Research On Technologies Of Spectrum Sensing In Multi-Antenna Cognitive Radio Networks

Cognitive radio networks (CRN) are considered to solve the problems of the current low spectrum utilization rate. Multi-antenna technology can bring the benefits of the spatial diversity and spatial multiplexing. Meanwhile, the combination of cognitive radio technology and mul-ti-antenna technology has broad prospects in the research. In the spectrum overlay system, CR needs accurate spectrum sensing before accessing the idle spectrum. By introducing mul-ti-antenna technology into traditional spectrum sensing, sensing performance can be enhanced significantly by fighting against fading. But the problem of communication overhead still exists, and many new problems like sensing scheduling, sensing-throughput trade off and energy effi-ciency are introduced into a new multiple antenna model at the same time. Some related discus-sion and research are done in this dissertation to solve the above problems. The main contribu-tions are outlined as follows:To solve the current contradiction between sensing performance and cooperation overhead of cooperative spectrum sensing, a hierarchical fusion cooperative spectrum sensing scheme based on double thresholds with multi-antenna cognitive radios is proposed in this paper. Addi-tional improvement of local sensing performance of CRs is provided by taking advantage of space diversity gain for employing square law combining (SLC) with multiple antennas. The local sensing results are divided into one-bit and two-bit decisions by setting thresholds. And then by limiting the lowest number of cooperating CRs through fusion center (FC), CRs make a hierarchical report and fusion center (FC) makes a hierarchical fusion based on the sensing report. Simulation results show that the proposed scheme achieves a good tradeoff between sensing performance and communication overhead and is more robust to the distance between double thresholds. In addition, when most of the detected values fall between the two thresholds, the probability of sensing failure problem caused by malfunctioning CR is greatly reduced.In order to make full use of space diversity, an optimization model is proposed for mul-ti-channel coordinated spectrum sensing in multiple antenna based CRN on the basis of flexible selection of antenna combination to apply spatial diversity. Branch and bound based method and greedy based heuristic method are proposed to solve the problem, both with the constraint of the maximum false alarm probability of each channel and goal of minimizing the sum of misdetec-tion probability over all the channels. Besides, the effects of the number of CRs and the value of signal to noise ratio are studied on the performance of two methods. The former can get the op-timal strategy with high complexity, while the latter can reduce the complexity obviously at the price of little loss of performance of detection and achieve a good trade-off between performance of detection and complexity, and it cares about the fairness to a certain degree on protecting pri-mary users from interference of cognitive radios on each channel.Considering sensing-throughput trade off and energy efficiency, a new sensing frame of multi-antenna CR is proposed based on flexible spatial diversity. A nonconvex optimization model is formulated and two algorithms (outer linearization (OL) in conjunction with branch and bound algorithm and genetic algorithm (GA)) are introduced to solve it. Both algorithms solve the optimization problem of the sensing threshold, antenna assignment, sensing time and transmit power of CRN system based on CR with multiple antennas for maximization of the op-portunistic system throughput under the constraints of the maximum sensing energy, probability of missed detection and transmit power in each sensing period. The former can get the better performance than genetic algorithm with high complexity, while the latter can reduce the com-plexity obviously at the price of little loss of opportunistic throughput. Compared with some current methods, the new sensing frame improves the overall performance of system obviously. Besides, the effects of the number of CRs, the value of signal to noise ratio and the length of si-lent sensing time are studied on the performance of two methods.