Research On The Spectrum Sensing In Cognitive Radio

Cognitive radio technology can sense the radio environment in outside world, discover and use the unused spectrum bands in an opportunistic way in order to relieve the contradiction between radio resources and scarcities. Spectrum sensing is one of the key technologies of cognitive radio. The research on this dissertation is focused on spectrum sensing in cognitive radio.First, the research background and significance of spectrum sensing are introduced, besides, the technologies of narrowband spectrum sensing and wideband spectrum sensing are analyzed and summarized respectively.Secondly, spectrum sensing technologies based on the random matrix theory are studied. Some basic narrowband spectrum sensing methods are introduced. In order to enhance the performance of the difference between the maximum and the minimum eigenvalue (DMM) algorithm, an improved cooperative sensing algorithm based on heightening the estimation accuracy of eigenvalue is presented. It can improve the matrix dimensions and increase the number of logic signals by the decomposition and recombination (DAR) of the sensing signals. Depending on the kind of decompositions, it can be divided into order-DAR and interval-DAR, which have a different influence on the correlation of signals. Theoretical analysis and simulations all show that the performance of the proposed method is superior to that of the DMM algorithm while keep the same complexity. Then, a spectrum sensing algorithm using double eigenvalue limiting distributions is presented. This algorithm uses the double eigenvalue to estimate a threshold based on the double eigenvalue limiting distributions to heighten an accuracy of the threshold. Simulations results show that the proposed algorithm can gain better performance against the short of sampling number and low false alarm probability than the DMM algorithm.Next, spectrum sensing technologies based on the stochastic resonance are presented. It is proved that the bistable stochastic resonance (BSR) system can enlarge the power of signal and improve signal noise ration (SNR) in theory. Then, in order to resolve the problem that conventional BSR system is only suitable for low frequency signals, re-scaling frequency technique is adopted to compress the frequency. Simulations show that the performance of the proposed method is superior to that of DMM algorithm for the same probability of false alarm under low SNR. Meanwhile, in order to validate decompositions have different influence on various algorithm further, simulation experiments are designed for DMM algorithm and energy detection algorithm respectively.At last, the wideband spectrum sensing based on compressed sensing is researched. According to the small variety of received signals in a short timeslot, the orthogonal matching pursuit spectrum sensing algorithm based on differential signal is proposed. It differentiates the previous signal and current signal, and detects its spectrum to acquire the information of channel variation, then, the current channel state is exclusive OR of the previous channel state and channel variation, and the current occupied spectrum can be obtained by the channel state. On account of the estimation error of sparsity, a spectrum sensing algorithm using weak matching pursuit based on Bayesian predictive densities (BPD) is proposed. A penalty function is derived by applying BPD, and the weak matching strategy is applied, which could enhance the estimation performance of spectrum support set and reduce the influence of estimation error of sparsity. Simulations results show that both proposed algorithms can greatly achieve a better performance.