Effects of Location Awareness on Concurrent
Transmissions for Cognitive Ad Hoc Networks
Overlaying Infrastructure-Based Systems
Li-Chun Wang, Senior Member, IEEE, and Anderson Chen, Student Member, IEEE
Abstract—Through wideband spectrum sensing, cognitive radio (CR) can identify the opportunity of reusing the frequency spectrum of other wireless systems. To save time and energy of wideband spectrum, we investigate to what extent a CR system incorporating the location awareness capability can establish a scanning-free region where a peer-to-peer ad hoc network can overlay on an infrastructure-based network. Based on the carrier sense multiple access with collision avoidance (CSMA/CA) medium access control (MAC) protocol, the concurrent transmission probability of a peer-to-peer connection and an infrastructure-based connection is computed. It is shown that the frequency band of the legacy system can be reused up to 45 percent by the overlaying cognitive ad hoc network when CR users have the location information of the primary and secondary users.Index Terms—Ad hoc networks, cognitive radio, carrier sense multiple acces, MAC protocol.
Ç
1
I
NTRODUCTIONC
OGNITIVE radio (CR) has attracted a great deal ofattention from both academia and industry recently because unlicensed spectrums become very crowded but some licensed spectrums are not fully utilized [1], [2]. In addition to Federal Communication Council [3], the Defense Advance Research Projects Agency (DARPA) also initiates the next-generation communications (XG) program to develop the so-called opportunistic spectrum access techni-ques for military and emergency applications [4]. The authors in [5] and [6] provide inspiring and comprehensive overviews on the related research issues in a CR system. To establish a harmless communication link in the presence of the existing legacy systems, a CR user is required to
1. sense wideband spectrums [7], [8], [9],
2. identify the primary user’s spectrum usage in terms of locations and time [10], [11], [12], [13], [14], and 3. realize the spectrum sharing opportunities between
the primary and secondary users by adjusting the transmission parameters accordingly [15], [16], [17], [18], [19], [20].
However, wideband spectrum sensing requires sophisti-cated and energy-consuming signal processing [16].
Instead of developing another efficient spectrum sen-sing technique, in this paper, we discuss a challenging but fundamental issue—Can CR users effectively identify the available spectrum holes without wideband spectrum sensing? Intuitively, when the secondary CR users are far away from the primary user of the legacy system, both
CR and the primary users can concurrently transmit their data without causing interference. If a CR device knows the region where it can concurrently transmit with the primary user, a CR system does not need to rely on the time- and energy-consuming wideband spectrum scanning to detect the spectrum holes. In addition, it is clear that concurrent transmission can enhance the overall through-put. In this sense, identifying concurrent transmission opportunity shall be given a higher priority over spectrum sensing for a CR user.
The next important issue is how to identify the concurrent transmission region where CR users will not cause interference to the legacy wireless systems. In this paper, we propose to utilize the location awareness techniques to help CR users identify the concurrent transmission opportunity. Our specific goal is to dimension the concurrent transmission region where CR devices can establish an overlaying ad hoc network on top of an infrastructure-based legacy system. The overlaying ad hoc network can be considered an important application for CR devices because it can reuse the underutilized spectrum and significantly improve the efficiency of the frequency band. A possible example is that a group of users who want to share data with each other may require to set up such an overlaying ad hoc network. Suppose that their communication devices only can access the licensed spectrum. Obviously, all of them may not want to pay just for a temporary data transfer. To solve this situation, an overlaying ad hoc link in the presence of primary connection may be the network architecture that can satisfy their requirement. We believe this network topology will become more and more attractive because of the spectrum scarcity as the advance of the wireless technology. We will also investigate the throughput improvement resulting from concurrent transmissions based on the carrier sense multiple access with collision avoidance (CSMA/CA) medium access control (MAC) protocol.
In the literature, the coexistence issue of the hybrid infrastructure-based and overlaying ad hoc networks has
. The authors are with the Department of Communication Engineering, National Chiao-Tung University, 1001 Ta Hsueh Road, Hsinchu 300, Taiwan, ROC.
E-mail: [email protected], [email protected].
Manuscript received 19 Jan. 2007; revised 13 Dec. 2007; accepted 20 Feb. 2008; published online 25 Apr. 2008.
For information on obtaining reprints of this article, please send e-mail to: [email protected], and reference IEEECS Log Number TMC-0023-0107. Digital Object Identifier no. 10.1109/TMC.2008.72.
been addressed but in different scenarios. In [21], [22], [23], and [24], the idea of combining ad hoc link and infra-structure-based link was proposed mainly to extend the coverage area of the infrastructure-based network. That is, the coverage area of ad hoc networks is not overlapped with that of the infrastructure-based network. In the present hybrid ad hoc/infrastructure-based network, as shown in Fig. 1, the peer-to-peer CR users are located within the coverage area of the existing legacy wireless network. In [25], to further improve the throughput of a wireless local area network (WLAN), it was suggested that an access point (AP) could dynamically switch between the infrastructure mode and the ad hoc mode. In our considered scenario, the decision of establishing ad hoc connections is made by the CR users in a distributed manner.
The main contribution of this paper is to provide the idea of utilizing location awareness to facilitate frequency sharing in a concurrent transmission manner. Specific achievements are summarized in the following:
. We show that a CR device having location informa-tion of other nodes can concurrently transmit a peer-to-peer data in the presence of an infrastructure-based connection in some region. We also dimension the concurrent transmission (or the scanning-free) region for CR users. Note that a concurrent transmis-sion region of a CR system is equivalent to a scanning-free region. Nevertheless, the wideband spectrum sensing procedure is still needed but is initiated only when the CR user is outside the concurrent transmission region. Therefore, the en-ergy consumption of CR systems with location awareness capability can be reduced significantly. . Based on the CSMA/CA MAC protocol, a physical/
MAC cross-layer analytical model is developed to compute the coexistence probability of a peer-to-peer connection and an infrastructure-based connection.
Based on this analytical model, we find that concurrent transmission of the secondary CR users and the primary users in the legacy system can significantly enhance the total throughput over the pure legacy system.
The rest of this paper is organized as follows: Section 2 describes the system model. Section 3 analyzes the coex-istence probability of both the infrastructure and ad hoc modes sharing the same frequency spectrum simulta-neously. The impacts of shadowing on this coexistence probability are investigated in Section 4. The physical (PHY) and MAC cross-layer performance analysis is provided in Section 5. Section 6 shows the numerical results. The concluding remarks are given in Section 7.
2
S
YSTEMM
ODEL ANDD
EFINITIONFig. 1 illustrates a hybrid ad hoc/infrastructure-based network consisting of two CR devices (MS1 and MS2) and
a primary user MS3. Assume that the secondary CR users
MS1and MS2try to make a peer-to-peer connection, and the
primary user MS3 has been connected to the base station
(BS) or AP of the legacy infrastructure-based system. In the figure, MS1, MS2, and MS3are located at ðr1; 1Þ, ðr2; 2Þ, and
ðr3; 3Þ, respectively; the coverage area of the BS is R2. All
the primary and secondary users stay fixed or hardly move. We assume that the CR devices can perform the positioning technique to acquire their relative or absolute position by using GPS or detecting the signal strength from the BSs of legacy systems [26], [27], [28], [29], [30], [31]. The location information is broadcasted by using the geogra-phical routing protocols [32], [33], [34]. Although both the positioning and geographical routing may waste time and consume energy, they have no need to be processed for every data transmission. They are only performed when a new node joins or the node changes its position. Further-more, with the help of upper layers, the location informa-tion is already stored in the device. Therefore, compared to the spectrum sensing at every transmission, we believe that the additional energy consumption and memory space due to the positioning and location update is relatively small. The overhead and optimal reserved resources for acquiring the location information are beyond the scope of this paper, but the relative research works have been studied in [35] and [36].
Based on the CSMA/CA MAC protocol, multiple users contend the channel, and only one mobile station within the coverage of the BS can establish an infrastructure-based communication link at any instant. To set up an extra peer-to-peer ad hoc connection in the same frequency band of the primary user, the secondary users not only require to ensure that the current infrastructure-based link quality cannot be degraded but also has to win the contentions between other feasible secondary users. Here, we consider that both primary and secondary users have identical transmit power. It is reasonable to assume that only one secondary user can establish a link after the contention at one instance due to the similar interference range. Denote SIRi and
SIRa as the received signal-to-interference ratios (SIRs) of
the infrastructure-based and ad hoc links, respectively. Then, we can define the coexistence (or concurrent
Fig. 1. An illustrative example for the coexistence of two CR devices establishing a peer-to-peer ad hoc link and a primary user connecting to the infrastructure-based network, where all the devices (MS1, MS2,
transmission) probability ðPCTÞ of the infrastructure-based
link and CR-based ad hoc link in an overlapped area as follows:
PCT ¼ P ðSIRf i> ziÞ \ ðSIRa> zaÞg; ð1Þ
where zi and za are the required SIR thresholds for the
infrastructure-based and ad hoc links, respectively. To obtain the concurrent transmission region, it is crucial to calculate the coexistence probability of both the infrastruc-ture and ad hoc links. If the link quality of the primary user cannot be guaranteed, CR devices have to sense and change to other frequency bands.
We consider the two-ray ground reflection model in which there exists two paths between the transmitter and receiver [37]. One is the line-of-sight, and the other is reflected from ground. Thus, the received power can be written as
Pr ¼
Pth2bsh2msGbsGms10
10
r ; ð2Þ
where Pr and Pt are the received and transmitted power
levels at a mobile station, respectively; hbsand hmsrepresent
the antenna heights of the BS and the mobile station, respectively; Gbsand Gmsstand for the antenna gains of the
BS and the mobile station, respectively; r is the distance between the transmitter and receiver; is the path loss exponent; and 10=10is the lognormally distributed
shadow-ing component.
3
S
IGNAL-
TO-I
NTERFERENCER
ATIOA
NALYSIS3.1 Uplink SIR Analysis
In the uplink case, when the primary user MS3transmits data
to the BS, denote SIRðuÞi as the uplink SIR of MS3and let P30
and P10be the received power from MS3and MS1at the BS,
respectively. Then, from (2), we have SIRðuÞi ¼ r1 r3 ¼P30 P10 ; ð3Þ where r1and r3are the distances between MS1and MS3 to
the BS, respectively. Similarly, the SIR of a peer-to-peer ad hoc link from MS1to MS2can be written as
SIRa¼ P12 P32 ¼ d23 d12 ; ð4Þ where P12is the received power at MS2from MS1and P32is
the interference power from MS3; d12 and d23 are the
distances from MS1 and MS3 to MS2, respectively.
Sub-stituting (3) and (4) into (1), the concurrent transmission probability PCTðuÞ in the uplink case can be written as
PCTðuÞ¼ P r3z1=i < r1< R \ d12< d23 za1= ¼R ðuÞ CT R2: ð5Þ
Note that RðuÞCT denotes the concurrent transmission region where MS1 can connect to MS2 without interfering the
uplink signal of MS3 to the BS. As shown in Fig. 2, the
condition ðr3z1=i < r1< RÞ leads to a donut-shaped area
consisting of two circles centered at the BS with the radii of r3z1=i and R, respectively. Meanwhile, the condition ðd12<
d23=z1=a Þ yields the circular area centering at MS2 with a
radius of d23=z1=a . From the figure, the region R ðuÞ CT can be computed as RðuÞCT ¼ d23 z1=a !2 A1 A2; ð6Þ where A1¼ d23 z1=a !2 ð 0Þ R2þ 2 ð7Þ and A2¼ d23 z1=a !2 r3z 1= i 2 0 20: ð8Þ The definitions of parameters , 0, , 0, , and 0and the
detailed derivation of (6), (7), and (8) are discussed in Appendix A.
3.2 Downlink SIR Analysis
Now, we consider the downlink case when the BS sends data to the primary user MS3. Denote SIRðdÞi as the infrastructure
link’s SIR in the downlink direction. Then, from (2), we have SIRðdÞi ¼P03 P13 ¼ hbs hms 2 d13 r3 ; ð9Þ where P03and P13 are the MS3’s received powers from the
BS and MS1, respectively; d13 stands for the distance from
MS1 to MS3; hbs, hms, and r3 are given in (2) and (3).
Similarly, the ad hoc link’s SIR from MS1 to MS2 can be
expressed as SIRa¼ P12 P02 ¼ hms hbs 2 r2 d12 ; ð10Þ where P12and P02are the received powers at MS2from MS1
and the BS, respectively; r2represents the distance between
MS2and the BS; d12, hbs, and hms are defined in (4) and (2).
Substituting (9) and (10) into (1), we can obtain the concurrent transmission probability PCTðdÞ of a CR-based
Fig. 2. Physical representation of the coexistence probability for the concurrent transmission of hybrid CR-based ad hoc link and infra-structure uplink transmission.
peer-to-peer ad hoc link and an infrastructure-based down-link transmission as follows:
PCTðdÞ¼ P d13> r3z01=i \ d12< r2z01=a \ rð 1< RÞ n o ¼ R ðdÞ CT R2; ð11Þ where z0 i¼ zi ðh2ms=h2bsÞ, and z0a¼ ð1=zaÞ ðh2ms=h2bsÞ. From
(11), the concurrent transmission region RðdÞCT in the downlink
case is shown in Fig. 3. The criterion ðd13> r3z01=i Þ results in
the region outside the circle centered at MS3 with a radius
of r3z01=i , while the criterion ðd12< r2z01=a Þ yields the region
inside the circle centered at MS2with radius r2z01=a . At last,
r1< Rbecause MS1is assumed to be uniformly distributed
within a cell of radius R.
The coexistence probability of the CR-based ad hoc link and the infrastructure-based downlink can be obtained by
calculating the area of RðdÞCT. The distances from the AP to the intersections of the two circles with radii of r3z01=i and
r2z01=a are denoted by rþ and r as shown in Fig. 3. In
Appendix B, we have shown that rþ¼ 1 d2 23 ( sinð2 3Þ þ r2r3 h r2r3 z 02 a þ z0 2 i þ cosð2 3Þ d223 r 2 2z 02 a r23z0 2 i i) ð12Þ and r¼ 1 d2 23 ( sinð2 3Þ r2r3 h r2r3 z 02 a þ z0 2 i þ cosð2 3Þ d223 r 2 2z 02 a r23z0 2 i i) ; ð13Þ where ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 2r2 3z 02 i d223þ r22z 02 a d2 23 r22z 02 a 2 r4 3z 04 i r : ð14Þ With the values of rþand r, RðdÞ
CT can be calculated in the
following two cases:
1. maxðrþ; rÞ R: In this case, referring to Fig. 3a, the
area of RðdÞCT can be expressed as
RðdÞCT ¼ d23z01=a 2 A1 A2; ð15Þ where A1¼ r2z01=a 2 ð 0Þ R2þ 2; ð16Þ A2¼ r2z01=a 2 r3z01=i 2 0 20: ð17Þ
2. maxðrþ; rÞ > R: As shown in Fig. 3b, the area of
RðdÞCT can be expressed as RðdÞCT ¼ d23z01=a 2 A1 A2þ A3; ð18Þ where A1¼ r2z01=a 2 ð 0Þ R2þ 2; ð19Þ A2¼ r2z01=a 2 r3z01=i 2 0 20; ð20Þ A3¼ 00þ r3z01=i 2 2 1 2 r3z 01= i 2 sin 2 þ r2z01=a 2 3 1 2 r2z 01= a 2 sin 3 R2 1 1 2R 2sin 1 : ð21Þ
The detailed derivations of (15) and (18) and the definitions of the parameters , 0, , 0,
1, 2, 3, , 0,
and 00 are given in Appendices C and D, respectively.
Fig. 3. The area of concurrent transmission region RCT in downlink
cases. (a) The case when maxðrþ; rÞ R. (b) The case when
3.3 Multiple Ad Hoc Connections Coexisting with One Infrastructure Link
After evaluating the concurrent transmission probability of the infrastructure link and the one overlaying CR-based ad hoc link, one may be interested in knowing how many secondary users can concurrently establish ad hoc links together with the primary user. This question is nontrivial since it needs to consider the interference from a set of ad hoc links to the infrastructure link, and vice versa. Besides, different from the pure infrastructure network, both the locations of the transmitter and receiver in an ad hoc link are random.
Instead of calculating the maximum number of ad hoc links, we suggest constructive procedures enabling CR devices to establish ad hoc links in the presence of an infrastructure transmission. The detailed procedures are summarized as follows:
1. Consider a network in which all the primary and secondary users are fixed, and the CR device can learn the locations of its receiver and neighbors by the routing mechanism [17]. Here, we assume that l ad hoc links have been established and coexisted with the infrastructure link at the same time. Before establishing a new ad hoc connection, the CR device has to overhear the channel and memorize the locations of all the existing transmitters.
2. With the location information, the new CR device starts evaluating the concurrent transmission region RCT. The device should consider the interference
from the infrastructure link as well as other existing ad hoc links, and vice versa. Denote the indices fp; m; n; kg as the primary user, the transmitter and receiver of the new ad hoc link, and the transmitter of other existing ad hoc link, respectively. Using similar procedures in deriving (5), the three conditions in the infrastructure uplink case can be written by
rm 1 1 zi 1 rp Pk 1 rk 0 B @ 1 C A 1 ; ð22Þ dmn 1 za d1pn þPk 1 dkn 0 B @ 1 C A 1 ; ð23Þ rm R; ð24Þ
where ri and dij are the distances between the BS
and CR device j to i, respectively. Similarly, from (11), the three criteria in the downlink case are
dmp 1 1 zi 1 rp Pk d1kp 0 B @ 1 C A 1 ; ð25Þ dmj 1 za r1n þPk 1 dkn 0 B @ 1 C A 1 ; ð26Þ rm R: ð27Þ
3. Since the concurrent transmission regions RðuÞCT and RðdÞCT are known, the CR device can determine whether it can concurrently transmit data together with the infrastructure link and other ad hoc connections by the primary user and other second-ary CR users.
4
S
HADOWINGE
FFECTSIn the previous section, we only consider the impact of path loss on the concurrent transmission probability of CR-based network overlaying the infrastructure-CR-based system. However, even though the CR device is located inside the concurrent transmission region RCT, a peer-to-peer ad hoc
connection may not be able to coexist together with the primary user’s infrastructure link due to shadowing. Thus, it is important to investigate the reliability of concurrent transmissions of the hybrid infrastructure and CR-based ad hoc network when shadowing is taken into account.
Shadowing can be modeled by a lognormally distrib-uted random variable [38]. Represent 10ij=10 as the
shadowing component in the propagation path from users i to j, where ij is a Gaussian random variable with zero
mean and standard deviation of . Thus, the uplink and
downlink SIRs in both the infrastructure-based connection and CR-based ad hoc link are modified as follows:
. uplink case: SIRðuÞi ð30; 10Þ ¼ 103010=r 3 101010=r 1 ; ð28Þ SIRðuÞa ð12; 32Þ ¼ 101210=d 12 103210=d 23 ; ð29Þ . downlink case: SIRðdÞi ð03; 13Þ ¼ 100310=r 3 101310=d 13 ; ð30Þ SIRðdÞa ð12; 02Þ ¼ 101210=d 12 100210=r 2 : ð31Þ Note that the index 0 represents the BS and 30of (29) in the
uplink case is equivalent to 03of (31) in the downlink case.
Let ¼ ð10; 30; 12; 32Þ and 0¼ ð13; 03; 12; 02Þ and
as-sume that these shadowing components are identical and independently distributed. Taking shadowing into account, the concurrent transmission probability PCT can be
repre-sented by . uplink case: PCTðuÞðÞ ¼ P ( r3 zi10 1030 10 1= < r1< R \ d12< d23 za10 1232 10 1= ) ; ð32Þ
. downlink case: PCTðdÞð0Þ ¼ P ( d13> r3 zi10 1303 10 1= \ d12< r2 za10 1202 10 1= \ rð 1< RÞ ) : ð33Þ We define the reliability of uplink concurrent transmis-sion FCTðuÞðÞ as the probability that, in the region RCT, a
CR device can successfully establish an ad hoc link in the presence of the primary user’s uplink transmission subject to the shadowing effect. That is,
FCTðuÞðÞ ¼ P ( SIRðuÞi ð30; 10Þ > zi \ SIRðuÞa ð12; 32Þ > z jMS12 RðuÞCT ) : ð34Þ
Note that FCTðuÞðÞ ¼ 1 when shadowing is not considered.
Substituting (28) and (29) into (34), we can obtain FCTðuÞðÞ ¼ P ( 3010> 10 log10 zi r3 r1 \ 1232> 10 log10 za d12 d23 jMS12 RðuÞCT ) : ð35Þ Assume that ij have the same standard deviation for all
i and j and let ðuÞi ¼ 30 10, ðuÞa ¼ 12 32. Then, ðuÞi
and ðuÞa become Gaussian random variables with
Nð0; 2Þ. Hence, it follows that
FCTðuÞðÞ ¼ P ðuÞi 10 log10 zi
r3 r1 jMS12 RðuÞCT P aðuÞ 10 log10 za d12 d23 jMS12 RðuÞCT ¼ Q 10 log10 zi r3 r1 2pffiffiffi2 0 @ 1 A Q 10 log10 za d12 d23 2pffiffiffi2 0 @ 1 A; ð36Þ where QðxÞ ¼ ð1=Þ Rx1expx2dx.
Following similar procedures in the uplink case, we can also obtain the reliability of downlink concurrent transmis-sion as follows: FCTðdÞð0Þ ¼ P ( SIRðdÞi ð03; 13Þ > zi \ SIRðdÞa ð12; 02Þ > z jMS12 RðdÞCT ) ¼ Q 10 log10 zi dr133 2pffiffiffi2 0 @ 1 A Q 10 log10 za d12 r2 2pffiffiffi2 0 @ 1 A: ð37Þ
5
MAC L
AYERT
HROUGHPUTA
NALYSISIn this section, the MAC layer throughput performance of the considered hybrid infrastructure and overlaying CR-based ad hoc network is evaluated from a PHY/MAC cross-layer perspective. The main task here is to incorporate the interference from both the infrastructure and ad hoc links into the throughput evaluation model in the MAC layer.
In this paper, the CSMA/CA MAC protocol with the binary exponential back-off algorithm is considered because it is widely deployed in many license-exempt frequency bands. However, the CSMA/CA MAC protocol may not be used to establish the CR-based ad hoc link since the clear channel assessment (CCA) by measuring the received signal strength (RSS) may forbid the transmissions in the presence of infrastructure link. To remove this constraint, we use the location and channel station information to replace the RSS measurement for CCA in the traditional CSMA/CA MAC protocol. Therefore, the CR device can establish the ad hoc connection once the new connection does not injure the existing primary infrastructure link.
Next, we first summarize the calculation of saturation throughput in the traditional CSMA/CA MAC protocol [39], [40]. Assume N stations always transmit data packets in the network, and let W and 2bW be the minimum and
maximum back-off window sizes, respectively. Given the stationary transmission probability that a station transmits packet in a given slot and the successful transmission probability psðNÞ, the throughput SðNÞ of the CSMA/CA
MAC protocol can be expressed as [39], [40] SðNÞ ¼ ptrpsðNÞE½P
ð1 ptrÞ þ ptrð1 psðNÞÞTcþ ptrpsðNÞTc
; ð38Þ where ptr¼ 1 ð1 ÞN; E½P, Ts, Tc, and are the average
payload size, the average successful transmission duration, the average collision duration, and the slot duration. Note that the average successful transmission and collision duration have included the DIFS and SIFS waiting time [39], [40]. The stationary transmission probability is a function of the packet loss probability pL, that is
ðpLÞ ¼ 2 1þ W þ pLWP b1 i¼0 ð2pLÞi : ð39Þ
Note that both the packet loss probability pL and the
successful transmission probability psðNÞ are influenced by
the radio channel effect and the multiuser capture effect in the physical layer [40].
Then, we evaluate the total throughput performance of the concurrent transmission in the hybrid infrastructure and overlaying CR-based ad hoc network. Here, we assume NCR
CR devices and N non-CR devices using the same frequency band in the coverage of a BS. Since the CR device can establish an ad hoc connection without interfering the existing infrastructure link, the total throughput of such a hybrid network SCT is independently contributed by the two links.
The throughput of the infrastructure link and the CR-based ad hoc connection are denoted by Siand Sa, respectively. The
total throughput SCT can then be expressed as
Since N non-CR devices contend for data transmission to the BS, the throughput of infrastructure-based link Si is the
same as (38). As for the throughput of the cognitive overlaying ad hoc connection Sa, it is similar to Si, but the
number of contending stations changes to NCRPCT because
only NCRPCT CR devices have the opportunity to establish
the connection. Although the secondary user employs the location information to replace the carrier sensing in the CSMA/CA, it still needs to wait for the DIFS and SIFS durations and makes sure if other secondary users are transmitting data. Once the user finds another one estab-lishes a link, it still has to freeze its back-off counter and resumes the back-off countdown after the link is complete, just like what the CSMA/CA MAC does. Therefore, all the throughput calculations for the secondary users is similar to that for the CSMA/CA MAC protocol. The only discrepancy between the two systems is merely the number of contend-ing stations. Because only the secondary users within the same concurrent transmission region have the right to access the spectrum, the number of users for the overlaying ad hoc network is the function of the coexistence probability.
6
N
UMERICALR
ESULTSIn this section, we first investigate the concurrent transmission probability of the infrastructure and over-laying CR-based ad hoc network. Then, we apply the proposed cross-layer analytical model to evaluate the total throughput performance in this hybrid network. Fig. 1 illustrates the considered network topology, where MS1,
MS2, and MS3 are the CR-based ad hoc transmitter,
receiver, and infrastructure primary user, respectively. The stations MS2 and MS3 are, respectively, located at
(r2;=2) and ðr3; =2Þ, where r2 and r3 are the distances
between the BS to MS2and MS3; whereas MS1is uniformly
distributed in the cell with radius R ¼ 100 m. In addition, we also perform the simulation to verify the proposed analytical model for the concurrent transmission prob-ability PCT. In the simulation, 104 points, which are
uniformly distributed in the region R2, represent the
possible locations of the ad hoc transmitter MS1. The
probability PCT is calculated by counting the number of
points where MS1can successfully establish an ad hoc link
to MS2 in the presence of the infrastructure link (BS to
MS3). As shown in the following figures, the results in the
analytical model agrees well with that in the simulation. The other system parameters are listed in Table 1.
6.1 Uplink Concurrent Transmission Probability Fig. 4 shows the impact of the primary user’s location on the uplink concurrent transmission probability PCTðuÞ, where the transmission power Pt¼ 20 dBm and noise power
N0¼ 90 dBm, respectively; the required link SIR threshold
is 0 or 3 dB. First, one can see that the analytical results match the simulation results well. Second and more importantly, there exists an optimal concurrent transmission probability PCTðuÞagainst the distance r3from the primary user MS3to the
BS. Note that, for zi ¼ 0 dB, the maximal PCTðuÞ ¼ 0:45
at r3 ¼ 40 m, and for zi ¼ 3 dB, the maximal PCTðuÞ ¼ 0:22 at
r3¼ 26 m. This phenomenon can be explained as follows: On
the one hand, when MS3approaches the BS, it is also closer to
the CR-based ad hoc receiver, thereby causing higher interference and decreasing the concurrent transmission probability. On the other hand, when MS3moves away from
the BS, its uplink SIR decreases due to the weaker signal strength and thus yields a lower PCTðuÞ. Hence, an optimal
primary user’s location can be found in the sense of max-imizing the uplink concurrent transmission probability PCTðuÞ. Fig. 5 shows the impact of MS2’s locations on the
uplink concurrent transmission probability PCTðuÞ. As shown in the figure, as the CR-based ad hoc user moves away from the BS, the concurrent transmission probability monotonically increases from 10 percent to 50 percent because the interference from the infrastructure-based link to the ad hoc connection decreases.
6.2 Downlink Concurrent Transmission Probability Fig. 6 shows the downlink concurrent transmission prob-ability PCTðdÞversus the distance r3of the primary user MS3to
the BS when user MS2 is located at ð50; =2Þ. For the SIR
requirement zi¼ za¼ 0 dB, PCTðdÞ¼ 25% is a constant in the
range of r3 100 m. This is because the interference
transmitted from the BS to the ad hoc users is independent of the locations of the primary user, MS3. However, a more
stringent SIR requirement zi ¼ za¼ 3 dB yields a lower and
TABLE 1 System Parameters
Fig. 4. The concurrent transmission probability PCTðuÞ versus the infrastructure uplink user’s locations as the ad hoc receiver MS2 is
located atð50; =2Þ, where r3is the distance between the BS and the
decreasing downlink concurrent transmission probability when r3 increases.
Fig. 7 shows the impact of CR user MS2’s locations on
the downlink concurrent transmission probability. Similar to Fig. 5, PCTðdÞ also monotonically increases when CR user
MS2moves away from the BS. However, comparing Figs. 5
and 7, the uplink’s concurrent transmission probability is higher than that of the downlink’s. For zi¼ za¼ 0 dB and
r2¼ 100 m, PCTðuÞ¼ 49 percent and P ðdÞ
CT ¼ 39 percent,
re-spectively. This is because in the considered scenario the interference to the ad hoc user from the infrastructure-based uplink transmission is weaker than that from the downlink transmission.
6.3 Effects of Shadowing on the Concurrent Transmission
Figs. 8a and 8b illustrate the reliability of the concurrent transmissions with various shadowing standard deviations versus r3 and r2, respectively. In general, comparing
¼ 6 dB and ¼ 1 dB, one can find that the larger
shadowing variance leads to a lower reliability for both uplink and downlink concurrent transmissions. For exam-ple, in Fig. 8a, when the primary user’s distance to the BS r3
Fig. 6. Impact of primary user MS3’s location on the downlink concurrent
transmission probability PCTðdÞas the ad hoc receiver MS2 is located at
ð50; =2Þ.
Fig. 7. Impact of CR-based ad hoc receiver MS2’s location on the
concurrent transmission probability PCTðdÞas the infrastructure downlink user MS3is located atð50; =2Þ.
Fig. 8. Impacts of shadowing on the reliability of downlink FCTðdÞ(solid line) and uplink FCTðuÞ (dotted line) concurrent transmission against the locations of (a) the primary user MS3 and (b) the ad hoc user MS2in
the cases of ¼ 1 and 6 dB, respectively.
Fig. 5. The concurrent transmission probability PCTðuÞversus the CR-based ad hoc receiver’s location as the infrastructure uplink user MS3is located
atð50; =2Þ, where r2is the distance between the BS and ad hoc link
is in the range of 0 100 m, FCTðdÞ is larger than 0.9 for ¼ 1 dB, whereas it decreases to 0.6 0.7 for ¼ 6 dB.
However, when the primary user moves to the cell edge, the reliability of uplink and downlink concurrent transmissions decreases due to shadowing and weaker RSS. As shown in Fig. 8a, for ¼ 6 dB, FCTðdÞand FCTðuÞdecrease from 0.7 to 0.5
and 0.4, respectively. Since the uplink signal strength is weaker than the downlink signal, the reliability of uplink concurrent transmission is usually more sensitive to shadowing effects than downlink concurrent transmission, especially when the primary user is at the cell edge. In Fig. 8b, it is shown that, subject to the influence of shadowing, the reliability of uplink and downlink con-current transmissions increases when the receiver MS2 of
the ad hoc link approaches the cell edge. For ¼ 1 dB, FCTðuÞ
and FCTðdÞ increase from 0.4 and 0.63 to 0.89 and 0.92 as r2
increases to 100 m; for ¼ 6 dB, FCTðuÞand F ðdÞ
CT also increase
from 0.29 and 0.4 to 0.54 and 0.62. Clearly, the interference from the primary user to the ad hoc user becomes weaker when ad hoc users moves away from the BS. As a result, the reliability of concurrent transmission increases and the shadowing effect on the reliability remains constant as r2>
30m for ¼ 1 dB and r2> 60m for ¼ 6 dB.
6.4 Total Throughput of Cognitive Ad Hoc Networks Overlaying Infrastructure-Based System with Concurrent Transmission
Fig. 9 demonstrates the total throughput of the CR-based ad hoc link and the infrastructure-based uplink transmis-sions for various numbers of ad hoc users and different locations of primary users. The total throughput is normalized to the infrastructure-based uplink capacity. As shown in the figure, in the worst case at r3¼ 50 m the
total throughput with the concurrent transmission is still 145 percent higher than the pure infrastructure-based uplink, and the total throughput reaches a maximum of 173 percent at r3¼ 10 m.
Fig. 10 shows the total throughput performance of the concurrent transmission of infrastructure-based downlink and ad hoc link. In this case, the concurrent transmission
probability is constant for various locations of primary users as shown in Fig. 6. Thus, the throughput is mainly affected by the number of ad hoc users. For NCR¼ 50, the
total throughput is 157 percent when 10 < r3< 100 m.
However, when r3¼ 50 m, the total throughput improves
from 148 percent to 173 percent as NCR is changed from
100 to 10.
7
C
ONCLUSIONSIn this paper, we identified a critical region RCT in which
the overlaying cognitive ad hoc users and the primary user can concurrently transmit data without causing interfer-ence to each other. If the location information of other nodes is available, such a concurrent transmission region can be easily identified. There are two major advantages of identifying the concurrent transmission opportunity. First, the overall throughput of the concurrently transmitted data obtained by combining both the overlaying cognitive ad hoc networks and the legacy infrastructure-based system is much higher than that of the pure infrastruc-ture-based system. Our numerical results show that, in the uplink case, the concurrent transmission region subject to 1 and 6 dB shadowing standard deviation can be up to 45 percent out of the entire cell area with about 90 percent and 60 percent reliability, respectively. Second, if such a concurrent transmission opportunity can be identified first, it is clear that the need of the time- and energy-consuming wideband spectrum scanning process required by most existing CR systems can be reduced dramatically.
In summary, the results presented in this paper provide a design paradigm for CR systems from an alternative perspective—identifying spectrum opportunity cannot just rely on wideband spectrum sensing techniques. Location awareness also provides another important resource to help a CR system identify the opportunities of concurrently using the spectrum of the primary user in the spatial domain, instead of just sharing the spectrum in the time domain. It is hoped that our work has shed some light in this promising area of research. Nevertheless, there are still many un-solved open issues to realize the idea of exploiting the
Fig. 9. Total throughput performance of the uplink concurrent transmission.
Fig. 10. Total throughput performance of the downlink concurrent transmission.
concurrent transmission opportunity by taking advantages of location awareness. Many interesting research topics, including the concurrent transmission MAC protocol de-sign, the efficient mechanism of location information exchange, and the theoretical upper bound on the max-imum concurrent transmission links overlaying the legacy infrastructure-based systems, are worthwhile to be inves-tigated further.
A
PPENDIXA
D
ERIVATION OF(6)
The region RðuÞCT shown in (6) is composed of three sections with the areas of ðd23=z1=a Þ
2
, A1, and A2. Fig. 11 shows all
the parameters used in deriving the area of A1and A2in (7)
and (8), respectively. In (7), section A1is composed of two
fan-shaped areas with the measures of ðd23=z1=a Þ 2
ð 0Þ
and R2, and two identical triangles with the lengths of R,
r2, and d23=z1=a , where ¼ cos1R 2þ r2 2 d23 z1=a 2Rr2 ; ð41Þ 0¼ cos1 r22þ d23 z1=a 2 R2 2r2 d23 z1=a ; ð42Þ ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sðs RÞðs r2Þ s d23 z1=a ! v u u t ; ð43Þ and s¼ Rþ r2þzd1=23 a 2 : ð44Þ
Similarly, in (8), section A2 is also made up of two
fan-shaped areas with the measures of ðd23=z1=a Þ 2
and ðr3z1=i Þ
20, and the triangle with the lengths of r
2, r3z1=i , and d23=z1=a , where ¼ cos1 r2 2þ d23 z1=a 2 r3z1=i 2 2r2 d23 z1=a ; ð45Þ 0¼ cos1 r2 2þ r3z 1= i 2 d23 z1=a 2 2r2 r3z1=i ; ð46Þ 0¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi s0ðs0 r 2Þ s0 r3z 1= i s0 d23 z1=a ! v u u t ; ð47Þ and s0¼ r2þ r3z 1= i þ d23 z1=a 2 : ð48Þ
A
PPENDIXB
D
ERIVATION OF(12)
AND(13)
The distances between the BS and the intersections by the two circles with the radii of r3z01=i and r2z01=a and centered
at ðr3; 3Þ and ðr2; 2Þ are denoted by rþand r, respectively.
Given the locations of intersections ðx; yÞ, they can be obtained by jointly solving the equations as follows:
ðx r2cos 2Þ2þ ðy r2sin 2Þ2¼ r2z01=a
2
; ðx r3cos 3Þ2þ ðy r3sin 3Þ2¼ r3z01=a
2
: (
ð49Þ The distances between the BS and the intersections rþand
r, respectively, shown in (12) and (13), can be obtained by solving the locations ðx; yÞ from (49).
A
PPENDIXC
D
ERIVATION OF(15)
In the infrastructure downlink case, when maxðrþ; rÞ R, the region RðdÞCT is composed of three sections with the areas of ðd23z01=a Þ
2
, A1, and A2. Fig. 12 details the
parameters used to calculate the areas of these sections in (16) and (17). In (16), the region A1 is made up of two
fan-shaped areas with the measures of ðr2z01=a Þ 2
ð 0Þ
Fig. 11. Definition of parameters in the calculation of the area of region RðuÞCT in the infrastructure uplink case.
Fig. 12. Definition of parameters in the calculation of the area of region RðdÞCT in the infrastructure downlink case when maxðrþ; rÞ R.
and R2, and two identical triangles with the lengths of R, r2, and r2z01=a , where ¼ cos1R2þ r22 r2z01=a 2Rr2 ; ð50Þ 0¼ cos1r 2 2þ r2z01=a 2 R2 2r2 r2z01=a ; ð51Þ ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sðs RÞðs r2Þ s r2z01=a r ; ð52Þ and s¼Rþ r2þ r2z 01= a 2 : ð53Þ Similarly, section A2 in (17) is also composed of two
fan-shaped areas with the measures of ðr2z01=a Þ 2
and ðr3z01=i Þ
2
0, and the triangle with the lengths of d23, r2z01=a ,
and r3z01=i , where ¼ cos1 d2 23þ r2z01=a 2 r3z01=i 2 2d23 r2z01=a ; ð54Þ 0¼ cos1 d2 23þ r3z01=i 2 r2z01=a 2 2d23 r3z01=i ; ð55Þ 0¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi s0ðs0 d 23Þ s0 r3z01=i s0 r 2z01=a r ; ð56Þ and s0¼d23þ r2z 01= a þ r3z01=i 2 : ð57Þ
A
PPENDIXD
D
ERIVATION OF(18)
When maxðrþ; rÞ > R, the region RðdÞ
CTcan be separated into
four sections with the areas of ðd23z01=a Þ 2
, A1, A2, and A3.
Fig. 13 shows all the parameters in deriving (19), (20), and
(21). In (19), section A1can be divided into two fan-shaped
areas with the measures of ðr2z01=a Þ 2
ð 0Þ and R2, and the
triangle with the lengths of R, r2, and r2z01=a , where
¼ cos1R 2þ r2 2 r2z01=a 2Rr2 ; ð58Þ 0¼ cos1r 2 2þ r2z01=a 2 R2 2r2 r2z01=a ; ð59Þ ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi sðs RÞðs r2Þ s r2z 01 a r ; ð60Þ and s¼Rþ r2þ r2z 01 a 2 : ð61Þ
In addition, section A2 is made up of two fan-shaped
areas with the measures of ðr2z01=a Þ 2
and ðr3z01=i Þ 2
0, and the triangle with the lengths of d23, r3z01=i , and r2z01=a ,
where ¼ cos1 d2 23þ r2z01=a 2 r3z01=i 2 2d23 r2z01=a ; ð62Þ 0¼ cos1 d2 23þ r3z01=i 2 r2z01=a 2 2d23 r3z01=i ; ð63Þ 0¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi s0 s0 d 23 ð Þ s0 r 3z 01 i s0 r 2z 01 a r ; ð64Þ and s0¼d23þ r2z 01 a þ r3z 01 i 2 : ð65Þ
At last, in (21), section A3 is constructed by three
arc-shaped areas with the measures of ðr3z01=i Þ 2 2 ðr3z01=i Þ 2sin 2 h i =2, ðr2z01=a Þ 2 3 ðr2z01=a Þ 2sin 3 h i =2, and R2
1 R½ 2sin 1=2, and the triangle with the lengths of
a, b, and c, where 1¼ cos1 R2þ r2 2 r22z 02 a 2Rr2 þ cos1R2þ r23 r23z 02 i 2Rr3 cos1r 2 2þ r23 d223 2r2r3 ; ð66Þ 2¼ cos1 d2 23þ r23z 02 i r22z 02 a 2d23 r3z 01 i þ cos1d223þ r23 r22 2d23r3 cos1r23þ r23z 02 i R2 2r3 r3z 01 i ; ð67Þ
Fig. 13. Definition of parameters in the calculation of the areas of sections A1 and A2 in (18) in the infrastructure downlink case when
3¼ cos1 d2 23þ r22z 02 a r23z0 2 i 2d23 r2z 01 a þ cos1d 2 23þ r22 r23 2d23r2 cos1r 2 2þ r22z 02 a R2 2r2 r2z 01 a ; ð68Þ 00¼pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffis00ðs00 aÞðs00 bÞðs00 cÞ; ð69Þ s00¼aþ b þ c 2 ; ð70Þ a¼ 2R sin 1 2 ; ð71Þ b¼ 2 r3zi0 1 sin 2 2 ; ð72Þ and c¼ 2 r2z 01 a sin 3 2 : ð73Þ Fig. 14 shows all the parameters in deriving (21).
A
CKNOWLEDGMENTSThis work was jointly supported by the National Science Council, Taiwan and the MOE ATU Program under Contracts NSC-96-2221-E-009-061, 96-2628-E-009-004-MY3, and 96W803, respectively.
R
EFERENCES[1] E.M. Noam, “Taking the Next Step beyond Spectrum Auctions: Open Spectrum Access,” IEEE Comm. Magazine, vol. 33, pp. 66-73, Dec. 1995.
[2] T.A. Weiss and F.K. Jondral, “Spectrum Pooling: An Innovative Strategy for the Enhancement of Spectrum Efficiency,” IEEE Comm. Magazine, vol. 42, pp. s8-s14, Mar. 2004.
[3] F.C. Commission, Notice of Proposed Rule Making and Order, Report ET Docket 03-108, Dec. 2003.
[4] The XG Vision, http://www.darpa.mil/ato/programs/XG/, 2008. [5] S. Haykin, “Cognitive Radio: Brain-Empowered Wireless Com-munications,” IEEE J. Selected Areas in Comm., vol. 23, no. 2, pp. 201-220, Feb. 2005.
[6] I.F. Akyildiz, W.Y. Lee, M.C. Vuran, and S. Mohanty, “NeXt Generation/Dynamic Spectrum Access/Cognitive Radio Wireless Networks: A Survey,” Computer Networks J., vol. 50, pp. 2127-2159, Sept. 2006.
[7] D. Cabric, S.M. Mishra, and R.W. Brodersen, “Implementation Issues in Spectrum Sensing for Cognitive Radios,” Proc. 38th Asilomar Conf. Signals, Systems, and Computers (ASILOMAR ’04), vol. 1, pp. 772-776, Nov. 2004.
[8] T. Weiss, J. Hillenbrand, A. Krohn, and F.K. Jondral, “Mutual Interference in OFDM-Based Spectrum Pooling Systems,” Proc. 59th IEEE Vehicular Technology Conf. (VTC ’04), vol. 4, pp. 1873-1877, May 2004.
[9] M. Oner and F. Jondral, “Extracting the Channel Allocation Information in a Spectrum Pooling System Exploiting Cyclosta-tionarity,” Proc. 15th IEEE Int’l Symp. Personal, Indoors, and Mobile Radio Comm. (PIMRC ’04), vol. 1, pp. 551-555, Sept. 2004. [10] F. Capar, I. Martoyo, T. Weiss, and F. Jondral, “Comparison of
Bandwidth Utilization for Controlled and Uncontrolled Chan-nel Assignment in a Spectrum Pooling System,” Proc. 55th IEEE Vehicular Technology Conf. (VTC ’02), vol. 3, pp. 1069-1073, May 2002.
[11] F. Capar, I. Martoyo, T. Weiss, and F. Jondral, “Analysis of Coexistence Strategies for Cellular and Wireless Local Area Networks,” Proc. 58th IEEE Vehicular Technology Conf. (VTC ’03), vol. 3, pp. 1812-1816, Oct. 2003.
[12] B. Aazhang, J. Lilleberg, and G. Middleton, “Spectrum Sharing in a Cellular System,” Proc. IEEE Int’l Symp. Spread Spectrum Techniques and Applications (ISSSTA ’04), pp. 355-359, Aug. 2004. [13] T. Weiss, M. Spiering, and F.K. Jondral, “Quality of Service in
Spectrum Pooling Systems,” Proc. 15th IEEE Int’l Symp. Personal, Indoors, and Mobile Radio Comm. (PIMRC ’04), vol. 1, pp. 345-349, Sept. 2004.
[14] Q. Zhao, L. Tong, and A. Swami, “Decentralized Cognitive MAC for Dynamic Spectrum Access,” Proc. First IEEE Int’l Symp. New Frontiers in Dynamic Spectrum Access Networks (DySPAN ’05), pp. 224-232, Nov. 2005.
[15] T.W. Rondeau, B. Le, C.J. Rieser, and C.W. Bostian, “Cognitive Radios with Genetic Algorithms: Intelligent Control of Software Defined Radios,” Proc. Software Defined Radio Forum Technical Conf. (SDR ’04), pp. C3-C8, 2004.
[16] Q. Zhao, L. Tong, and A. Swami, “A Cross-Layer Approach to Cognitive MAC for Spectrum Agility,” Proc. 39th IEEE Asilomar Conf. Signals, Systems, and Computers (ASILOMAR ’05), pp. 200-204, Nov. 2005.
[17] S. Krishnamurthy, M. Thoppian, S. Venkatesan, and R. Prakash, “Control Channel Based MAC-Layer Configuration, Routing and Situation Awareness for Radio Networks,” Proc. IEEE Military Comm. Conf. (MILCOM ’05), pp. 1-6, Oct. 2005.
[18] C. Doerr, M. Neufeld, J. Fifield, T. Weingart, D.C. Sicker, and D. Grunwald, “MultiMAC—An Adaptive MAC Framework for Dynamic Radio Networking,” Proc. First IEEE Int’l Symp. New Frontiers in Dynamic Spectrum Access Networks (DySPAN ’05), pp. 548-555, Nov. 2005.
[19] S. Sankaranarayanan, P. Papadimitratos, A. Mishra, and S. Hershey, “A Bandwidth Sharing Approach to Improve Licensed Spectrum Utilization,” Proc. First IEEE Int’l Symp. New Frontiers in Dynamic Spectrum Access Networks (DySPAN ’05), pp. 279-288, Nov. 2005. [20] Y. Xing and R. Chandramouli, “Dynamic Spectrum Access in
Open Spectrum Wireless Networks,” IEEE J. Selected Areas in Comm., vol. 24, pp. 626-637, Mar. 2006.
[21] H. Wu, C. Qiao, S. De, and O. Tonguz, “Integrated Cellular and Ad Hoc Relaying Systems: iCAR,” IEEE J. Selected Areas in Comm., vol. 19, no. 10, pp. 2105-2115, Oct. 2001.
[22] E. Yanmaz, O.K. Tonguz, S. Mishra, H. Wu, and C. Qiao, “Efficient Dynamic Load Balancing Algorithms Using iCAR Systems: A Generalized Framework,” Proc. 56th IEEE Vehicular Technology Conf. (VTC ’02), vol. 1, pp. 586-590, Sept. 2002.
[23] E.H.-K. Wu, Y.-Z. Huang, and J.-H. Chiang, “Dynamic Adaptive Routing for Heterogeneous Wireless Network,” Proc. IEEE Global Telecomm. Conf. (GLOBECOM ’01), vol. 6, pp. 3608-3612, Nov. 2001.
Fig. 14. Definitions of parameters in the calculation of the area of section A3 in (18) in the infrastructure downlink case when
[24] Y.D. Lin and Y.C. Hsu, “Multihop Cellular: A New Architecture for Wireless Communications,” Proc. IEEE INFOCOM, vol. 3, pp. 26-30, Mar. 2000.
[25] J. Chen, S.-H.G. Chan, J. He, and S.C. Liew, “Mixed-Mode WLAN: The Integration of Ad-Hoc Mode with Wireless LAN Infrastructure,” Proc. IEEE Global Telecomm. Conf. (GLOBECOM ’03), vol. 1, pp. 231-235, Dec. 2003.
[26] D. Niculescu and B. Nath, “Ad Hoc Positioning System (APS),” Proc. IEEE INFOCOM, vol. 3, pp. 1734-1743, Mar. 2001.
[27] J. Hightower and G. Borriello, “Location Systems for Ubiquitous Computing,” Computer, vol. 34, no. 8, pp. 57-66, Aug. 2001. [28] D. Niculescu and B. Nath, “Ad Hoc Positioning System (APS)
Using AOA,” Proc. IEEE INFOCOM, vol. 3, pp. 1734-1743, Mar. 2003.
[29] S.J. Ingram, D. Harmer, and M. Quinlan, “Ultrawideband Indoor Positioning Systems and Their Use in Emergencies,” Proc. Position Location and Navigation Symp. (PLANS ’04), pp. 706-715, Apr. 2004. [30] D. Madigan, E. Einahrawy, R.P. Martin, W.-H. Ju, P. Krishnan, and A.S. Krishnakumar, “Bayesian Indoor Positioning Systems,” Proc. IEEE INFOCOM, vol. 2, pp. 1217-1227, Mar. 2005.
[31] G. Sun, J. Chen, W. Guo, and K.-J.R. Liu, “Signal Processing Techniques in Network-Aided Positioning: A Survey of State-of-the-Art Positioning Designs,” IEEE Signal Processing Magazine, vol. 22, no. 4, pp. 12-23, July 2005.
[32] Y.-C. Tseng, S.-L. Wu, W.-H. Liao, and C.-M. Chao, “Location Awareness in Ad Hoc Wireless Mobile Networks,” Computer, vol. 34, no. 6, pp. 46-52, June 2001.
[33] M. Mauve, A. Widmer, and H. Hartenstein, “A Survey on Position-Based Routing in Mobile Ad Hoc Networks,” IEEE Network, vol. 15, no. 6, pp. 30-39, Nov./Dec. 2001.
[34] X. Hong, K. Xu, and M. Gerla, “Scalable Routing Protocols for Mobile Ad Hoc Networks,” IEEE Network, vol. 16, no. 4, pp. 11-21, July/Aug. 2002.
[35] T. Park and K.G. Shin, “Optimal Tradeoffs for Location-Based Routing in Large-Scale Ad Hoc Networks,” IEEE/ACM Trans. Networking, vol. 13, no. 2, pp. 398-410, Apr. 2005.
[36] H. Celebi and H. Arslan, “Adaptive Positioning Systems for Cognitive Radios,” Proc. IEEE Int’l Symp. New Frontiers in Dynamic Spectrum Access Networks (DySPAN ’07), pp. 78-84, Apr. 2007. [37] T.S. Rappaport, Wireless Communications: Principle and Practice,
second ed. Prentice Hall, 2002.
[38] G.L. Stu¨ber, Principle of Mobile Communication, second ed. Kluwer Academic, 2001.
[39] G. Bianchi, “Performance Analysis of IEEE 802.11 Distributed Coordination Function,” IEEE J. Selected Areas in Comm., vol. 18, no. 3, pp. 535-547, Mar. 2000.
[40] L.C. Wang, S.Y. Huang, and A. Chen, “A Cross-Layer Throughput Performance Investigation for CSMA/CA-Based Wireless Local Area Network with Directional Antennas and Capture Effect,” Proc. IEEE Wireless Comm. and Networking Conf. (WCNC ’04), vol. 3, pp. 1879-1884, Mar. 2004.
Li-Chun Wang received the BS degree in electrical engineering from the National Chiao-Tung University, Hsinchu, Taiwan, in 1986, the MS degree in electrical engineering from the National Taiwan University, Taipei, in 1988, and the MSc and PhD degrees in electrical engineer-ing from Georgia Institute of Technology, Atlanta, in 1995 and 1996, respectively. From 1990 to 1992, he was with Chunghwa Telecom. In 1995, he was with Northern Telecom, Richardson, Texas. From 1996 to 2000, he was with AT&T Laboratories, where he was a senior technical staff member in the Wireless Communications Research Department. Since August 2000, he has been an associate professor in the Department of Communication Engineering, National Chiao-Tung University, where he has been a full professor since August 2005. He was a corecipient of the Jack Neubauer Best Paper Award from the IEEE Vehicular Technology Society in 1997. His current research interests are in the areas of cellular architectures, radio network resource management, and cross-layer optimization for cooperative and cognitive wireless networks. He is the holder of three US patents with three more pending. He is a senior member of the IEEE.
Anderson Chen received the BS, MS, and PhD degrees in communication engineering from the National Chiao-Tung University, Hsinchu, Taiwan, in 1998, 1999, and 2007, respectively. He is currently working on his national service. His research interests include wireless net-works, cross-layer design, and cognitive wire-less system. He is the holder of a US patent with three more pending. He is a student member of the IEEE.
. For more information on this or any other computing topic, please visit our Digital Library at www.computer.org/publications/dlib.
