A Cooperative Spectrum Sensing Scheme Based on Trust and Fuzzy Logic for Cognitive Radio Sensor Networks

Transcription

1 IJCSI Internatona Journa of Computer Scence Issues, Vo., Issue, No, January 23 ISSN (Prnt: ISSN (Onne: A Cooperatve Spectrum Sensng Scheme Based on Trust and Fuzzy Logc for Cogntve Rado Sensor Networks Yonghua Wang,2, Yuehong L, Fe Yuan 3 and Jan Yang Schoo of Automaton, Guangdong Unversty of Technoogy Guangzhou, 56, Chna 2 Shenzhen Key Laboratory of Hgh Performance Data Mnng Shenzhen, 5855, Chna 3 Schoo of Informaton Scence and Technoogy,Sun Yat-sen Unversty Guangzhou, 56, Chna Abstract Ths paper proposes a cooperatve spectrum sensng scheme based on trust and fuzzy ogc for Cogntve Rado Sensor Networks (CRSN. The CRSN nodes use the T-S fuzzy ogc to make oca decsons on the presence or absence of the prmary user s (PU sgna, and then use a censorng method to ony aow the reatvey reabe decsons sent to the fuson center. Utzng a trust evauaton scheme based on the factors such as oca sensng dfference, sensng ocaton factors, and sensng channe condtons for each node. Combng the majorty rue and the trust vaues of the nodes, the fuson center makes the fna decson. Smuaton resuts show that the proposed scheme coud mprove the detect probabty effectvey. Keywords: Cogntve Rado Sensor Networks, trust, fuzzy ogc, Cooperatve Spectrum Sensng.. Introducton Most Wreess Sensor Networks (WSNs operate n the uncensed ISM (Industra, Scentfc and Medca frequency band, for exampe, the 2.4 GHz band. Whe such bands are aso used by other wreess appcatons such as WF, Buetooth, cordess phones, RFID, mcrowave ovens, etc. And n a arge-scae wreess sensor network, due to the event-drven nature, when an event occurs many WSN nodes need to transmt the event sgna data smutaneousy. Therefore, the nterference or coson probabty ncreases. The Cogntve Rado (CR [] that enabes hgher spectrum effcency by dynamc spectrum access [2][3] can be expoted by WSN to sove ths probem. The WSN comprsed of sensor nodes equpped wth CR s defned Cogntve Rado Sensor Networks (CRSN [4]. To mtgate the probem of uncertanty n spectrum sensng n a cogntve rado network, cooperatve spectrum sensng can be used. Dfferent technques were proposed for cooperatve spectrum sensng. The smpest method s to use an OR or AND operaton among the receved sensng resuts [5]. Combng technques based on ma rato combnng (MRC and equa gan combnng (EGC were nvestgated n [6], an optma near cooperaton scheme base on a kehood rato test (LRT has been proposed n [7]. In [8], the censor-based cooperatve spectrum sensng has been proposed to save energy. And a censor-based cooperatve spectrum sensng scheme usng Takag and Sugeno s (T-S fuzzy ogc for cogntve rado sensor networks was proposed n [9]. But n these schemes, the CRSN nodes are often assumed to be trustworthy. In practce, there are macous CRSN nodes sendng fase reportng vaues to the Fuson Center (FC [], whch w nduce the FC to make wrong decsons. Based on the prevous work n [9] and [], we proposed a cooperatve spectrum sensng scheme based on trust and fuzzy ogc for CRSN. The nodes use the T-S fuzzy ogc to make oca decsons on the presence or absence of the prmary user s (PU sgna, and then use a censorng method to ony aow the reatvey credbe decsons sent to FC. And a trust scheme on the bass of the factors such as oca sensng dfference, sensng ocaton factors, and sensng channe condtons for each node s mpemented. The FC makes the fna decson accordng to the majorty rue and the trust weghts of the nodes. Ths method can mprove the sensng performance whe saves the node s energy. Copyrght (c 23 Internatona Journa of Computer Scence Issues. A Rghts Reserved.

2 IJCSI Internatona Journa of Computer Scence Issues, Vo., Issue, No, January 23 ISSN (Prnt: ISSN (Onne: Cooperatve Spectrum Sensng Scheme 2. Loca Spectrum Sensng and Decson Suppose each CRSN node usng the energy detecton. The hypotheses f the prmary user s present ( H or not ( H are as foows: n (: t H x ( t ( h(( t s t n(: t H Where x ( t s the receved sgna by the -th node, n ( t s the therma nose, h ( t s the channe gan from the PU to the -th node, and s( t s the transmt sgna from the PU. The oca test statc of the -th node usng energy detecton s: N k 2 x x ( k,,2,, M (2 E Where x ( k s the k -th sampe of receved sgna at the -th node, M s the number of nodes, and N s the number of sampes, N 2TW, where T and W are detecton tme and sgna bandwdth, respectvey [2]. Let the two fuzzy sets Low and Hgh depct the -th node detected energy x E, and ther membershp functons are [2] :, f x 2 E ( ( x E Low xe (3 2 2, otherwse e Hgh ( x E, f x 2 E ( x E (4 2 2, otherwse e And the Fg. shows the shapes of the above functons [9] [2]. Fg. The membershp functons of the two fuzzy sets Based on the T-S fuzzy method, the oca decson of the -th node [9] s: Low( xe Hgh( xe Ld (5 ( x ( x Low E Hgh E And the fuzzy rue sets are [9] [2] : Rue : IF ( xe s Low THEN ( Ld Ld Rue 2: IF ( xe s Hgh THEN ( Ld Ld Where Ld s the oca decson of the -th node, Ld and Ld are the mum and mum vaue of Ld respectvey. Ld means the PU s present and Ld means the PU s absent. Gven the censorng threshod vaue C (,,the oca soft decson Ld s transmtted to the FC f and ony f Ld C or Ld C. 2.2 The Trust Vaue of CRSN Nodes Combne the factors such as the dfference of oca sensng; sensng ocaton, sensng channe condton [], the FC estabshes trust vaue for each node. The dfference of oca sensng s the dfference of the sensng vaue of a snge node wth the average vaue of a nodes. The smaer the dfference, the more reabe of the node. The dfference of oca sensng of the -th node s: Ns D Ld Ld (6 Ns Where Ns s the number of nodes that sent the oca decson to the FC. The sensng ocaton factor ncudes two aspects: the dstance of the -th node to PU, the dstance of the -th node to the FC, and denoted by LOa and LOb respectvey. The sensng ocaton factor s: LO LO LO (, 2,, Ns (7 a b Because the sensng channe s non-dea, so the nformaton transmsson s prone to error. Based on the channe SNR, defne the sensng channe condton factor of the -th node: CC SNR (, 2,, Ns (8 Copyrght (c 23 Internatona Journa of Computer Scence Issues. A Rghts Reserved.

3 IJCSI Internatona Journa of Computer Scence Issues, Vo., Issue, No, January 23 ISSN (Prnt: ISSN (Onne: Carry out the standardzng to the three factors, and get the foowng expressons [] : ' D D D (9 D D ' LO LO LO ( LO LO ' SC SC SC ( SC SC Where D and D s the mum and mum vaue of the sensng ocaton factor respectvey, and the others are smar. The trust vaue of the -the node s: ' ' ' T D LO SC (2 Where. 2.3 Data Fuson at the FC The FC receves Ns oca decsons from CRSN nodes. Accordng to the majorty rue [9], the fna decson fuson rue s: Ns H : Ld jtj H j (3 H : otherwse 3. Smuaton Resuts and Anayss the effect on the detecton performance, and the oca sensng dfference has the mum effect whe the sensng channe condton factor has the greatest effect. The scheme has the best sensng performance when the /3. Goba detecton probabty Pd α=,β=,γ= α=,β=,γ= α=,β=,γ= α=/3,β=/3,γ=/ Goba fase aarm probabty Pf Fg.2. The effect of the,, Secondy, the mpact of the SNR on the performance s evauated. The smuaton was carred out under condtons that the SNRs of the frst sx nodes are -5,- 4,-3,-2,-,- db, respectvey, and the SNR at the 7th node s -4,-2,-,-8,-6 db respectvey, /3,and C.2. From Fg.3, t can be seen that P M ncreases whe the SNR at the 7th node decreases. To evauate the performance of the proposed cooperatve spectrum sensng agorthm for CRSN and compare the performance wth some extng methods, the Monte-Caro smuatons are carred out wth, sampes under the foowng condtons: The number of CRSN nodes M s 7, the PU sgna s key-equay BPSK sgna, The noses at the sensng and contro channes are Gaussan wth zero mean and unt varance, The number of sampes N s 3. Goba mss probabty Pm CU7=-6dB CU7=-8dB CU7=-dB CU7=-2dB CU7=-4dB The performance of the scheme s compared wth the performance of the OR/AND fuson rues, the equa gan combnaton (EGC based scheme [6], and the fuzzy ogc scheme [9]. Frsty, the performance of the factors that oca sensng dfference, sensng ocaton factors, and sensng channe condton s evauated. Let the SNRs for the nodes are - 2,-8,-6,-4,-2, -,-8 db, respectvey, and C.2. From Fg.2, t can be seen that the three factors a have Goba fase aarm Pf Fg.3. The mpact of the SNR Fnay, the smuatons are conducted under the /3, the SNRs for the nodes are -2,-8,- 6,-4,-2,-,-8 db, respectvey, and C.2. The ROC curves of our scheme and compared schemes are Copyrght (c 23 Internatona Journa of Computer Scence Issues. A Rghts Reserved.

4 IJCSI Internatona Journa of Computer Scence Issues, Vo., Issue, No, January 23 ISSN (Prnt: ISSN (Onne: depcted n Fg. 4. It can be seen that the proposed scheme has the hghest detecton probabty than the OR/AND fuson rues based scheme, the EGC based scheme, and the fuzzy ogc scheme. It means that the method of node trust evauaton mproves the FC decson accuracy. Goba detecton probabty Pd Proposed scheme AND rue OR rue EGC Fuzzy ogc scheme Goba fase aarm probabty Pf Fg.4 ROC curves of proposed scheme vs. comparson schemes 4. Concusons By ntroducng the trust scheme nto the fuzzy ogc scheme to represent the reabty of the nodes, a cooperatve spectrum sensng scheme based on trust and fuzzy ogc for CRSN was proposed n ths paper. Anayss and smuatons show that whe keep the merts the fuzzy ogc scheme that reduce the number of nodes reportng ts oca decson to the FC and save the energy of the CRSN nodes, the proposed scheme can mprove the detecton probabty effectvey compared wth the OR/AND fuson rues based scheme, the EGC based scheme, and the fuzzy ogc scheme. It s usefu for the CRSN because the hardware, ow energy restrcton and the performance are favorabe. Acknowedgments Ths research was supported by The Natona Natura Scence Foundaton of Chna. (No. 6234,Shenzhen New Industry Deveopment Fund under grant (No.CXB25252A, the Deveopng Fund for Innovatve Research Team of Guangdong Unversty of Technoogy(No. GDUT2-, and the Ph.D. Programs Foundaton of Guangdong Unversty of Technoogy ( No References [] J. Mtoa, G. Q. Magure, "Cogntve Rados:Makng Software Rados More Persona",IEEE Persona Communcatons,Vo. 6,No. 4, 999,pp. 3-8,. [2] Chaoub, Abdeaa; Ibn-Ehaj, Ehassane, "Comparson between Possonan and Markovan Prmary Traffcs n Cogntve Rado Networks ",Internatona Journa of Computer Scence Issues, Vo.9, No.2, 22, pp [3] Yadav, Shrkrshan;Roy, Krshna Chandra, "Tradeoff Anayss of Bt-Error-Rate (BER n Cogntve Rado Based on Genetc Agorthm",Internatona Journa of Computer Scence Issues, Vo. 9, No., 22, pp [4] O.B. Akan,O.B.Kar,O.Ergu, "Cogntve Rado Sensor Networks",IEEE Networks,Vo.23, No.4, 29,pp [5] Tevfk Yucek, Huseyn Arsan, "A Survey of Spectrum Sensng Agorthms for Cogntve Rado Appcatons",IEEE Communcatons Surveys & Tutoras, Vo., No., 29, pp [6] J.Ma,G.Zhao,Y.L, "Soft Combnaton and Detecton For Cooperatve Spectrum Sensng n Cogntve Rado Networks", IEEE Transactons on Wreess Communcatons, Vo.7, No., 28, pp: [7] Z. Quan, S. Cu, and A. H. Sayed, "Optma Lnear Cooperaton for Spectrum Sensng n Cogntve Rado Networks", IEEE Journa of Seected Topcs n Sgna Processng, Vo. 2, No., 28, pp [8] S.Maek, A.Pandharpande, G.Leus,"Energy-Effcent Dstrbuted Spectrum Sensng for Cogntve Sensor Networks", IEEE Sensors Journa, Vo., No.3,2, pp [9] T.K.XUAN,I.KOO, "A Censor-Based Cooperatve Spectrum Sensng Scheme Usng Fuzzy Logc for Cogntve Rado Sensor Networks", IEICE Transctons on communcatons, Vo. E93-B, No.2, 2, pp [] W.Lang, Y.Zhu, H.L,"Trust Leve Based Byzantne Attack- Excuson Cooperatve Spectrum Sensng Scheme for Wreess Cogntve Sensor Networks n Smart Grds",n 22 IEEE Internatona Conference on Cyber Technoogy n Automaton, Contro, and Integent Systems (CYBER, 22, pp:2 25. []Zhang Guanghua, Dng Renshuang, Huang Ljng, "Usng Trust to Estabsh Cooperatve Spectrum Sensng Framework",Proceda Engneerng, Vo.5, 2,pp [2]T.K.XUAN,I.KOO, "Cooperatve Spectrum Sensng usng Kaman Fter based Adaptve Fuzzy System for Cogntve Rado Networks", KSII Transctons on Internet and Informaton System,Vo. 6, No., 22, pp Yonghua Wang has receved hs B.S. degree n Eectrca Engneerng and Automaton from Hebe Unversty of Technoogy n 2, the M.S. degree n Contro Theory and Contro Engneerng from Guangdong Unversty of Technoogy n 26, and Ph. D degree n Communcaton and Informaton System from Sun Yat-sen Unversty n 29.Currenty he s a ecturer n schoo of automaton n Guangdong Unversty of Technoogy snce 29. Hs current research nterests ncude cogntve networks, RFID, etc. Yuehong L has receved hs B.S. degree n Eectrca Engneerng Copyrght (c 23 Internatona Journa of Computer Scence Issues. A Rghts Reserved.

5 IJCSI Internatona Journa of Computer Scence Issues, Vo., Issue, No, January 23 ISSN (Prnt: ISSN (Onne: and Automaton from Hangzhou Danz Unversty n 2.He s now workng towards hs M.S. degree n Contro Theory and Engneerng at Guangdong Unversty of Technoogy. Fe Yuan has receved hs B.S.degree n Communcaton Engneerng and M.S. degree n Contro Theory and Contro Engneerng from Guangdong Unversty of Technoogy n 27,2 respectvey. He s now workng towards hs Ph.D. degree n Communcaton and Informaton system at Sun Yat-sen Unversty. Hs research nterests ncude wreess networks,data ng, etc. Jan Yang has receved hs B.S. and M.S. degree n Bomedca Engneerng from Northeastern Unversty n 23,26 respectvey. He has receved Ph. D degree n Communcaton and Informaton system from Sun Yat-sen Unversty n 29. Currenty he s a ecturer n schoo of automaton n Guangdong Unversty of Technoogy snce 29. Hs research nterests ncude RFID,wreess communcatons etc. Copyrght (c 23 Internatona Journa of Computer Scence Issues. A Rghts Reserved.