences of the hopping sequence of receiver r, i.e. Perkr for
0t(N1), to form N2 channel hopping sequence as indi-
cated in Eq. (2). Given the permutation of SUi, Peri with time
shift duration t, 0t(N1), will be the same as the permuta-
tion Pertr. Thus Peqtr Seqs for all t. The total length of Seqs
is N2 and the length of each Perkr equals to N. Hence the sender Fig. 6 Time slot architecture of SU
Peer-to-Peer Netw. Appl.

Fig. 7 SU options for the appearance of PU (a) option 1: wait scheme, (b) option 2: jump scheme

specific period. The benefit of continuous rendezvous is when 3.2 Threshold-controlled-skip-based channel access and data
the system is in heavy loading; therefore, the proposed SCHC transmission scheme
R scheme allows SUs to achieve rendezvous in continuous N
time slots with degree K. According to this characteristic, it PUs can use channels exclusively whenever they must trans-
was assumed that (N-1) PUs occupy the (N-1) time slots which mit data. Unlike PUs, SUs must contend for access to a
comprise the SUs rendezvous time slots. In other channel for either rendezvous or data transmission. To appro-
words, SUs only have a time slot to rendezvous and priately utilize channels, the threshold concept is proposed to
the condition of rendezvous we define is the worst case differentiate the usage of the channel. PUs have the highest
when system is in heavy loading. Thus, an SU fails to priority to use a channel and an SU that has achieved rendez-
achieve rendezvous only when all channels are occupied vous with a receiver can exclusively transmit data in that
in N consecutive time slots. Let be the number of channels channel if a PU is absent. An SU that is seeking rendezvous
being occupied by PUs, where {0, 1, 2, , N1 }; the contends for a channel, and a sender SU can only verify
average TTR with PU interference can then be derived. whether a receiver SU occupies the channel after it gains
Let TTR(PUi(j)) be the average TTR with PU interfer- access to the channel. The proposed TCS channel-access
ence. The following equation can be obtained by referring scheme separates the time slot of each channel into the PU
to (3): sensing period, the continuity sensing period, and the conten-
tion or data transmission period based on the PU sensing and
  
TTRPU i j PU i j  E TT Ravg continuity sensing thresholds, as presented in Fig. 6. If a PU
  must transmit data, it directly transmits data at the beginning
1PU i j  E TT Ravg 5 of the channel. SUs can only sense whether a PU occupies a
channel in the PU sensing period. If it is sensed that a PU is
where PUi(j) is a Bernoulli random variable that refers to absent, the SU, which has already achieved rendezvous with
whether PU occupies in channel i at time slot j. The value of the receiver in the channel, can then directly transmit data to
PUi(j) will be 1 if PU occupies. Otherwise, the value will be 0. the receiver in this channel. An SU that is seeking a receiver
For example, if there are total 5 channels SUs rendezvous and can contend for access to this channel only when it senses no
PUs occupy in previous four channels, the value of is 4, the transmission carriers in the PU sensing and continuity sensing
average TTR with interferences is TTR(PUi(j))= E[TTRavg]+ periods. In this study, it was assumed that each time slot was
4. If there is no PU activity, the TTR can be estimated without 10 ms in length and the fast sensing technique of IEEE 802.22
PU interference, denoted as TTR(PUi(j))= E[TTRavg]. There- [16], which requires a 1 ms sensing time for PU sensing and
fore, the upper bound of the average TTR in a high-loading continuity sensing, was applied. When the sender SU achieves
system can be estimated using (5). rendezvous with its receiver in channel i, it starts to transmit
data in channel i in the next time slot if there is no PU

Table 1 The simulation schemes

Schemes SCHCR channel TCS with Policy when


hopping continuity SU PU interrupts
mechanism data transmission

CPJ Y Y Jump
CPW Y Y Wait
NC Y N Jump
CPJ_random N (random hopping) Y Jump
CPW_random N (random hopping) Y Wait
NC_random N (random hopping) N Jump
Fig. 8 Efficiency of rendezvous - the average time of SU rendezvous
Peer-to-Peer Netw. Appl.

Table 2 The simulation parameters channel) transmits data in the same channel in current time
SU sender and receiver pairs Case 1: 10~60 slot without contending with other SUs by issuing its role as
Case 2: 40 the continued SU (C_SU) in the continuity sensing period.
SU avg. data inter-arrival time 50 time slots (exp.)
Algorithm 3 presents the proposed TCS-based channel access
SU avg. data length 10 time slots / data
scheme.
PU avg. data inter-arrival time Case 1: 30 time slots (exp.)
Algorithm 3: Threshold Controlled Skip (TCS)
Mechanism
Case 2: 10~60 time slots (exp.)
01: set threshold = Pinitial
PU avg data length 10 time slots / data
02: if SU is a C_SU
03: SU can ready to transmit data
interference. It is clear that no collision occurs in the first two 04: else
periods. If either a PU or an SU claims to transmit data in the 05: SU random a number x between 0 and 1
channel, it transmits data for the rest of period in that channel 06: while x<threshold
and there is no contention period. For the contention period, to 07: if there is no collision detect
prevent collisions, the opportunity access request approach is 08: SU can ready to transmit data
adopted for SUs in a rendezvous state. Assume that the 09: else
contention period of a channel is divided into minislots. Each 10: if current frame is finish
SU in a rendezvous state contends for each minislot in a 11: go to next sequence in next time slot
probabilistic manner. The probability of contending for ren- 12: else
dezvous increases when the SU fails to contend for the 13: setting new threshold according to
minislot. Let Pt denote the probability of the tth contention Eq. (6)
period of the SU to issue the contention; 14: SU random a number x between 0 and 1
 15: go to next sequence in next time slot
Pt Min Pt ; Pmax 6 The data transmission of a sender SU is interrupted if a PU
appears in the transmission channel. In this case, two options
where P*t = Pt1 +Pstep, where Pstep refers to the increasing are discussed in this paper. In the first option, as presented in
value after each failed contention. Let P0 = Pinitial, where Fig. 7(a), both sender and receiver SUs continue waiting in the
Pinitial is the initial contention threshold for the first contention current channel until the PU leaves and then resume the
period. The maximum contention threshold Pmax restricts transmission. The sender SU can sense the existence of PU
within a boundary of succeeded contention probability to during the PU sensing period. Alternatively, upon the appear-
prevent all contended SUs fail to access this channel. After ance of PU, the sender and receiver SUs leave the current
achieving rendezvous, an SU enters the data transmission state channel and start to hop channels according to the Perr se-
to transmit data. In this state, the sender continues transmitting quence. The predominance of accessing channels in the con-
data to a receiver in the rendezvous channel until the occur- tinuity sensing period is not applicable in this scenario. Thus,
rence of PU interference. Thus, the SU that has transmitted SUs must contend for access to a channel to achieve rendez-
data in previous time slot in channel i (i.e. the rendezvous vous with other SUs, as indicate in Fig. 7(b).

Fig. 9 System throughput v.s.


SU numbers
Peer-to-Peer Netw. Appl.

Fig. 10 Channel utilization v.s.


SU numbers

4 Experimental simulations off modes. The ratio of the PUs inter-arrival time (off dura-
tion) to service time (on duration) was varied to adjust the PU
To investigate the performance of the proposed scheme, the load to between 20 % and 80 % in the simulations. It indicates
efficiency of the proposed rendezvous algorithm and the that the simulation results of the proposed SCHCR scheme
threshold-based transmission schemes were analyzed through were considerably close to the ideal numerical results obtained
exhaustive simulations. The TV band was adopted as the radio in Eq. (3). It also clearly indicates that, regardless the change
environment during the simulations. The 64-QAM modulation of PU load, the proposed scheme demonstrates more stable
with an inner coding rate of 1/2, which provides 14.24 Mbps average TTR than that of the SSB and SB schemes proposed
data transfer rate, was adopted for transmission in the channel in [14] and [15]. This is mainly due to the characteristic of
[17] and the number of channels in the CRN was assumed to be continuous rendezvous with degree K in the SCHCR scheme.
10. The simulation results were examined from two perspec- It is also noted that the proposed SCHCR scheme achieves
tives: the efficiency of TTR and data transmission. much lower average TTR when the PU load is getting higher.

4.1 Efficiency of time to rendezvous 4.2 Performance of data transmission in cognitive radio
network
The simulation result of the average rendezvous time, which
compares with short-sequence-based (SSB) approach [14] and Several performance indices of complete ad hoc transmission
sequence-based (SB) approach [15], versus the PU load is in CRN were investigated through exhaustive simulations.
presented in Fig. 8. The rendezvous procedure was also included in the simula-
Figure 8 presents the average rendezvous time for various tion. Because the proposed framework provides an SCHCR
PU loads. The PU appeared in each channel in alternating on/ scheme, TCS-based continuity sensing for data transmission,

Fig. 11 Time to rendezvous v.s.


SU numbers
Peer-to-Peer Netw. Appl.

Table 3 Comparisons of total


transmission time in time slots No. of SUs 10 20 30 40 50 60
CPW scheme 140.848 142.798 142.974 142.451 142.491 143.092
NC scheme 2235.36 3623.23 4969.58 6493.94 8128.44 10028.6

and various policies regarding PU interference of SU data However, the SCHCR-based schemes achieved higher sys-
transmission. Accordingly, the interference of PU and the tem throughput when the number of SU pairs exceeded 30.
contention of SU when accessing the same channel were Thus, the proposed SCHCR approach can effectively reduce
considered in the following simulations. In addition, the effi- the TTR for large numbers of SUs and thereby improve
ciency facilitated by each characteristic was examined and the system throughput. The CPW and CPW_random approaches
various options of the schemes, as listed in Table 1, were demonstrated lower average TTRs compared with the CPI and
compared in the simulations. CPJ_random approaches because, when a PU appears, an SU
The number of channels in CRN was assumed to be 10 and in a data transmission state remains in the channel to wait for
the length of time slot was assumed to be 10 ms. The the PU to leave, and therefore SUs do not contend for channel
contended mechanism parameters of the TCS scheme were access, thus alleviating the collision condition. Figure 10
as follows: Pinitial = 0.5, Pstep = 0.1, and Pmax = 0.9. The other compares the channel utilization of the proposed CPW- and
parameters are listed in Table 2. Five simulations were con- CPJ-related schemes and the NC-related schemes. In this case,
ducted to examine the effect of various parameters on the average data length and average data interarrival time of
performance. PU were 10 and 30 time slots, respectively. Thus a PU
exclusively utilized 33 % of the total bandwidth. The results
(1) Case 1 indicated that the utilization of NC-related schemes was below
10 % and the proposed CPW- and CPJ-related schemes can
Figure 9 compares the system throughput of the six ap- share approximately half of the total utilization. The CPW
proaches by varying the number of PU pairs. The figure shows scheme demonstrated the highest utilization among those
that the NC and NC_random schemes exhibited the lowest schemes because the use of continuity channel access with
system throughput among all schemes because they did not the option of waiting in a same channel, effectively preventing
adopt the continued mechanism for SU. Consequently, the the channel contention overhead. The channel utilization also
SUs had to contend for channel resources in each time slot saturates approximately 65%, primarily because of the over-
which is wasted for collision and, therefore, the SU channel head of rendezvous.
utilization as considerably low, as indicated in Fig. 10. For the The proposed CPJ- and CPW-related schemes demonstrat-
other four schemes, it indicates that the random-channel- ed substantially lower TTRs and higher system throughput
selection schemes (i.e., CPJ_random and CPW_random) ob- than the NC-related schemes did. The CPW and NC schemes
tained a slightly higher throughput than the SCHCR-based were used as representatives to examine the total transmission
schemes did when the number of SU pairs was less than 30; times, as listed in Table 3. The total transmission time was
the primary reason is that it is easy to achieve rendezvous with measured from the sender SUs search for a receiver to the
a receiver when the number of SU pairs is small, as indicated completion of data transmission. Because the proposed SCHC
in Fig. 11. R scheme decreases the TTR and the TCS-based scheme gives

Fig. 12 System throughput v.s.


mean data inter arrival time of PU
Peer-to-Peer Netw. Appl.

Fig. 13 Time to rendezvous v.s.


mean data inter arrival time of PU

the exclusively utilizes the channel in the absence of PU 10 time slots, on average, and the mean data length was also
interference, the CPW schemes achieved a much shorter total 10 time slots; hence, PUs generated consecutive data in each
transmission time than the NC scheme did. The total trans- channel. Therefore, it was considerably difficult for SUs to
mission of the CPW scheme was maintained within the range locate spectrum holes for achieving rendezvous and transmit-
of 140144 time slots regardless of the number of SU pairs; ting data. The system throughput of SUs increased when the
the primary reason is that the CPW scheme assigned a higher mean data interarrival time increased. The NC schemes ob-
channel access priority to SUs in a data transmission state tained the lowest system throughput among the six schemes.
without incurring contention, except with the interference of This phenomenon is similar to that observed in Case 1 and the
PUs. TTR was longer than that in the proposed CPW and CPJ
schemes, presented in Fig. 13. The proposed CPW scheme
(2) Case 2 exhibited the highest system throughput. In addition to
the advantages of SCHCR and continuous channel ac-
The appearance of a PU affected the TTR and the trans- cess, waiting in a channel for PUs to leave reduced SU
mission performance of SUs. The purpose of the simulation in contention compared with the CPJ scheme. The spec-
this case was to examine whether the proposed scheme can trum utilization of the CPW and CPJ schemes was
dynamically adjust resource utilization according to the PU further investigated, as indicated by Fig. 14. The results
load. The data interarrival time of a PU was varied to examine revealed the PU and SU utilizations of both schemes.
the performance effects of the six schemes. Figure 12 com- The PU utilization decreased as the mean data interarrival time
pares the system performance among the schemes. The figure increased and the utilization of the proposed schemes
illustrates that the system throughput of an SU was consider- increased accordingly. Thus, in the proposed schemes,
ably small when the mean data interarrival time was 10 time SUs can cooperate with PUs to facilitate peer-to-peer
slots. Because, under this condition, PUs generated data every transmission.

Fig. 14 Spectrum utilizations v.s.


mean data inter arrival time
Peer-to-Peer Netw. Appl.

5 Conclusion 6. Cormio C, Chowdhury KR (2010) Common control channel design


for cognitive radio wireless ad hoc networks using adaptive frequen-
cy hopping. Ad Hoc Netw 8:430438
In this paper, the complete peer-to-peer transmission process of 7. Kim MR, Yoo SJ (2009) Distributed coordination protocol for com-
a CRN was discussed. A novel shift-based channel-hopping mon control channel selection in multichannel ad-hoc cognitive radio
scheme for achieving continuous rendezvous of SUs was pro- networks. Wireless and Mobile Computing, Networking and
Communications (WIMOB), pp. 227232
posed and the threshold concept was applied to facilitate con-
8. Cormio C, Chowdhury KR (2010) An adaptive multiple rendezvous
tinuous data transmission when PUs were absent. In addition to control channel for Cognitive Radio wireless ad hoc networks.
maximizing the rendezvous degree, the proposed SCHCR Pervasive Computing and Communications Workshops (PERCOM
scheme enables consecutive rendezvous time slots in various Workshops), pp. 346351
9. Liu Q, Pang D, Hu G, Zhou X (2012) A neighbor cooperation
channels. Because SUs must avoid PUs even when the sender
framework for time-efficient asynchronous channel hopping rendez-
and receiver occupy the same channel, this property is helpful vous in cognitive radio networks. Dynamic Spectrum Access
in reducing the average TTR. When compared with other Networks (DYSPAN), pp. 529539
schemes, the simulation results revealed that the proposed 10. Bian K, Park JM (2011) Asynchronous channel hopping for estab-
lishing rendezvous in cognitive radio networks. IEEE INFOCOM,
scheme can stably achieve the aforementioned objective. The
pp. 236240
proposed TCS scheme appropriately arranges SUs in a data 11. Silvius MD, Ge F, Young A, MacKenzie AB, Bostian CW (2008)
transmission state without SU contention when PUs are absent. Smart radio: spectrum access for first responders. Proc. SPIE 6980,
Six schemes with various parameters were compared through Wireless Sensing and Processing III, 698008, pp. 112. doi:10.1117/
12.777678
exhaustive simulations. The simulation results revealed that the
12. Jia J, Zhang Q (2013) Rendezvous protocols based on message
property of continuous rendezvous in various channels was passing in cognitive radio networks. IEEE Trans Wirel Commun
more helpful when the PU load was higher. The TCS approach 12(11):55945606
greatly increased the system throughput and reduced the data 13. Yu L, Liu H, Leung YW, Chu X, Lin Z (2013) Multiple radios for
transmission time. The simulation results also revealed that for effective rendezvous in cognitive radio networks. IEEE Int Conf
Commun, ICC, p. 28572862
SUs in a data transmission state, it is better to wait in the same 14. Reguera V, Guerra EO, Souza RD, Fernandez E, Brante G (2014)
channel for a PU to leave. Because the wait policy combined Short channel hopping sequence approach to rendezvous for cogni-
with the designed continuity sensing period not only assigns tive networks. IEEE Commun Lett 18(2):289292
high channel-access priority to SUs in a data transmission state 15. DaSilva L, Guerreiro I (2008) Sequence based rendezvous for dy-
namic spectrum access. IEEE Intl Symp New Front Dyn Spectr
but also reduces the occurrence SU contention that is in the Access Netw pp. 17
rendezvous state. In this study, SUs were assumed to occupy 16. Wang YC (2011) The analysis of shift sequence based channel
only one access channel; however, multiple antennas can be selection in cognitive radio network. Master Thesis, National
equipped in advanced devices. Devices with more than one Central University
17. IEEE 802.22 Working Group on Wireless Regional Area Networks.
antenna might improve the flexibility of rendezvous in CRNs http://www.ieee802.org/22/
(e.g., relay-based rendezvous), and this constitutes a future 18. ECMA Std. 392 (2009) MAC and PHY for Operation in TV White
research direction. Space

Acknowledgments This research work was supported in part by the


grants from the National Science Council (NSC) (grant numbers: NSC
98-2221-E-008-063, NSC 99-2218-E-159-001, NSC 100-2221-E-008-
Yen-Wen Chen received the Ph.D.
097, and NSC 101-2221-E-159-026), Taiwan, ROC.
degree in Electronic Engineering
from National Taiwan University of
Science and Technology (NTUST) in
References 1997. During 1983 to 1998, he
worked at Chunghua Telecommuni-
cation Laboratories, Taiwan and was
1. Wang B, Liu KJR (2011) Advances in cognitive radio networks: a a project manager of the broadband
survey. IEEE J Sel Top Signal Process 5:523 switching systems. From August
2. Dhar S, Ray A, Bera R (2013) Cognitive vertical handover engine for 1998 to July 2000, he joined the De-
vehicular communication. Peer-to-Peer Netw Appl 6(3):305324 partment of Information Manage-
3. FCC (2003) Facilitating opportunities for flexible, efficient and reli- ment, Central Police University. Since
able spectrum use employing cognitive radio technologies: Notice of August 2000, Dr. Chen has joined
proposed rule making and order, ET Docket No 03222 the Department of Communication
4. Akyildiz IF, Lee WY, Vuran MC, Mohanty S (2006) NeXt Engineering of National Central Uni-
generation/dynamic spectrum access/cognitive radio wireless net- versity. Currently, he is a professor. His research interests include broadband
works: a survey. Comput Netw 50:21272159 mobile networks, QoS management, sensor networks, network applications,
5. Theis NC, Thomas RW, DaSilva LA (2011) Rendezvous for cogni- and multimedia networks. Dr. Chen is a member of the IEEE communication
tive radios. IEEE Trans Mob Comput 10(2):216227 society.
Peer-to-Peer Netw. Appl.

Po-Yin Liao is a Ph.D. Candidate Ying-Cheng Wang He received


in the Department of Communi- the M.S. degree in Communica-
cation Engineering of National tion Engineering from National
Central University, Taiwan. His Central University in 2012. His
research interests include peer to research interests are cognitive ra-
peer application, cognitive radio, dio and wireless sensor network.
and wireless sensor network.

You might also like