2012 zou-yao-zheng-cmag-98

IEEE Communications Magazine • April 201298 0163-6804/12/$25.00 © 2012 IEEE

INTRODUCTION

Cognitive radio is emerging as a promising tech-nology for future radio networks [1, 2], whichenables an unlicensed cognitive user to recog-nize (by spectrum sensing) and access a spec-trum hole that is a radio frequency band licensedto a primary user but not used by that user. In[3], code-division multiple access (CDMA) has

been investigated for spectrum sharing betweentwo heterogeneous systems (i.e., a cellular net-work and a microwave network) through inter-ference suppression. Differing from [3], cognitiveradio typically exploits the spectrum sensing todetect spectrum holes for secondary transmis-sions so that the primary and secondary userscan share spectrum resources without harmfulinterference. In general, ideal spectrum sensing(without miss detection and false alarm) cannotbe realized in practice and thus mutual interfer-ence between the primary and cognitive usersexists, which severely impairs and limits the sec-ondary transmission performance under a targetprimary quality of service (QoS) requirement,especially with a strict primary QoS constraint[4]. In order to achieve a reliable cognitive radiosystem with high data rates, it requires both arobust spectrum sensing algorithm and an effi-cient secondary transmission mechanism so as todetect and utilize the spectrum holes in the mostefficient way. To that end, cooperative relaytechnology is considered as an effective methodto improve the performance of both spectrumsensing and secondary transmissions.

The spectrum sensing with cooperative relays(also known as cooperative sensing) has beenstudied to combat wireless fading effects [5–8].As is known, each cooperative sensing processrequires two essential phases: the phase of theprimary user’s signal detection by cognitive usersand the phase of the initial detection resultreporting to a fusion center, referred to as detec-tion and reporting phases, respectively. In [5]and [6], a dedicated channel is considered forreporting the initial detection results from cogni-tive users to the fusion center in order to avoidinterfering with the primary user. However, thisapproach requires extra spectrum resources andinvolves complex networking/signaling issue dueto the dedicated channel resource management.

ABSTRACT

Cognitive radio is a promising technologythat enables an unlicensed user (also known as acognitive user) to identify the white space of alicensed spectrum band (called a spectrum hole)and utilize the detected spectrum hole for itsdata transmissions. To design a reliable and effi-cient cognitive radio system, there are two fun-damental issues: to devise an accurate and robustspectrum sensing algorithm to detect spectrumholes as accurately as possible; and to design asecondary user transmission mechanism for thecognitive user to utilize the detected spectrumholes as efficiently as possible. This article inves-tigates and shows that cooperative relay technol-ogy can significantly benefit the above-mentioned two issues, spectrum sensing and sec-ondary transmissions. We summarize existingresearch about the application of cooperativerelays for spectrum sensing (referred to as thecooperative sensing) and address the relatedpotential challenges. We discuss the use of coop-erative relays for the secondary transmissionswith a primary user’s quality-of-service (QoS)constraint, for which a diversity-multiplexingtrade-off is developed. In addition, this articleshows a trade-off design of cognitive transmis-sions with cooperative relays by jointly consider-ing the spectrum sensing and secondarytransmissions in cognitive radio networks.

ADVANCES IN COOPERATIVE WIRELESSNETWORKING: PART 2

Yulong Zou, Nanjing University of Posts and Telecommunications, Stevens Institute of Technology

Yu-Dong Yao, Stevens Institute of Technology

Baoyu Zheng, Nanjing University of Posts and Telecommunications

Cooperative Relay Techniques forCognitive Radio Systems: Spectrum Sensing andSecondary User Transmissions

YAO LAYOUT_Layout 1 3/22/12 3:20 PM Page 98

IEEE Communications Magazine • April 2012 99

Therefore, in [7, 8], we study a selective-relay-based cooperative sensing (SRCS) scheme with-out a dedicated reporting channel and examineits receiver operating characteristics (ROC) per-formance by jointly considering the detectionand reporting phases. It is shown in [7] that theSRCS scheme can save dedicated reportingchannel resources without sacrificing ROC per-formance as compared to the conventionalapproach [5]. Moreover, we show that the coop-erative sensing performance can be further opti-mized through a trade-off design in determiningthe time durations for the detection and report-ing phases.

Cooperative relay technology has been stud-ied extensively for traditional wireless networks[9]. It, however, faces an additional challengingissue in cognitive radio networks, i.e., mutualinterference between the primary and cognitiveusers. Specifically, although a spectrum sensingmodule is utilized in cognitive radio, it is notpractical to achieve ideal spectrum sensing with-out any miss detection and false alarm. Thiswould introduce mutual interference betweenthe primary and cognitive users. In [4], we studythe use of cooperative relays for secondary trans-missions and illustrate that, given a primary out-age probability requirement, the mutualinterference causes an outage probability floorfor secondary transmissions in high signal-to-noise ratio (SNR) regions. Using the outageprobability floor as a performance metric, wegeneralize the traditional diversity gain defini-tion and show that the full diversity is stillachieved for secondary transmissions with a tar-get primary outage probability requirement. Inaddition, we provide a generalized multiplexinggain definition in [10] and develop the diversity-multiplexing trade-off of secondary transmissionswith a primary QoS constraint.

A reliable cognitive radio system with highdata rates is achievable by using cooperativerelays for both the spectrum sensing and sec-ondary transmissions. However, the two individ-ual designs of spectrum sensing and secondarytransmissions cannot be optimized separately,since they affect each other [11, 12]. For exam-ple, when an available spectrum hole is notdetected by spectrum sensing during a certainobservation window, the spectrum hole utiliza-tion would decrease. To alleviate this issue, wemay increase the observation time for the spec-trum sensing phase, which, however, comes atthe cost of degradation in secondary transmis-sion performance since less time is now avail-able for the secondary transmission phase. In[12], we investigate the sensing and transmis-sion trade-off in a multiple-relay cognitive radionetwork and show that a significant perfor-mance improvement is obtained by using coop-erative relays in terms of the spectrum holeutilization.

The rest of this article is organized as follows.We describe related cooperative sensing workwith a dedicated channel for reporting initialdetection results to a fusion center. We also pro-pose an alternative SRCS framework without thededicated reporting channel and show its ROCperformance comparison with the conventionalcooperative sensing (with a dedicated reporting

channel). We investigate the use of cooperativerelays for secondary transmissions with a prima-ry QoS constraint and illustrate the diversity-multiplexing trade-off. Next, we provide atrade-off design between the spectrum sensingand secondary transmissions in multiple-relaycognitive radio networks. Finally, we make con-cluding remarks.

COOPERATIVE RELAYS FORSPECTRUM SENSING

Spectrum sensing is regarded as a mandatoryfeature in cognitive radio networks [2], for whichthree typical signal detection approaches areavailable: energy detection, matched filter detec-tion, and feature detection. These detectorswork well in Gaussian noise scenarios, which,however, are not appropriate to be directly uti-lized in wireless fading environments. To thatend, cooperative sensing strategies have beenstudied to combat the wireless fading in [5] and[6], where multiple cognitive users (CUs) inde-pendently detect the licensed primary channelusing a certain detector (e.g., energy detector)and report their initial detection results to afusion center (FC).

Typically, each cooperative sensing processconsists of two essential phases:• Detection phase, in which all CUs attempt

to detect the presence of a primary user(PU) within a certain time duration (calledsignal detection overhead)

• Reporting phase, in which the initial detec-tion results are forwarded to the FC overthe remaining time

A joint detection and reporting analysis is essen-tial to optimize the cooperative sensing perfor-mance, since the two phases affect each otherand cannot be optimized in isolation. Althoughthe initial detection results contain only a fewinformation bits in an information-theoreticalsense, CUs scan the licensed channel periodical-ly and have a non-negligible transmission rate ofinitial detection results, which would lead totransmission errors in the reporting phase con-sidering the noise and fading effects. This mayaffect the final fusion result at the FC anddegrade the overall spectrum sensing perfor-mance. While increasing the time period for thereporting phase can improve the reliability ofthe initial detection result transmission, it comesat the expense of a reduction in individualdetection performance since less time is nowavailable for the detection phase. It would alsodegrade the overall spectrum sensing perfor-mance at the FC.

In addition, when CUs forward their initialdetection results to the FC in the reportingphase, they may interfere with the primary user.To avoid such interference, previous research[5, 6] assumed that there is a dedicated channelfor reporting the initial detection results, which,however, requires extra channel resources. Tothat end, we propose a selective-relay basedcooperative sensing (SRCS) scheme in [7] and[8] to mitigate the interference from CUs to PUin the reporting phase without a dedicated chan-nel, where each CU forwards its initial detection

A reliable cognitive

radio system with

high data rates is

achievable by using

cooperative relays for

both the spectrum

sensing and sec-

ondary transmissions.

However, the two

individual designs of

spectrum sensing

and secondary trans-

missions cannot be

optimized separately,

since they affect

each other.


IEEE Communications Magazine • April 2012100

result in a selective manner depending on if theabsence of PU is detected or not. Specifically, ifthe ith CU (denoted CUi) detects the absenceof PU in its detection phase, it transmits a cyclicredundancy check (CRC)-encoded indicator sig-nal to the FC over the ith orthogonal subchan-nel of the primary channel, instead of adedicated channel. Otherwise, nothing is trans-mitted from CUi to avoid interfering with PU.At the FC side, if the CRC checking is success-ful over i-th orthogonal sub-channel, FC consid-ers the absence of PU as the initial resultdetected by CU i; otherwise, it considers thepresence of PU as the CUi’s initial detectionresult. Accordingly, in the SRCS scheme, a CUmay interfere with PU only if it fails to detectthe presence of PU given that PU is active. Asdiscussed in [7], this interference is controllableand can be reduced to satisfy a given primaryQoS requirement.

In Fig. 1, we show the ROC performancecurves of the traditional cooperative sensing [5]and the SRCS [7] schemes with a primary sig-nal-to-noise ratio of 5dB considering a logic“AND” fusion rule, where the three ROC curvepairs correspond to the different number ofCUs, M = 1, 2 and 3, respectively. From Fig. 1,one can observe that in the low detection proba-bility regions, the false alarm probabilities ofthe SRCS scheme are larger than these of thetraditional scheme. However, in higher detec-tion probability regions, it is shown from Fig. 1that ROC performance of the SRCS scheme isnearly identical to that of the traditional scheme.Notice that practical cognitive radio systemsrequire the detection probability to be above arelatively large value (e.g., at least 0.9 [2]) forthe primary user protection. In this sense, theSRCS scheme saves the dedicated reporting

channel resources without sacrificing ROC per-formance as compared to the traditional coop-erative sensing.

Figure 2 illustrates the false alarm probabil-ity versus signal detection overhead of theSRCS scheme for different number of CUs, M= 1, 2, and 3. From Fig. 2, one can see that anoptimal signal detection overhead exists tominimize the false alarm probability under atarget detection probability 0.99, i.e., a mini-mum false alarm probability can be achievedthrough an optimal allocation of the time dura-tions between the detection and reportingphases. As shown in Fig. 2, the optimal signaldetection overhead value decreases with anincreasing number of CUs. This is due to thefact that, as the number of CUs increases, eachCU is allocated with less bandwidth resourcesfor its initial detection result reporting andthus a longer time duration is needed in thereporting phase, resulting in the decrease ofoptimal signal detection overhead.

COOPERATIVE RELAYS FORSECONDARY TRANSMISSIONS WITH

A PRIMARY QOS CONSTRAINT

In this section, we discuss the application ofcooperative relays for secondary data transmis-sions with a primary QoS constraint. As men-tioned, ideal spectrum sensing without any missdetection and false alarm is not achievable inpractice. This introduces the mutual interferencebetween the primary and cognitive users, result-ing in performance floor (e.g., outage probabilityfloor) for the secondary transmissions in highSNR regions with a primary QoS constraint. Inthe following, we show that cooperative relayscan significantly improve the secondary transmis-sion performance without affecting the primaryQoS.

Figure 3 shows the coexistence of a primarynetwork with a cognitive radio network, where aprimary source (PS) may send data to a primarydestination (PD). Meanwhile, in the cognitiveradio network, a cognitive source (CS) firstdetects the activity (presence or absence) of PSin search of a spectrum hole. Given a spectrumhole detected, CS starts its data transmissions toa cognitive destination (CD). We consider thatmultiple cognitive relays (CRs) are available toassist the secondary data transmissions. General-ly speaking, there are two basic relaying proto-cols, amplify-and-forward (AF) anddecode-and-forward (DF), and the latter proto-col is considered in this article.

Given multiple CRs available to assist a CS’sdata transmissions, a straightforward way is toallow all the relays to participate in forwardingthe CS’s signal to the CD. Although thisapproach can achieve full diversity gain, it comesat the expense of a reduction in multiplexinggain, since the multiple relays shall transmit onorthogonal channels to avoid interfering witheach other. To overcome this issue, an alterna-tive protocol, called the best relay selection, hasbeen studied in cognitive radio networks [4],where only the “best” cognitive relay is selected

Figure 1. The false alarm probability vs. detection probability of the traditionalcooperative sensing and the SRCS schemes for different number of CUs.

Detection probability

The number ofCUs, M = 3, 2, 1

Practical ROC regions(e.g., Pd > 0.9 [2])

0.10 1

10-8

10-10

Fals

e al

arm

pro

babi

lity

10-6

10-4

10-2

100

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

SRCS schemeTraditional scheme



to forward the CS’s signal; thus, only two chan-nels (i.e., the best relay link and direct link) arerequired regardless of the number of cognitiverelays. It has been shown that the best-relayselection protocol can still achieve the full diver-sity in cognitive radio with a primary QoS con-straint [4].

Although the best relay transmission achievesthe full diversity, one-half of the multiplexinggain is sacrificed since two orthogonal channelsare required. In contrast, the non-relay directtransmission inherits the full multiplexing butwithout any diversity gain. In [10], we explore aselective best relay cooperation framework byjointly considering the best-relay transmissionmode and non-relay direct transmission mode.We examine two specific selective best relaycooperation schemes with and without anacknowledgment (ACK) from the cognitive des-tination with regard to success or not in decod-ing, which are called ACK- and non-ACK-basedselective best relay cooperation, respectively.

For the non-ACK-based selective best relaycooperation, CD does not acknowledge whetheror not it decodes successfully, and only the CRs’outcomes of decoding CS’s signal are consideredin determining a selective relaying strategy.Specifically, CS first broadcasts a data messageto CD and all CRs, and those CRs that can cor-rectly decode CS’s signal constitute a so-calleddecoding set. Then, if the decoding set is notempty, the best cognitive relay among the decod-ing set is chosen to forward its decoded result toCD. If the decoding set is empty, let CS transmitits subsequent new message which is after themessage transmitted earlier.

Differing from the non-ACK scheme, ACK-based selective best relay cooperation allows CDto transmit an ACK to inform CS and all CRswhether it successfully decodes CS’s signal ornot. To be specific, if CD succeeds in decodingits received signal directly from CS, it broadcastsan ACK signal to CS and CRs so that a newdata message would be scheduled for transmis-sion at CS, instead of repeating the previousmessage by a cognitive relay. Otherwise, in thecase of CD failing to decode CS’s signal, theremaining process of the ACK-based selectivebest relay cooperation is same as that of thenon-ACK-based scheme.

Figure 4 compares the diversity-multiplexingtrade-offs of non-cooperation, and non-ACKand ACK-based selective best relay cooperationschemes, where M represents the number ofCRs. One can observe from Fig. 4 that, as themultiplexing gain approaches zero, the maximumdiversity gains of the non-ACK and ACK-basedselective cooperation schemes with M = 1, 2,and 4 are equal to 2, 3, and 5, respectively, show-ing the full diversity achieved by both the non-ACK and ACK schemes. On the other hand, asthe diversity gain decreases to zero, a multiplex-ing gain of only 1/2 is achieved by the non-ACK-based cooperation scheme regardless of thenumber of CRs. For the ACK-based cooperationscheme, as the number of CRs increases from M= 1 to 4, the maximum multiplexing gainsdecrease from 2/3 to 5/9, which are always largerthan 1/2, showing its advantage over the non-ACK-based scheme [10].

TRADE-OFF BETWEEN SPECTRUMSENSING AND SECONDARY

TRANSMISSIONSIn this section, we study a trade-off design of thecognitive transmissions by jointly considering thespectrum sensing and secondary transmissions.As discussed in [8] and [12], the two individualphases cannot be optimized separately, sincethey affect each other. In [12], we have exploredthe cognitive transmissions with multiple relaysthat are utilized for both the spectrum sensingand secondary transmissions. As shown in Fig. 5,each cognitive transmission process consists oftwo phases (i.e., the spectrum sensing and sec-ondary transmission phases), where the parame-ter α is referred to as spectrum sensingoverhead, which is to be adjusted to optimizethe cognitive transmission performance [12].

Figure 5 shows that, in the spectrum sensingphase, there are two subphases: the primary userdetection and initial detection result reporting

Figure 2. The false alarm probability vs. signal detection overhead of the SRCSscheme for different number of CUs, M, with a target detection probability of0.99.

Signal detection overhead0.10

10-4

Fals

e al

arm

pro

babi

lity

10-5

10-3

10-2

10-1

100

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

M = 1M = 2M = 3

Figure 3. A cognitive radio network with multiple cognitive relays.

PS PD

CD

... ...

PS: Primary sourcePD: Primary destinationCS: Cognitive sourceCR: Cognitive relayCD: Cognitive destination

CS

Primarynetwork

Cognitiveradio

networkCR1 CRi

CRM


IEEE Communications Magazine • April 2012102

phases. To be specific, in the detection sub-phase, a cognitive source and all cognitive relaysindependently detect the presence or absence ofthe primary user. In the subsequent reportingsubphase, all CRs forward their initial detectionresults to CS for fusion. Here, two fusion strate-gies are considered: fixed fusion and selectivefusion, which lead to fixed fusion spectrum sens-ing (FFSS) and selective fusion spectrum sensing(SFSS) schemes, respectively. In the FFSSscheme, CRs transmit their initial detectionresults to CS without using any forward errordetection code, and CS decodes the received ini-tial detection results from CRs and combines allthe decoded outcomes. In contrast, the SFSSscheme employs an additional forward errordetection code (such as cyclic redundancy checkcode) at CRs to encode their initial detectionresults. The encoded signals are transmitted toCS, which decodes the received signals and com-bines the successfully decoded outcomes only;that is, only the successfully received initialdetection results are selected and used forfusion.

As shown in Fig. 5, the secondary transmis-sion phase also consists of two subphases. Givena spectrum hole detected earlier in the spectrumsensing phase, CS starts transmitting its data toCD and CRs in the first secondary transmission

subphase. Then, in the subsequent sub-hase, abest relay data transmission (BRDT) strategy isused for the secondary transmission. Specifically,all CRs attempt to decode their received signals(from CS), and those CRs that decode success-fully constitute a decoding set. If the decodingset is not empty, we choose the best relay withinthe decoding set to forward its decoded result toCD. If the decoding set is empty (i.e., no relay isable to decode the CS’s signal successfully), letCS repeat the transmission of the original signalto CD through its direct link. Finally, CD com-bines the two copies of the received signals usinga certain combining method and gives an estima-tion of the original signal.

In Fig. 6, we show the spectrum hole utiliza-tion vs. spectrum sensing overhead of the FFSS-BRDT and SFSS-BRDT schemes for differentdata rates R, R = 1 b/s/Hz and 1.5 b/s/Hz. Notethat the spectrum hole utilization is used toquantify the percentage of spectrum holes uti-lized for secondary data transmissions withoutoutage events. Figure 6 illustrates that an opti-mal spectrum sensing overhead exists to maxi-mize the spectrum hole utilization for both theFFSS-BRDT and SFSS-BRDT schemes. Hence,a joint analysis of the spectrum sensing and sec-ondary transmission phases is essential to opti-mize the overall performance of cognitive radiotransmissions. One can also observe from Fig. 6that, no matter which scheme (FFSS-BRDT orSFSS-BRDT) is used, the optimal spectrumsensing overhead corresponding to R = 1.5b/s/Hz is smaller than that for R = 1 b/s/Hz. Thisis because, as the data rate increases, the sec-ondary transmission phase should be allocated arelatively longer time duration, which results inless time available for the spectrum sensingphase. In addition, it is shown from Fig. 6 thatthe spectrum hole utilization of the SFSS-BRDTscheme is always higher than that of the FFSS-BRDT scheme across the whole spectrum sens-ing overhead regions.

CONCLUSIONS

This article describes the application of coopera-tive relays for both the spectrum sensing andsecondary transmissions to achieve a reliable andefficient cognitive radio system. We present aselective-relay-based cooperative sensing schemewithout a dedicated channel to report initialdetection results for fusion and show its advan-tage over traditional cooperative sensing (with adedicated reporting channel). We also discussthe use of cooperative relays for the secondarytransmissions with a primary QoS constraint andprovide a diversity-multiplexing trade-off evalua-tion. In addition, it is shown that the cognitiveradio transmission performance can be opti-mized in terms of the spectrum hole utilizationthrough a joint analysis of the spectrum sensingand secondary transmissions.

REFERENCES[1] J. Mitola and G. Q. Maguire, “Cognitive Radio: Making

Software Radios More Personal,” IEEE Pers. Commun.,vol. 6, 1999, pp. 13–18.

[2] IEEE 802.22 Working Group, “IEEE P802.22/D1.0 Draft

Figure 4. Diversity-multiplexing trade-offs of the non-cooperative transmission,and the non-ACK and ACK-based selective best relay cooperation schemeswith different numbers of cognitive relays.

Multiplexing gain0.10

0.5

0

Div

ersi

ty g

ain

1

1.5

2

2.5

3

3.5

4

4.5

5

0.2 0.3 0.4 0.5 0.6

M = 1

0.7 0.8 0.9 1

ACK based cooperationNon-ACK based cooperationNon-cooperative direct transmission

M = 2

M = 4

Figure 5. Time slot structure of the cognitive transmissions with multiple relays.

t1/2

Best CRforwards

or CS repeatsCS and CRs

detect

Spectrum sensing phase Secondary transmission phase

CS transmits ...

CR1→CS

CRM→CS

1/2 1/21-α

1/2

Time slot (k)

α



Standard for Wireless Regional Area Networks Part 22:Cognitive Wireless RAN Medium Access Control (MAC)and Physical Layer (PHY) Specifications: Policies andProcedures for Operation in the TV Bands,” Apr. 2008.

[3] J. Wang, “CDMA Overlay Situations for MicrocellularMobile Communications,” IEEE Trans. Commun., vol.43, no. 234, Feb. 1995, pp. 603–14.

[4] Y. Zou et al., “An Adaptive Cooperation DiversityScheme with Best-Relay Selection in Cognitive RadioNetworks,” IEEE Trans. Signal Proc., vol. 58, no. 10,Oct. 2010, pp. 5438–45.

[5] Ghasemi and E. S. Sousa, “Collaborative Spectrum Sens-ing for Opportunistic Access in Fading Environment,”Proc. IEEE DySPAN 2005, pp. 131–36.

[6] J. Ma, G. Zhao, and Y. Li, “Soft Combination and Detec-tion for Cooperative Spectrum Sensing in CognitiveRadio networks,” IEEE Trans. Wireless Commun., vol. 7,no. 11, Nov. 2008, pp. 4502–07.

[7] Y. Zou, Y.-D. Yao, and B. Zheng. “A Selective-RelayBased Cooperative Spectrum Sensing Scheme WithoutDedicated Relay Channels in Cognitive Radio Net-works,” IEEE Trans. Wireless Commun., vol. 10, no. 4,Apr. 2011, pp. 1188–98.

[8] Y. Zou, Y.-D. Yao, and B. Zheng. “A Cooperative Sens-ing based Cognitive Relay Transmission Scheme With-out A Dedicated Sensing Relay Channel in CognitiveRadio Networks,” IEEE Trans. Signal Proc., vol. 59, no.2, Feb. 2011, pp. 854–58.

[9] J. N. Laneman et al., “Cooperative Diversity in WirelessNetworks: Efficient Protocols and Outage Behavior,”IEEE Trans. Info. Theory, vol. 50, no. 12, Dec. 2004, pp.3062–80.

[10] Y. Zou, Y.-D. Yao, and B. Zheng, “Diversity-Multiplex-ing Trade-Off in Selective Cooperation for CognitiveRadio,” IEEE Trans. Commun., under 2nd round review.

[11] A. Ghasemi and E. S. Sousa, “Optimization of Spec-trum Sensing for Opportunistic Spectrum Access inCognitive Radio Networks,” Proc. IEEE CCNC ’07, pp.1022–26.

[12] Y. Zou, Y.-D. Yao, and B. Zheng, “Cognitive Transmis-sions with Multiple Relays in Cognitive Radio Net-works,” IEEE Trans. Wireless Commun., vol. 10, no. 2,Feb. 2011, pp. 648–59.

BIOGRAPHIESYULONG ZOU received his B.Eng. degree in information engi-neering from Nanjing University of Posts and Telecommuni-cations (NUPT), China, in 2006. He is currently workingtoward his dual Ph.D. degree at the Institute of Signal Pro-cessing and Transmission of NUPT, and the Electrical andComputer Engineering Department of Stevens Institute ofTechnology (SIT), New Jersey. His research interests span awide range of topics in wireless communications and sig-nal processing, including cooperative communications,space-time coding, network coding, and cognitive radio.Recently, he has been working on cooperative relay tech-niques in cognitive radio networks, opportunistic distribut-ed space-time coding in cooperative wireless networks, andfull-diversity high-rate network coding for cellular systems(e.g., LTE/IMT-advanced and beyond).

YU-DONG YAO [S’88, M’88, SM’94, F’11] ([email protected]) has been with Stevens Institute ofTechnology, Hoboken, New Jersey, since 2000, and is cur-rently a professor and department director of electrical andcomputer engineering. He is also a director of Stevens’Wireless Information Systems Engineering Laboratory(WISELAB). Previously, from 1989 and 1990, he was at Car-leton University, Ottawa, Canada, as a research associateworking on mobile radio communications. From 1990 to1994, he was with Spar Aerospace Ltd., Montreal, Canada,where he was involved in research on satellite communica-tions. From 1994 to 2000, he was with Qualcomm Inc.,San Diego, California, where he participated in researchand development in wireless CDMA systems.

BAOYU ZHENG received B.S. and M.S. degrees from theDepartment of Circuit and Signal System, NUPT, in 1969and 1981, respectively. Since then, he has been engaged inteaching and researching signal and information process-ing. He is a full professor and doctoral advisor at NUPT. Hisresearch interests span the broad area of the intelligentsignal processing, wireless network and signal processingfor modern communication, and the quantum signal pro-cessing.

Figure 6. The spectrum hole utilization vs. spectrum sensing overhead of theFFSS-BRDT and SFSS-BRDT schemes with the number of cognitive relaysM = 4.

Spectrum sensing overhead0

0.1

0

Spec

trum

hol

e ut

iliza

tion

0.2

0.3

0.4

0.5

0.6

0.8

0.7

0.9

1

0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

R=1 bit/s/HzR=1.5 bit/s/Hz

0.9 1

SFSS-BRDT schemeFFSS-BRDT scheme


2012 zou-yao-zheng-cmag-98

Documents