IET Counications Research Article Cooperative spectru sensing with secondary user selection or cognitive radio networks over Nakagai- ading channels ISSN 1751-8628 Received on 14th May 2015 Revised on 3rd August 2015 Accepted on 8th Septeber 2015 doi: 10.1049/iet-co.2015.0461 www.ietdl.org Quoc-Tuan Vien 1, Huan X. Nguyen 1, Aruuga Nallanathan 2 1 School o Science and Technology, Middlesex University, The Burroughs, London, Hendon NW4 4BT, UK 2 Departent o Electrical and Electronic Engineering, King s College London, Strand, London WC2R 2LS, UK E-ail: q.vien@dx.ac.uk Abstract: This study investigates cooperative spectru sensing (CSS) in cognitive wireless radio networks (CWRNs). A practical syste is considered where all channels experience Nakagai- ading and suers ro background noise. The realisation o the CSS can ollow two approaches where the inal spectru decision is based on either only the global decision at usion centre (FC) or both decisions ro the FC and secondary user (SU). By deriving closed-or expressions and bounds o issed detection probability (MDP) and alse alar probability (FAP), the authors are able to not only deonstrate the ipacts o the -paraeter on the sensing perorance, but also evaluate and copare the eectiveness o the two CSS schees with respect to various ading paraeters and the nuber o SUs. It is interestingly noticed that a saller nuber o SUs could be selected to achieve the lower bound o the MDP rather using all the available SUs while still aintaining a low FAP. As a second contribution, they propose a SU selection algorith or the CSS to ind the optiised nuber o SUs or lower coplexity and reduced power consuption. Finally, nuerical results are provided to deonstrate the indings. 1 Introduction Cognitive radio (CR) has recently eerged as a novel technology to eiciently exploit spectru resource by ipleenting dynaic spectru access [1, 2]. The secondary users (SUs) can opportunistically utilise the licenced requency bands o the priary users (PUs) when they are not occupied. Thus, spectru sensing is a basic eleent required at the SUs to detect the occupation and reappearance o the PUs [3]. Incorporating relaying techniques in cognitive wireless radio networks (CWRNs), cooperative spectru sensing (CSS) has been proposed not only to help the shadowed SUs detect the licenced requency bands, but also to iprove sensing reliability o the SUs [4 9]. Basically, a CSS schee consists o sensing (SS) phase, reporting phase (RP) and backward (BW) phase. Every SU perors local spectru sensing (LSS) to deterine the availability o the licenced spectru in the SS phase and then orwards its local decisions to a usion centre (FC) in the RP phase. Collecting all LSS decisions, global spectru sensing (GSS) is then carried out at the FC to ake a global decision on the spectru availability, which is then broadcast back to all the SUs in the BW phase. To save the energy consuption o CSS in CWRNs, user selection approaches have been investigated in various works, such as [10 15]. Speciically, an energy-based user selection algorith was proposed in [10] given battery lie constraints o SUs. To deal with the dynaic changes o the network topology and channel conditions, a correlation-aware distributed user selection algorith was developed in [11] to adaptively select the uncorrelated SUs or the CSS. The overhead energy caused by the CSS was also dealt with in [12] where an energy-eicient node selection was proposed to select the best node based on the binary knapsack proble. In [13], the selection o sensing nodes can also be realised by linearly weighting the sensing data at all sensing nodes. Considering the scenario when only partial inoration o SUs and PUs is available in the wireless sensor networks, an energy-eicient sensor selection algorith has been proposed in [14] to iniise the energy consuption while still satisying the average detection probability. An optiisation raework has also been developed in [15] to solve the proble o joint sensing node selection, decision node selection and energy detection threshold aiing at saving energy consuption in cognitive sensor networks. In this paper, we irst analyse the probabilities o issed detection and alse alar o two CSS schees over Nakagai- ading channels, including (i) global decision-based CSS (GCSS): the GSS decision is the inal spectru sensing (FSS) decision at the SUs (e.g. [5]) and (ii) ixture o local and global decisions-based CSS (MCSS): both the LSS and GSS decisions are taken into account to ake the FSS decision at the SUs (e.g. [9]). In particular, we consider a practical scenario where all SS, RP and BW links suer ro ading and noises. Our work neither assues the RP/BW channels are error-ree [4, 5, 7, 12] nor ipractically requires instantaneous signal-to-noise ratio (SNR) inoration and excessive overhead [16, 17]. Our schee requires a inial 1-bit overhead or the report and eedback o sensing decisions. While a general criteria or decision-approach selection has been analytically derived in the presence o realistic channel propagation eects in [18], our work speciically ais at analysing and understanding behaviour o sensing perorance over Nakagai- channel with inial overhead. By deriving closed-or expressions o issed detection probability (MDP) and alse alar probability (FAP), we irst copare the sensing perorance achieved with the above CSS schees and evaluate the eects o the nuber o SUs and the ading channel paraeters on the perorance. It is observed that GCSS schee achieves a lower FAP while MCSS schees iprove the MDP. The ading paraeters o the RP and BW channels are shown to have eects on both the MDP and FAP, while those o the SS channels only aect the MDP. Furtherore, the bounds o the MDP and FAP are then derived or a large nuber o SUs, which allows us to develop a SU selection algorith or reducing the coplexity and power consuption in the CSS. The reaining o the paper is organised as ollows. In Section 2, we introduce the syste odel o CWRNs and the process o GCSS and MCSS schees. Section 3 derives the expressions and bounds o FAP and MDP or both CSS schees over Nakagai- ading & The Institution o Engineering and Technology 2016 91
channels. The SU selection algorith or the CSS is presented in Section 4. Nuerical results are presented and discussed in Section 5. Finally, Section 6 concludes the paper. 2 Syste odel and CSS 2.1 Syste odel The syste odel o a CWRN under investigation is illustrated in Fig. 1 consisting o PU, SU 1, SU 2,..., SU N } and FC. We assue there are K non-overlapping licenced requency bands 1, 2,, K. Two hypothesis that the kth requency band is occupied and unoccupied by PU are denoted by H 1,k and H 0,k, respectively. For convenience, let us deine a spectru indicator vector (SIV) s (M) A, M L, G, F j}, A, FC}, i =1, 2,, N, j =1, 2, o length K (in bits) to report the availability o the licenced spectru in the LSS at, the GSS at FC and the FSS at using schee j. The unavailable and available requency bands are represented in s (M) A by bits 0 and 1, respectively. The channel or a link A 1 A 2 is denoted by h A1 A 2,A 1, A 2 } P, S i, F}, A 1 A 2, and assued to suer ro quasi-static slow Nakagai- ading. Coplex Gaussian noise vector n (T) A, T SS, RP, BW}, is assued at receiver node A in phase T, in which each entry has zero ean and variance o s 2 0. 2.2 Cooperative spectru sensing Three phases o CSS can be briely described as ollows: (i) SS phase LSS: The signal sensed at, i =1,2,, N, at k, k =1,2,, K, can be expressed as r (SS) [k] = h PS i x[k] + n (SS) [k], H 1,k, n (SS) [k], H 0,k, where x[k] is the transitted signal ro PU. Then, detects the availability o k by coparing the energy o the received signal in (1) with an energy threshold ɛ i [k] via an energy easureent j[ ] as ollows s (L) [k] = 0, i j[r(ss) [k]] 1 i [k], 1, otherwise. (ii) RP GSS: The received signal at FC ro, i =1,2,, N,at k, k =1,2,, K, can be written by i [k] = L i r (RP) hsi F x(l) (1) (2) [k] + n (RP) FC [k], (3) where Λ i is the transission power o and x (L) [k] is the binary phase-shit keying (BPSK) odulated version o s (L) [k] [see (2)]. Then, FC decodes and cobines all the decoded SIVs (denoted by s (RP) i [k]}) ro all } using the OR rule to ake a global decision as FC [k] = 0, i N s(rp) i [k], N, 1, otherwise. s (G) (iii) BW phase FSS: The received signal at, i =1,2,, N,ro FC with respect to k, k =1,2,, K, is given by [k] = L FC r (BW) hfsi x (G) FC where Λ FC is the transission power o FC and x (G) FC (4) [k] + n(bw) [k], (5) [k] is the BPSK decodes the odulated version o s (G) FC [k] [see (4)]. Then, received signal as s (BW) [k]. GCSS schee: The GSS decision received ro FC is also the FSS at. Thus, we have s (F 1 ) [k] = s (BW) [k]. (6) MCSS schee: cobines its local SIV with the global SIV received ro FC as [9] s (F 2 ) [k] = 3 Perorance analysis 0, i (s(l) [k] + s (BW) [k]), 2, 1, otherwise. In this section, we analyse the FAP and MDP o two CSS schees. Without loss o generality, a speciic requency band is considered and thus the index o the requency band (i.e. k) is oitted in the rest o the letter. The Nakagai ading paraeters o the SS, RP and BW channels are denoted by ss, rp and bw, respectively. We assue that the RP and BW channels o the sae link have the sae Nakagai ading paraeters (i.e. rp = bw ) and all the SUs have the sae energy threshold (i.e. ɛ i [k]=ɛ[k] i = 1, 2,..., N). Deine a W 1/(2s 2 0) and b i W ss s 2 0/( ss s 2 0 + g PSi ), i =1,2,, N, where g PSi is the average SNR at over h PSi. The FAP and MDP o the LSS are given by [19] P () P ( ) (7) = Prs (L) = 0 H 0 } = G u (r, a), (8) G(r) = Prs (L) = 1 H 1 } = 1 q i,1 q i,2, (9) where Fig. 1 Syste odel o cognitive wireless relay network q i,1 = e ( ab i / ss ) [b ss 1 i L ss 1 ( a(1 b i )) + (1 b i ) ss 2 (10) b j i L j ( a(1 b i ))], q i,2 = b ss i e a r 1 j=1 j=0 a j j! 1F 1 ( ss ; j + 1; a(1 b i )), (11) ρ denotes the tie-bandwidth product o the energy detector, Γ( )is the gaa unction [20, eq. (8.310.1)], Γ u (, ) is the upper incoplete gaa unction [20, eq. (8.350.2)], 1 F 1 ( ; ; ) is the conluent hypergeoetric unction [20, eq. (9.210.1)] and L i ( ) is the Laguerre polynoial o degree i [20, eq. (8.970.2)]. 92 & The Institution o Engineering and Technology 2016
Over a Nakagai- ading channel h AB, the average bit error rate (BER) or BPSK odulation with respect to the average SNR o g AB is obtained as in [21] and given below ( P b (E AB ) = 1 + g ) AB AB G( AB + 1/2) AB 2 p G(AB + 1) 2 F 1 ( AB,1/2; AB + 1; 1/(1 + g AB / AB )) W c AB, (12) where 2 F 1 (, ; ; ) is the Gauss hypergeoetric unction [20, eq. (9.100)]. Considering the GSS at FC, we obtain the ollowing: Lea 1: The FAP and MDP o the GSS are deterined by = 1 1 [G(r)] N N = N [G l (r, a)(1 c Si F ) + G u (r, a)c S i F ] (13) [(1 q i,1 q i,2 )(1 c Si F) + (q i,1 + q i,2 )c Si F], (14) where c Si F is given by (12) and Γ l (, ) is the lower incoplete gaa unction [20, eq. (8.350.1)]. Proo: Fro (4), the FAP and MDP at FC are given by = Prs (G) FC = 0 H 0} = 1 N = Prs (G) FC = 1 H 1 } = N Prs (RP) i = 1 x = 0}, (15) Prs (RP) i = 1 x = 0}. (16) Thus, over the Nakagai- ading RP channels, we have = 1 N = N [(1 P ( ) )(1 c Si F ) + P( ) c Si F ], (17) [P ( ) (1 c Si F) + (1 P ( ) )c Si F]. (18) Substituting (8) and (9) into (17) and (18) with the act Γ u (ρ, α)+ Γ l (ρ, α)=γ(ρ) [20, eq. (8.356.3)], the lea is proved. In the BW phase, we have the ollowing indings: Lea 2: The FAP and MDP o the FSS at,, 2,, N, using GCSS schee are deterined by P ( ),1 P ( ) = 1 [(1 )(1 c FSi ) + c FSi ], (19),1 = P (FC) (1 c FS i ) + (1 P (FC) )c FS i. (20) Proo: Fro (6), the proo can be straightorwardly obtained as in Lea 1. Lea 3: The FAP and MDP o the FSS at,, 2,, N, using MCSS schee are deterined by P (),2 = 1 1 G(r) [G l (r, a)(1 c FS i ) + G u (r, a)c FSi ] [(1 )(1 c FSi ) + c FSi ], P (),2 = [(1 q i,1 q i,2 )(1 c FSi ) + (q i,1 + q i,2 )c FSi ] [P (FC) (1 c FS i ) + (1 P (FC) )c FS i ]. Proo: Fro (7), P ( ),2 and P ( ),2 can be given by P ( ),2 = Prs (F 2 ) = 0 H 0 } = 1 Prs (L) = 1 x = 0}Prs (BW) = 1 x = 0}, P (),2 = Prs (F 2 ) = 1 H 1 } = Prs (L) = 1 x = 0}Prs (BW) = 1 x = 0}. Thus, over the Nakagai- BW channels h FSi, P ( ),2 and P ( ) be obtained by (21) and (22), respectively. (21) (22) (23) (24),2 can Reark 1: (Lower FAP with GCSS schee and lower MDP with MCSS schee). Fro (19) (22) in Leas 2 and 3, it can be easily shown that P ( ),1, P ( ),2 and P ( ),1. P ( ),2, i =1,2,, N. Reark 2: (Lower MDP but higher FAP with increased nuber o SUs): Fro (13) and (14) in Lea 1, it can be seen that and P (FC) onotonically increase and decrease, respectively, over N. Thus, ro (19) (22), the increased nuber o SUs helps both CSS schees iprove the MDP, however, causing a higher FAP. Reark 3: (Ipact o Nakagai- ading paraeters on MDP and FAP). Both the MDP and FAP decrease when the ading paraeters o RP and BW channels increase, while only MDP is iproved with increased ading paraeters o SS channels. In act, it is known that the BER o a Nakagai- ading channel h AB onotonically decreases as AB increases [see (12)]. Thus, ro (19) (22), it can be proved that P ( ),j and P ( ),j, i =1, 2,, N, j =1, 2, onotonically decrease as either rp or bw increases. In addition, as shown in (8) and (9), P ( ), i =1, 2,, N, o the LSS is independent o ss, while a lower P ( ) is achieved as ss increases. Bounds o FAPs and MDPs: According to Reark 2, there is a signiicant ipact o the nuber o SUs on FAPs and MDPs. For the sake o providing insightul eanings o the above derived expressions or the FAPs and MDPs o the two CSS schees, let us investigate a speciic scenario o identical channels, i.e. g PSi W g ss, g Si F W g rp, g FS i W g bw, i = 1, 2,..., N. Thus, ro (10) and (11), we can rewrite q i,1 = q 1 and q i,2 = q 2. Lea 4: When the nuber o SUs is very large, i.e. N, FAP and MDP o GCSS schee approach,1 N1 and,1 N1, respectively, where,1 N1 = 1 c bw, (25),1 N1 = c bw. (26) Proo: As N, ro Lea 1, it can be seen that 1 and P (FC) 0. Substituting into (19) and (20), we obtain,1 N1 and,1 N1 as shown in (25) and (26). & The Institution o Engineering and Technology 2016 93
Fig. 2 SU selection algorith or the CSS Lea 5: When the nuber o SUs is very large, i.e. N, FAP and MDP o MCSS schee approach,2 N1 and,2 N1, respectively, where,2 N1 = 1 G l (r, a) c G(r) bw G u (r, a) G l (r, a) c 2 bw G(r) (27),2 N1 = (1 q 1 q 2 )c bw (1 2q 1 2q 2 )c 2 bw. (28) Proo: The proo is siilarly obtained as in Lea 4. Reark 4: (Lower MDP with MCSS schee and approxiately siilar FAPs)In act, ro (26) and (28), it can be easily shown that,2 N1,,1 N1, which eans that a lower MDP bound is achieved with MCSS schee. Considering the FAP bound, it is noted that Γ l (ρ, α) Γ(ρ) as a = 1/(2s 2 0 ) 1. Also, we have c 2 bw c bw, 1. Thus, ro (27), we have 1 c bw =,1 N1.,2 N1 4 SU selection or the CSS Fro Leas 4 and 5, it is noted that, given a large nuber o available SUs or CSS, we can select a saller nuber o SUs to achieve the lower bound o the MDP instead o using all the SUs while still guaranteeing an achievable lower FAP. Let N opt denotes the optiised nuber o SUs or the CSS. We achieve the ollowing: Lea 6: The optiised nuber o SUs or the CSS is deterined by ( ) N opt = n n [ ] n k ( 1) k c k rp + (q 1 + q 2 )(1 2c rp ) = t, (29) k=0 where t 0 + is an extreely sall positive nuber. Proo: In Leas 4 and 5, the lower bound o the MDP o Schee j, j = 1, 2, i.e., j N1, is achieved as 0. Fro (14) with identical ading channels, the MDP o the GSS can be rewritten as = [c rp + (1 q 1 q 2 )(1 2c rp )] N = [1 c rp (q 1 + q 2 )(1 2c rp )] N. (30) Applying Taylor expansion o power series [20, eq. (1.111)], we obtain = n k=0 ( ) n ( 1) k [c k rp + (q 1 + q 2 )(1 2c rp )] k. (31) Denote t as an extreely sall positive nuber, i.e. t 0 +. Since P (FC) 0, the optiised nuber o SUs can be ound by solving P (FC) = t. The lea is proved. It is noted in Lea 6 that N opt can be deterined using nuerical ethod. A SU selection algorith can thus be proposed as suarised in Fig. 2 [Note that the deterination o the optiised nuber o SUs and the user selection or the CSS can be carried out with the assistance o a coordinator, e.g. FC.]. 5 Nuerical results Fig. 3 shows the MDP against the FAP o two CSS schees (i.e. FCSS and MCSS) with respect to various ading paraeters and various values o the energy threshold. We assue there are ten SUs (i.e. N = 10) and the tie-bandwidth product o the energy detector is ρ = 5. The SNRs o the channels are set as ollows: g PSi } = 10, 8, 9, 12, 5, 7, 8, 4, 2, 6} db, g Si F} = 8, 7, 10, 4, 6, 8, 9, 11, 8, 10} db and g FSi } = 10, 11, 13, 9, 94 & The Institution o Engineering and Technology 2016
Fig. 3 Perorance coparison o two CSS schees Fig. 4 Perorance o MCSS schee 8, 14, 11, 10, 12, 7} db. Two Nakagai- ading scenarios are considered: (i) ss =3, rp = bw = 1 and (ii) ss =1, rp = bw = 2. It can be observed that, at a given energy threshold, the MCSS schee achieves a lower MDP than the FCSS schee, while a lower FAP is achieved with the FCSS copared to the MCSS. This observation conirs the stateent in Reark 1. In addition, the analytical results o the FAP and MDP or both CSS schees derived in Leas 2 and 3 are shown to be consistent with the siulation results. Investigating the ipact o Nakagai- ading paraeters on the sensing perorance o the CSS, Fig. 4 plots the MDP versus FAP o MCSS schee with respect to various ading scenarios [The ipact o the ading paraeters on the sensing perorance o GCSS schee can be siilarly observed, and thus is oitted or brevity.]. A total o ten SUs is considered and the SNRs o the SS, RP and BW channels are siilarly set as shown in Fig. 3. As shown in Fig. 4, given ixed ss, both the MDP and FAP are iproved as rp (or bw ) increases. Considering the scenario o ixed rp and bw, a lower MDP is achieved as ss increases, while the FAP is unchanged or all values o the energy threshold. These above coparisons veriy the stateent in Reark 3. The ipact o the nuber o SUs on the sensing perorance o various CSS schees is shown in Figs. 5 and 6 where the FAP and MDP o the two aoreentioned CSS schees are plotted as unctions o N. The SNRs o the SS, RP and BW links are set as 8, 10 and 12 db, respectively. We consider three ading scenarios: (i) ss =1, rp = bw = 2, (ii) ss =2, rp = bw = 1 and (iii) ss = 1, rp = bw = 10. It can be observed that both schees approach the siilar FAP upper bound as N is large, while the MDP o the MCSS schee approaches a lower MDP bound. This accordingly veriies the stateents in Rearks 2 and 4. Also, the FAP and MDP o the two CSS schees are shown to approach the bounds given by (25) (28) in Leas 4 and 5. Fig. 7 plots the optiised nuber o SUs as a unction o the SNR o SS links. Various ading scenarios are considered and the SNR o the RP and BW links are set as 6 and 4 db, respectively. As shown in Lea 6, N opt is deterined using nuerical ethod with t =10 5 and 100 available SUs (i.e. n ax = 100). It can be seen that a lower nuber o SUs can be selected to achieve the lower bound o the MDP rather than using all 100 SUs, which accordingly eans a lower coplexity and reduced power consuption are achieved with the proposed user selection or the CSS in CWRNs. & The Institution o Engineering and Technology 2016 95
Fig. 5 FAP o CSS schees over the nuber o SUs Fig. 6 MDP o CSS schees over the nuber o SUs Fig. 7 Optiised nuber o SUs over SNR o SS links 96 & The Institution o Engineering and Technology 2016
Moreover, a lower N opt is required as either the SS perorance iproves or the ading paraeters increase. 6 Conclusions In this paper, we have analysed the MDP and FAP or two CSS schees in CWRNs considering the practical scenario where all SS, RP and BW channels suer ro Nakagai- ading. The derived expressions have shown the MCSS schee achieves an iproved MDP while causing a higher FAP when copared to the GCSS schee. Both the MDP and FAP are iproved as the ading paraeters o the RP and BW channels increase, while the increased ading paraeters o SS channels only results in a lower MDP. Furtherore, the bounds o the MDP and FAP have been derived and an optiised nuber o SUs has been deterined to reduce the coplexity and power consuption in the CSS. For uture work, we will investigate the perorance o the CSS schees along with the SU selection taking into account various scenarios o channel quality and node location. 7 Reerences 1 Haykin, S.: Cognitive radio: brain-epowered wireless counications, IEEE J. Sel. Areas Coun., 2005, 23, (2), pp. 201 220 2 Mitola, I., Maguire, G.Q.: Cognitive radio: Making sotware radios ore personal, IEEE Pers. Coun., 1999, 6, (4), pp. 13 18 3 Yucek, T., Arslan, H.: A survey o spectru sensing algoriths or cognitive radio applications, IEEE Coun. Surv. Tutor., 2009, 11, (1), pp. 116 130 4 Ganesan, G., Li, Y.: Cooperative spectru sensing in cognitive radio, part I: Two user networks, IEEE Trans. Wirel. Coun., 2007, 6, (6), pp. 2204 2213 5 Zhang, W., Letaie, K.: Cooperative spectru sensing with transit and relay diversity in cognitive radio networks, IEEE Trans. Wirel. Coun., 2008, 7, (12), pp. 4761 4766 6 Ben Letaie, K., Zhang, W.: Cooperative counications or cognitive radio networks, Proc. IEEE, 2009, 97, (5), pp. 878 893 7 Noh, G., Wang, H., Jo, J., et al.: Reporting order control or ast priary detection in cooperative spectru sensing, IEEE Trans. Veh. Technol., 2011, 60, (8), pp. 4058 4063 8 Zou, Y., Yao, Y.-D., Zheng, B.: Cooperative relay techniques or cognitive radio systes: spectru sensing and secondary user transissions, IEEE Coun. Mag., 2012, 50, (4), pp. 98 103 9 Vien, Q.-T., Stewart, B.G., Tianield, H., et al.: Eicient cooperative spectru sensing or three-hop cognitive wireless relay networks, IET Coun., 2013, 7, (2), pp. 119 127 10 Moneian, M., Mahdavi, M.: Sensing user selection based on energy constraints in cognitive radio networks. Proc. IEEE WCNC 2014, Istanbul, Turkey, April 2014, pp. 3379 3384 11 Cacciapuoti, A., Akyildiz, I., Paura, L.: Correlation-aware user selection or cooperative spectru sensing in cognitive radio ad hoc networks, IEEE J. Sel. Areas Coun., 2012, 30, (2), pp. 297 306 12 Hasan, N., Ejaz, W., Lee, S., et al.: Knapsack-based energy-eicient node selection schee or cooperative spectru sensing in cognitive radio sensor networks, IET Coun., 2012, 6, (17), pp. 2998 3005 13 Zhou, Y., Zhou, Z., Li, B.: Sensing nodes selection and data usion in cooperative spectru sensing, IET Coun., 2014, 8, (13), pp. 2308 2314 14 Najii, M., Ebrahizadeh, A., Hosseini Andargoli, S., et al.: Energy-eicient sensor selection or cooperative spectru sensing in the lack or partial inoration, IEEE Sens. J., 2015, 15, (7), pp. 3807 3818 15 Ebrahizadeh, A., Najii, M., Andargoli, S., et al.: Sensor selection and optial energy detection threshold or eicient cooperative spectru sensing, IEEE Trans. Veh. Technol., 2015, 64, (4), pp. 1565 1577 16 Ma, J., Zhao, G., Li, Y.: Sot cobination and detection or cooperative spectru sensing in cognitive radio networks, IEEE Trans. Wirel. Coun., 2008, 7, (11), pp. 4502 4507 17 Liu, X., Sirkeci-Mergen, B.: Group-orthogonal MAC or cooperative spectru sensing in cognitive radios. Proc. IEEE MILCOM 2010, San Jose, CA, USA, October 2010, pp. 1221 1226 18 Cacciapuoti, A.S., Calei, M., Paura, L., et al.: Decision aker approaches or cooperative spectru sensing: participate or not participate in sensing?, IEEE Trans. Wirel. Coun., 2013, 12, (5), pp. 2445 2457 19 Digha, F.F., Alouini, M.-S., Sion, M.K.: On the energy detection o unknown signals over ading channels, IEEE Trans. Coun., 2007, 55, (1), pp. 21 24 20 Gradshteyn, I.S., Ryzhik, I.M.: Table o integrals, series, and products (Acadeic Press, 2007, 7th edn.) 21 Shin, H., Lee, J.H.: On the error probability o binary and M-ary signals in Nakagai- ading channels, IEEE Trans. Coun., 2004, 52, (4), pp. 536 539 & The Institution o Engineering and Technology 2016 97