Research On Technology Of Spectrum Sensing In Multi-antenna Cognitive Radio System

Scarcity of spectrum resource becomes increasingly critical due to the rapid growth ofwireless communications and sharp accretion of wireless communication business. Thedemand-supply contradiction of spectrum resource, which is the bottleneck of the growth ofwireless communications, needs to be solved urgently. Cognitive radio is a promising techniqueto deal with this problem, and has been of catholic and deep concern.A critical presupposition of cognitive radio is spectrum detection, which is the basis ofopportunistic spectrum access. Currently existing spectrum detection techniques can be classifiedinto three categories: standalone detection, cooperative detection and blind detection. In thispaper, these techniques are combined with spatial diversity capacity, aiming at studyingmulti-antenna spectrum detection method which can improve the performance of spectrumdetection further. Main work in this paper is organized as follows:1. The design principle as well as technical limit of conventional standalone detection andcooperative detection in cognitive radio spectrum detection technology is analyzed. The cut-inpoints of this paper is provided based on these analyses, which is the spadework of the followingstudy in multi-antenna spectrum detection technique.2. Study on the issue that conventional energy detection algorithm is difficult to determinatethe decision threshold and the performance is susceptible to noise uncertainty, and a AIC-CSSalgorithm is provided. The proposed algorithm applies the idea of spatial spectrum estimationinto spectrum detection. Using eigen-values obtained by Eigen-decomposition of receivingcorrelation matrix, spatial distribution of energy can be characterized and the existence ofprimary users can be determinate with AIC. Due to the idea of multi-antenna joint detection andjudgment of relative value, the algorithm gains robustness without setting the noise threshold.Simulations show that AIC-CSS algorithm can mitigate the influence of noise uncertainty, andperforms better than energy detection algorithm.3. Study on the issue that the performance of multi-antenna multi-user cooperative detectionmaybe affected by error transmission through the imperfect control channel between cognitiveusers and fusion center, and propose an optimal algorithm computing the amount of cooperativeusers. Targeting at minimizing the error detection probability with consideration of systemtransmission overhead, the optimal threshold, the optimal combination criterion and theclosed-form solution of optimal amount of cooperative users are derived respectively and theeffect of control channel error probability on system detection performance using the idea oflayered optimization and the method of maximum likelihood estimation. Cognitive users cancalculate statistical estimated energy value of each receiving antenna, then derives optimaldecision threshold to minimize the error detection probability. Theoretical analysis andsimulation results demonstrate that the proposed scheme is correct, and has better detectionperformance comparing with conventional cooperative detection algorithm.4. Study on the issue of low throughput of cognitive system and detection efficiency whencognitive radio detects every single channel and propose a multi-channel joint spectrum detection algorithm based on improved genetic algorithm. Given the detection probability, aoptimization model of optimize the amount of detection antennas allocated to each channel tomaximize cognitive channel total throughput is build. Considering non-convex relation betweenobjective function and antenna amount, genetic algorithm is introduced to solve the optimalantenna allocation method. Individual diversity is guaranteed by migration strategy, thus geneticalgorithm can evolve and recombine normally and early convergence is restrained, gaining betterperformance of total throughput. Simulations show that proposed algorithm almost close to theoptimal allocation algorithm, it can greatly improve cognitive channel total throughputcomparing with random allocation strategy.