Research On Centralized Cooperative Spectrum Sensing Algorithm For Cognitive Radio

With the high-speed development of wireless communication technologies andrapid increasing of wireless users, the scarcity of spectrum resources has been aserious problem. Cognitive radio based on software radio, is an intelligent wirelesstechnology which can improve the spectrum utilization effectively. In order to makefull use of the spectrum resources, cognitive radio can use the idle spectrum whichhas been allocated to the authorized user but not temporarily used throughcontinually sensing the external spectrum environment. Spectrum sensing as thecore of cognitive radio needs to identify the idle spectrum quickly and accurately.Since the single-user spectrum sensing will bring hidden terminal problem and thedistributed cooperative spectrum sensing will increase the designed complexity ofterminal, the centralized cooperative spectrum sensing is investigated in thisdissertation. Through the methods of threshold optimization, clustering, weighing,and double thresholds, the performance of cooperative spectrum sensing can beimproved. Through the optimization of periodic cooperative spectrum sensingmechanism, the sensing algorithms can be used in the actual scene effectively.Centralized cooperative spectrum sensing has two decision fashions: harddecision and soft decision. In hard decision, each cooperative cognitive user sends1-bit decision information to the fusion centre. The overhead of hard decision islittle, but since the fusion centre may acquire less information of authorized userand the combination cannot obtain enough information capacity, the detectedperformance of hard decision is restricted. In soft decision, each cognitive usersends the observed value of authorized user to the fusion centre, and since thecombination algorithm of the fusion center collects enough information ofauthorized user, the detected performance of soft decision is preferable, but theinformation capacity sent by each cognitive user is large.This dissertation respectively analyzes and investigates the algorithms of thecooperative spectrum sensing based on hard decision and soft decision in detail.After a full account of the influence of detected threshold, channel state, andsingle-user detection on cooperative spectrum sensing, the dissertation mainlyfocuses on the designing of the sensing algorithms, and the analysis and simulationof the sensing performance. In addition, in order to further apply the proposedsensing algorithm in the actual scene, the dissertation proposes the mechanism ofperiodic cooperative spectrum sensing, and the parameters of the sensingmechanism such as sensing period, local sensing time, number of cooperative users and search time etc, are respectively optimized and analyzed.Firstly, according to the type of spectrum sensing for cognitive radio includingtransmitter spectrum sensing and receiver spectrum sensing, the existing methods ofsingle-user spectrum sensing are analyzed and compared. The dissertation explainsthe implemental principle of each sensing method in detail, and compares theadvantages and disadvantages of these methods. The dissertation further analyzesthe hidden terminal problem brought by the single-user spectrum sensing, andintroduces the cooperative spectrum sensing which is used to conquer the hiddenterminal problem in detail.Secondly, in allusion to the spectrum sensing based on hard decision, thethreshold optimization algorithm of cooperative spectrum sensing by “AND Rule”,“OR Rule”, and “K-OUT-N Rule” is proposed. In this algorithm, each cognitiveuser adopts the local optimal threshold according to its received SNR and noisevariance, and the simulation shows that compared with the traditional cooperativespectrum sensing algorithm in which all the cognitive users adopt the uniformthreshold, when the received SNR and noise variance of each user is uniform, theperformance of the proposed algorithm slightly decreases, however, when thereceived SNR and noise variance of each user is different, the performance of theproposed algorithm can increase observably. Since the channel fading may decreasethe performance of cooperative spectrum sensing, the clustering cooperativespectrum sensing algorithm is proposed. In this algorithm, cognitive users aredivided into several clusters, and the cluster heads near the fusion centre send thedecision information of their local clusters to the fusion centre. The simulationshows that compared with the traditional cooperative spectrum sensing by “ORRule”, the detection probability of the proposed algorithm keeps invariable inperfect channel but increases in fading channel.Thirdly, according to the different influence of the decision result of each useron the combination decision of cooperative spectrum sensing based on soft decision,the algorithm of weighed cooperative spectrum sensing based on soft decision isproposed, and the algorithm represents the contribution of each user to thecombination decision through allocating the different weights to the cognitive users.In allusion to the military scene and civil scene, the dissertation respectivelyproposes the two algorithms of weight allocation including weighing based onmaximizing throughput and weighing based on minimizing interference capacity. Inaddition, the dissertation extends the algorithm to the wideband cooperativespectrum sensing, and proposes the weighed wideband cooperative spectrumsensing algorithm. The simulation shows that the proposed algorithm can achieve larger throughput, produce less interference to the authorized user, and get lessinfluence of the channel fading. In order to obtain the tradeoff between theinformation capacity sent by cognitive user and the detected performance ofcooperative spectrum sensing, the dissertation proposes the weigheddouble-threshold cooperative spectrum sensing algorithm. This algorithm combineshard decision and soft decision, and its sensing performance is between those ofhard decision and soft decision. The simulation shows that the detection probabilityof the proposed algorithm is higher than that of hard decision, while the sentinformation capacity of the proposed algorithm is less than that of soft decision.Lastly, in order to apply the cooperative spectrum sensing algorithms to theactual scene better, the mechanism of periodic single-channel cooperative spectrumsensing is proposed. In this mechanism, the communication process of cognitiveradio is divided into several periods, and in each period the cognitive user firstlysenses the authorized user. If the absence of the authorized user is detected, thecognitive user can transmit data during the following time of this period, otherwiseif the presence of the authorized user is detected, the cognitive user needs to searchfor another idle channel. In order to improve the performance of the spectrumsensing mechanism, the parameters of the sensing mechanism such as sensingperiod, local sensing time, number of cooperative users, and search time arerespectively optimized. The simulation shows that through the optimization of thesensing mechanism, cognitive radio can decrease the interference to the authorizeduser, improve the throughput, and decrease the search time. The dissertationexpands the periodic cooperative spectrum sensing mechanism to the widebandcooperative spectrum sensing, and the periodic wideband multi-slot cooperativespectrum sensing mechanism is proposed. Through the joint optimization of thenumbers of the sensing slots and cooperative users, the throughput of widebandcognitive radio can be improved observably with the increasing of the number ofthe sub-channels.