A Primary User Emulation Attack Countermeasure Strategy and Energy-Efficiency Analysis in Cognitive Radio Networks

Transcription

1 Journal o Communications Vol., No., January 7 A imary User Emulation Attack Countermeasure Strategy an Energy-Eiciency Analysis in Cognitive Raio Networks unchuan Wang, Xiaorong Xu, Weiwei Wu, an Jianrong Bao College o Telecommunication Engineering, angzhou Dianzi University, angzhou, Zheiang 38, China waynetlg@oxmail.com; xuxr@hu.eu.cn; @63.com; baor@hu.eu.cn Abstract Due to the act that security an energy eiciency are investigate separately in the conventional imary User Emulation Attack (PUEA) countermeasure strategy in Cooperative Spectrum Sensing (CSS) process, the traeo between security an energy eiciency is iscusse in this paper. CSS can improve etection perormance whereas energy consumption will ramatically boost up with the increasing o cooperative users. To tackle this problem, we set CSS energy eiciency in the presence o PUEA as optimal obective, whereas secure etection perormance an secure alse alarm threshol as the constraint conitions. The optimal problem is resolve to achieve a trae-o between energy eiciency an security in the presence o PUEA attack. The optimal usion threshol K an optimal cooperative user numbers can be obtaine in secure CSS with energy eiciency maximization, which guarantees energy eiciency as well as secure CSS etection perormance. The propose PUEA countermeasure strategy etection perormance is presente an compare with traitional Maximum Ratio Combining (MRC) usion rule. Simulation results show that, the propose strategy is immune to PUE intererence power, an it provies high robustness to PUEA attack, which implements the trae-o between security an energy eiciency in Cognitive Raio Network (CRN) eectively. by overlay or unerlay approaches with the premise o not aecting the authorize imary User (PU) normal communications. CR technology has greatly reuce the constraints o spectrum an banwith or the evelopment o wireless technology. It has become one o the potential key technologies in the 5G stanars [], []. CRN is one o the key technologies o 5G [3], owever, CRN is an open network environment, which is particularly vulnerable to malicious user attack. CRN not only aces the conventional wireless network security threats, such as eavesropping attack, tampering ata, etc [4], [5], but also conronts with some new threats such as PUEA [5] an Spectrum Sensing Data Falsiication (SSDF) [6]. The traitional wireless network security problems coul be mainly solve by the encryption o transmitting signals [3], [4], [7]. owever, energy consumption was signiicantly increase in secure transmission ue to the complexity o encryption algorithm an key management mechanism. ence, physical layer (P) security was regare as complement or substitution o conventional encryption techniques, which ha rawn much attention by network security researchers [8]-[]. The basic iea o P security is to make use o the ranom eature o the noisy channel to ensure that malicious users can not obtain transmit message rom the channel. In other wors, P security uses the inherent uncertainty o the noisy channel, in orer to ensure CRN ata security. P security consiere the security rom inormation theory an signal processing perspectives [8], [9]. P security containe two aspects. On the one han, CRN security capacity was theoretically analyze rom the inormation theory perspective. On the other han, secure transmit rates to achieve security capacity was investigate rom signal processing an optimization perspective [9], []. owever, CRN requires more on energy eiciency an system security in the pursuit o high spectrum utilization an high transmission eiciency. Recent research showe that, energy consume in spectrum sensing phase an ata transmission phase were the main energy cost in CRN []. With the increasing o SU ensity an the expansion o network coverage area, CRN energy consumption problem has rawn much attention. Namely, Green Communications has become one o the Inex Terms Cooperative Spectrum Sensing (CSS), imary User Emulation Attack (PUEA), countermeasure strategy, energy eiciency, etection perormance I. INTRODUCTION Cognitive Raio (CR) is a promising technique that enables a smart exploitation o the unuse portions o the license spectrum. In Cognitive Raio Networks (CRN), Seconary User (SU) can aaptively aust the operational parameters (such as the transmit power, carrier requency an moulation metho) via sensing the external wireless environments. SU can access to the authorize spectrum to realize ynamic spectrum sharing Manuscript receive August 3, 6; revise January, 7. This work was supporte in part by the ogram o Zheiang ovincial Natural Science Founation o China (Grant No. L5F8, LZ4F3), National Natural Science Founation o China (Grant No. 6475), the Grauate Scientiic Research Founation o angzhou Dianzi University in 6 (ZX663835) an oung Talent Cultivation oect o Zheiang Association or Science an Technology (Grant No. 6CGC9). Corresponing author xuxr@hu.eu.cn. oi:.7/cm Journal o Communications

2 Journal o Communications Vol., No., January 7 evelopment trens or CRN in the uture. Green CRN is the research hotspot in CR at present. It is also one o the key techniques to realize uture energy eiciency wireless communications [], []. In aition, ue to two ierent network architectures existe in CRN, namely, primary network an cognitive network, ata transmission security issue has become a new research irection in the iel o CRN [7]-[9]. P security is the emphasis in energy-eicient CRN. In [], the authors pointe out that there is an optimal trae-o between energy eiciency an transmission security in CRN. The system aopts ierent P security strategies in accorance with ierent malicious attacks, thus energy consumption varies or ierent security schemes. Thereore, in orer to ensure secure an reliable transmission, it is necessary to iscuss the optimal energy-eicient security strategy or a speciic attack. Re. [8] an Re. [9] investigate the common malicious attacks an the corresponing countermeasure strategies in CRN. In CRN spectrum sensing phase, SUs irstly ientiy legitimate SUs an malicious SUs. The presence o malicious SUs will inluence the accuracy o secure cooperative spectrum sensing [5], [6]. A propose scheme in [6] mainly stuie the trae-o between energy eiciency an security or SSDF attack. The scheme eectively etecte channel availability through reporting message authentication coe (MAC) to the usion center. owever, ash unction was require to generate MAC in this scheme, which inevitably resulte in the increasing o complexity. Furthermore, the propose scheme in t consier the inluence o legitimate SUs reporting error rate as well as reporting istance to CSS energy consumption [6]. In [5], the authors stuie eective CSS in the presence o PUEA. Local sensing inormation rom ierent SU was weighte at the usion center. The weights coul be optimize by maximizing the etection probability uner a constraint o alse alarm probability. Meanwhile, the inluence o channel estimation error on CSS etection perormance was also iscusse. In [3], the authors investigate PUEA attack an its countermeasure strategy in mobility CRN scenario. A hybri strategy base on PUEA countermeasure with CSS etection was propose, which was satisie the requirements o alse alarm probability an the etection perormance in PUEA attack with lower energy consumption. At the same time, the overall perormance o PUEA problem with game theory analysis was presente in the conitions o ierent attack overhea an etection overhea, which coul reuce the total resource consumption an increase CRN throughput [3], [4]. At present, P security in spectrum sensing phase mainly ocuses on the PUEA or SSDF countermeasure strategies [4], [5], [7]-[9], [3], [4]. Aime to the trae-o between security an energy eiciency in CRN, the paper mainly ocuses on energy-eicient PUEA 7 Journal o Communications countermeasure strategy with perormance analysis [3], [6], [], [5]. Speciically, energy eiciency o the propose PUEA countermeasure strategy is erive in etail. Energy eiciency is set as the optimal obective, whereas secure etection probability an secure alse alarm threshol as the constraint conitions. Then, the optimal usion threshol K an optimal SU number N are presente to achieve the maximization o energy eiciency by resolving the optimization problem, which guarantees energy eiciency as well as secure CSS etection perormance. The rest o this paper is organize as ollows, Section II escribes the system moel. Secure CSS uner PUEA is escribe in etail in the Section III. Section IV escribes the propose countermeasure strategy against PUEA an analyzes its energy eiciency perormance. Simulation results are presente an iscusse in Section V. Finally, conclusions are rawn in Section VI. II. SSTEM MODEL In this paper, we consier CSS in a CRN where N SUs trying to access a target spectrum. As shown in Fig., to avoi conlict with the PU, each SU etects the target spectrum within a speciic perio an uges whether the spectrum is occupie by PU or not. Then, SUs report local etection results to the Fusion Center (FC). Finally, the FC makes global etection to ecie whether the spectrum is occupie by PU via speciic usion rule. The PUE attacker mimics the eature o PU an launches the PU signal to eceive SUs, which makes SUs to uge that PU is occupying the authorize spectrum. Thereore, SUs are orce to switch to another spectrum holes. The key issue to een PUEA is that SUs shoul ientiy PU an PUEA, an make right ugment on PU an PUEA signals [5], [3], [4]. In this paper, energy etection is implemente in SU local etection, an har usion with K-out-o-N usion rule is applie at FC. It is assume that the report channels rom SUs to the FC are noisy. imary user PUE attacker Seconary user SU SU SU3 SUN Channel between attacker an seconary user Channel between primary an seconary user an reporting Fusion Center channel Fig.. System moel o cooperative spectrum sensing with PUEA in CRN. III. COOPERATIVE SPECTRUM SENSING IN TE PRESENCE OF PUEA SU local spectrum sensing consists o three common techniques. Namely, energy etection, matche iltering

3 Journal o Communications Vol., No., January 7 etection an cyclo-stationary eature etection [], []. In this paper, we use energy etection to uge the authorize spectrum utilization. Due to the presence o PUEA, or N, the signal receive by the th SU at the ith (i M ) time instant [5] can be written as P h x ( i) w ( i), y () i P h x ( i) P h x ( i) w ( i), m m, m, p p, p, m m, m, We enote that when PUEA is present, the absence an presence o PU can be expresse as two hypotheses respectively, that is, an. Signals transmitte by PU an PUE attacker with power P p an P m are x i an x, () i respectively. PU signal x, () i is p, () m assume to be inepenently an ientically istribute (i.i.) complex Gaussian ranom variable with zero mean an unit variance. Due to the similarity between malicious an primary signal in PUEA, the attacker s signal xm, () i also ollows the complex Gaussian istribution. hp, an hm, enote the instantaneous channel gain rom primary user to the th seconary user an rom PUE attacker to the th seconary user, respectively. In aition, all the channel links are assume as block aing channel, that is, hp, an h m, are constant within one time interval. w () i enotes the aitive white Gaussian noise at the th seconary user with zero mean an variance. The signal-to-noise ratio (SNR) receive at the th SU can be expresse as = h. p, w In this paper, we aopt energy etection metho in which M samples o signal energy are ae uring one etection interval, the power etecte by the th SU can be expresse as m, () w p M y ( i). Due to p, () i () x i an x i ollow the complex Gaussian istribution with zero mean an unit variance, the ecision statistic subect to the central Chi-square ( ) istribution with M egrees o reeom [6] where an respectively M ~ M ( ) y () i () i ~ M ( ) are the variance or an Ph m m, w P h P h p p, m m, w is (3) In spectrum sensing, alse alarm probability an etection probability eine as over one etection interval are (4) where is etection threshol. ence, the alse alarm probability an etection probability at the th SU are expresse as [], [], [7] ( M, ) ( M, ) Each SU issues a local binary ecision, the spectrum status. I (5) about, then, which enotes that the spectrum is occupie by PU. Otherwise, the spectrum is ientiie as unuse by PU [], [], [7]. All local ecisions are reporte to the FC via reporting channels. At the FC, a speciic usion rule (FR) is employe to process these ecisions in orer to make the inal global ecision. In this paper, we use K-out-o-N rule, where K is a preeine threshol on the number o SUs who etect PU on the spectrum (K N ). The iea behin this rule is to compare the number o SUs that report with K. I it is less than K, then the spectrum is unuse by PU. Otherwise, the spectrum is use by PU. In this paper, the reporting channel between SU an FC is assume to be noisy, which can be moelle as binary symmetric channel with error probability e, hence, we can moiy the expression o an as below [8]. ( ) ( P x e ( ) ( P y e The overall etection probability ( alse-alarm probability ( respectively as ollows N ( ) N ( ) N N x x K N N y y K (6) (7) ) an overall ) can be expresse IV. ENERG EFFICIENC ANALSIS OF PUEA COUNTERMEASURE STRATEG (8) Energy eiciency ( ) can be eine as the ratio o average throughput in bits to average energy consumption in Joule. The average throughput is eine as average number o successully transmitte bits, an the average consume energy is the energy consume uring CSS an reporting results transmission in the reporting phase. The average throughput can be expresse as ollows Th ( ) R (9) 7 Journal o Communications 3

4 Journal o Communications Vol., No., January 7 where is the probability that the spectrum is not occupie by PU. R enotes the ata rate, an T is the ata transmission time interval. The actor ( ) in (9) represents the probability o successul elivering the transmit ata. The average energy consumption by all SUs is given as ollows where es E Ne e () s unuse t is the energy consume uring CSS by one SU, an et is the energy consume uring reporting phase. Note that transmission occurs only i the spectrum is ientiie as unuse. Thus, unuse can be given as ollows ( ) ( ) unuse () where. Thereore, energy eiciency o eense strategy against PUEA can be expresse as ollows ( ) RT Ne e ( ) s t () The optimization o K or energy eiciency in the propose PUEA countermeasure strategy with constraint conitions on the an, which can be state as below ( ) RT max max K K Ne e ( ) s.t. an s t (3) The K value that satisies (3) without the constraint can be obtaine by the erivative o obective into zero, namely RT K Ne e ( ) s t ( ) RTet K K Nes et ( ) The erivatives o approximately as ollows an (4) can be written N ( K ) ( K) y ( y ) K K N ( K ) ( K) x ( x ) K K K N K K N K (5) Substitute (5) into (4) with some mathematical calculations, the optimal K that maximizes the energy eiciency can be presente in a close orm shown as below Ne s et ( ) y ln N ln et ( ) x K (6) x( y) ln y ( x ) ence, or a given N an a given alse alarm probability, the etection probability ecrease as K increases, the optimal K that maximizes energy eiciency while ulilling the etection probability requirements can be written as below. Kopt min K, K (7) where the value o the K that satisies the constraint conition in CSS etection. From [7], we can get the approximate expression o K shown as below. K N N (8) ( ) x( x) x.5 V. SIMULATION RESULTS AND ANALSIS In this section, we analyze the simulation perormance o the propose PUEA countermeasure strategy with energy eiciency optimization. We assume that there is one PUE attacker in the propose scheme. Sensing channels an reporting channels are i.i.. The number o samples within a etection interval is M 3 [5]. All simulation parameters are summarize in Table I. These parameters are assume to be ientical among all SUs. TABLE I: SIMULATION PARAMETERS Parameters Values Parameters Values N.6..5 e s Joule e t Joule T.3 sec R Kbps Fig. shows the receiver operation characteristics (ROC) perormance when the FC uses har FR or CSS in the presence o PUEA. In this igure, the average SNR is set to be B an the transmitting power o primary user an PUE attacker are ientical P P. From Fig., we in that the etection probability o non-cooperation scheme is severely reuce by PUEA attack. When we use K-out-o-N FR in FC, a signiicant etection perormance improvement is achieve, even better than the non-cooperation scheme when the PUE attacker is absent. From Fig., we can in that when we use K-out-o-N rule, or a given N, the threshol K also have great impact on the etection perormance. When the alse alarm probability is ixe, the overall etection perormance shows better with the ecreasing o threshol K. The reason is that the FR approaches to OR rule with the ecreasing o K, so that the overall etection perormance is enhance. p m 7 Journal o Communications 4

5 Journal o Communications Vol., No., January 7 Fig. 3 shows energy eiciency o the propose PUEA countermeasure strategy with a set o ierent threshol K an SU noes N. In this igure, the average SNR is set to be B, the constraint conitions are.6 an.. It is shown that, the maximum energy eiciency o the propose PUEA countermeasure 4 strategy achieves to 7.8 bit/ Joule with N 5 an K. As shown in Fig. 3, we can also in that the threshol K that achieves to the maximum energy eiciency increases with the increasing o SU noes in CSS. In aition, with the increasing o SU noes, the achievable maximum energy eiciency is lower. The reason is that, more SU noes will increase the energy consumption in the process o CSS an reporting phase, which will ecrease energy eiciency. Detection probability, No cooperation,attack No cooperation,no attack an rule or rule K=,N=6 K=4,N= False alarm probability, Fig.. ROC o CSS with har usion rule in the presence o PUEA. Energy Eiciency [bit/joule] x 4 N=4 N=5 N=6 N=7 N=8 N= K Fig. 3. Energy eiciency o the propose strategy with ierent K an N. From perormance analysis in Fig. 3, we oun that the propose PUEA countermeasure strategy can achieve the maximum energy eiciency with SU number N 5 an the optimal threshol K. Fig. 4 shows the ROC perormance o non-cooperation, K-out-o-N rule an MRC with/without PUEA attack. For the same FR, ROC perormance without PUEA is better than that with PUEA. When the PUE is present, or the same alse alarm probability, the etection probability o the propose PUEA countermeasure strategy with K-out-o-N FR outperorms the MRC scheme. When., the etection perormance o the propose strategy shows even better than the non-cooperation scheme without PUEA. owever, alse alarm probability increases with the increasing o etection probability. ence, it s necessary to make a trae-o between alse alarm probability an etection probability to ensure the ROC perormance o secure CSS. Fig. 5 shows secure CSS etection perormance o the propose PUEA countermeasure strategy in the presence o PUEA with ierent power ratio between PUE an PU. We eine the transmit power ratio between PUE an PU as P p (9) P Eq. (9) normalizes PUE attacker s power in terms o PU s power. A larger value o p inicates PUE attacker has more transmit power. In this igure, or a given alse alarm probability threshol., the SNR threshols or each SU are ientical. Secure CSS etection perormance o the propose scheme is compare with MRC scheme whereas p is set to be.// respectively. As p increases rom. to, etection perormance o both schemes are ecrease. p inicates PUE power is ominant over the noise power. ence, the etection perormance is eteriorate even when the average SNR is very high. We also in that the etection perormance o the propose secure CSS scheme base on K-out-o-N FR outperorms MRC FR signiicantly. What s more, the etection perormance o propose scheme is almost inepenent o the value p. That is, PUE power has harly aecte the propose PUEA countermeasure strategy, which inicates the propose scheme ensures the security o CSS eectively, an the propose scheme has robustness to een PUEA attack. m p VI. CONCLUSIONS This paper investigates secure CSS strategy that ensures energy eiciency an P layer security in CSS spectrum sensing uner the presence o PUEA in the CRN. We obtain the optimal threshol K an optimal SU numbers N o secure CSS base on K-out-o-N FR via the solution o energy eiciency optimization problem, in orer to ensure the CSS etection perormance simultaneously. In aition, we analyze the ROC perormance with the consieration o optimal energy eiciency. With the comparison o MRC FR, it is apparent that secure etection perormance o the propose scheme with K-out-o-N FR outperorms MRC FR scheme in terms o the same alse alarm probability scenario. Detection perormance o the propose PUEA countermeasure strategy can be signiicantly enhance in 7 Journal o Communications 5

6 Journal o Communications Vol., No., January 7 high SNR region. The propose scheme is almost immune to PUE intererence power, which ensures CSS security. Thereore, the propose strategy has higher robustness to PUEA attack, an it makes a trae-o between energy eiciency an etection perormance eectively. Detection probability, No cooperation,attack No cooperation,no attack. K=,N=5,attack K=,N=5,no attack. MRC,no attack MRC,attack False alarm probability, Fig. 4. ROC perormance o non-cooperation K-out-o-N rule an MRC with/without PUEA attack. Detection probability, MRC,p= MRC,p=. MRC,p= K=,N=5,p=. K=,N=5,p= K=,N=5,p= SNR(B) Fig. 5. Detection perormance o the propose PUEA countermeasure strategy with ierent power ratios. ACKNOWLEDGMENT The authors woul like to greatly appreciate anonymous reviewers or their valuable comments an constructive suggestions in helping to improve the quality o this paper. REFERENCES [] I. F. Akyiliz, W.. Lee, M. C. Vuran, an S. Mohanty, Next generation/ynamic spectrum access/cognitive raio wireless networks: A survey, Computer Networks, vol. 5, no. 3, pp. 7-59, September 6. [] I. F. Akyiliz, B. F. Lo, an R. Balakrishnan. Cooperative spectrum sensing in cognitive raio networks: A survey, Physical Communications, vol. 4, no., pp. 4-6, March. [3] X. ong, J. Wang, an C. X. Wang. Cognitive Raio in 5G: A perspective on energy-spectral eiciency traeo, IEEE Communication Magazine, vol. 5, no. 7, pp , 4. [4] T. Chen an G. Wei. Stuy on physical layer security o wireless networks, Doctoral Dissertation o South China University o Technology, 3. [5] C. Chen,. Cheng, an. D. ao. Cooperative spectrum sensing in cognitive raio networks in the presence o the primary user emulation attack, IEEE Transactions on Wireless Communications, vol., no. 7, pp. 35-4,. [6] S. Althunibat, V. Sucasas,. Marques, J. Roriguez, R. Taazolli, an F. Granelli, On the traeo between security an energy eiciency in cooperative spectrum sensing or cognitive raio, IEEE Communications Letters, vol. 7, no. 8, pp , 3. [7]. N. Li. Research on security an privacy in cognitive raio networks, Doctoral Dissertation o Xiian University, 3. [8] Z. Shu,. ian, an S. Ci. On physical layer security or cognitive raio networks, IEEE Network, vol. 7, no. 3, pp. 8-33, 3. [9] J. Li, Z. Feng, Z. Feng, an P. Zhang, A survey o security issues in cognitive raio networks, China Communications, vol., no. 3, pp. 3-5, 5. []. Zou, X. Wang, an W. Shen, Physical-layer security with multi-user scheuling in cognitive raio networks, IEEE Transactions on Communications, vol. 6, no., pp , 3. [] J. Wang, J.. Zhao, an J. J. Du, Survey on cognitive raio green networks with cross-layer technology, Telecommunication Science, vol. 3, no. 3, pp. 4-9, 4. [] S. Eryigit, G. Gur, S. Bayhan. Energy eiciency is a subtle concept: Funamental traeos or cognitive raio networks, IEEE Communication Magazine, vol. 5, no. 7, pp.3-36, 4. [3] F. J. Bao. Research on primary user emulation attack an countermeasures with mobile seconary users in cognitive raio networks, Master Dissertation o Zheiang University, 3. [4] F. Bao, L. Xie, an. Chen, Analysis o primary user emulation attack with motional seconary users in cognitive raio networks, in oc. IEEE 3r International Symposium on Personal Inoor an Mobile Raio Communications, September, pp [5] C. Comaniciu an. V. Poor, On energy-secrecy traeos or Gaussian wiretap channels, IEEE Transactions on Inormation Forensics an Security, vol. 8, no., pp , 3. [6] J. ang,, Chen, W. Shi, X. Dong, an T. Peng. Cooperative spectrum sensing against attacks in cognitive raio networks, in oc. IEEE International Conerence on Inormation an Automation, 4, pp [7] S. Althunibat, M. Di Renzo, an F. Granelli, Optimizing the K-out-o-N rule or cooperative spectrum sensing in cognitive raio networks, in oc. IEEE Global Communications Conerence (IEEE Globecom 3) Communications os, Reliability an Moeling Symposium, 3, pp [8] X. Xu, J. Bao,. Luo, an. Wang, Cooperative wieban spectrum etection base on maximum 7 Journal o Communications 6

7 Journal o Communications Vol., No., January 7 likelihoo ratio or CR enhance VANET, Journal o Communications, vol. 8, no., pp. 84-8, 3. unchuan Wang is with the College o Telecommunication Engineering, angzhou Dianzi University (DU), angzhou, China, as master. e receive the B. Eng. egree in Communication Engineering rom Tianin University o Technology, Tianin, China, in 5. e is currently working towar his Master egree in the College o Telecommunication Engineering, DU. is research interests emphasize on energy eiciency an P security in Cognitive Raios (CR). Xiaorong Xu is with the College o Telecommunication Engineering, angzhou Dianzi University (DU), angzhou, China, as associate proessor an master supervisor. e receive the B. Eng. egree in Communication Engineering an M. Eng. egree in Communication an Inormation System rom DU, angzhou, China, in 4 an 7, respectively. e receive Ph.D. egree maor in Signal an Inormation ocessing rom Naning University o Posts an Telecommunications (NJUPT), Naning, China, in. eviously, rom to 3, he was working as a postoctoral researcher in the Institute o Inormation an Communication Engineering, Zheiang University (ZJU), angzhou, China. During 3-4, he serve as a research scholar with the Electrical an Computer Engineering Department, Stevens Institute o Technology (SIT), oboken, NJ, USA. Currently, he is working as an associate proessor an master supervisor in DU. Dr. Xu s research interests emphasize on energy eiciency an P security in Cognitive Raios (CR) an energy eiciency in cooperative communications, etc. Weiwei Wu is with the College o Telecommunication Engineering, angzhou Dianzi University (DU), angzhou, China, as master. e receive the B. Eng. egree in Communication Engineering rom Anhui University o Technology, Anhui, China, in 5. e is currently working towar his Master egree in the College o Telecommunication Engineering, DU. is research interests emphasize on energy eiciency an optimal resource allocation in Cognitive Raios (CR). Jianrong Bao is with the College o Inormation Engineering, angzhou Dianzi University (DU), angzhou, China, as associate proessor, e receive the B. Eng. egree in Polymer Material Engineering an M. Eng. egree in Communication an Inormation System rom Zheiang University o Technology, angzhou, China, in an 3, respectively. e receive Ph.D. egree maor in Communication an Inormation System rom Tsinghua University, Beiing, China, in 9. eviously, rom to 3, he was working as a postoctoral researcher in the Institute o Inormation an Communication Engineering, Zheiang University (ZJU), angzhou, China. Currently, he is working as associate proessor in DU. Dr. Bao s research interests emphasize on energy eiciency in Cognitive Raios (CR) an cooperative communications, etc. 7 Journal o Communications 7