Spectrum Sensing In Cognitive Radio

Most of the current wireless communication services are fixed on a rigid spectrum allocation scheme, which leads to the spectrum shortage. Cognitive radio, as a revolutionary intelligence technology, can maximize the utilization of the frequency resources by allowing cognitive users access the allocated spectrum bands which are idle spatially or temporally. Spectrum sensing plays a vital role in cognitive radio realization, which needs to exploit the existence of unused spectrum bands and keep monitoring the activation of primary user in order for the cognitive users to vacate the occupied spectrum segments when an incumbent signal appears. In any practical communication system, background noise is neither Gaussian and white perfectly, nor stationary perfectly, and A/D converters have finite precision. These uncertainties will substantially degrade the sensing reliability and robustness of a detector. The dissertation obtains some simulation results which show that energy detector based on adaptive-threshold and double-threshold outperforms conventional energy detector.In cognitive radio networks, cooperative spectrum sensing can mitigate the effects of shadowing and fading, and reduce the detection time. The dissertation discusses the parameter selection when using k-out-of-N fusion and weighting scheme in cooperative sensing.The performance of most detecting schemes is vulnerable to the interference from the on-going transmission of the cognitive user networks. The direct way of avoiding this kind of influence is to set a series of quiet period to halt data transmission when sampling for detection. However, quiet period will cause lots of overheads and affect the quality-of-service of cognitive systems. The dissertation proposes a heuristic approach without quiet period based on estimating the number of source signals. When using the proposed algorithm, almost no prior information about primary signal and transmission channel is needed. The simulation results show the proposed scheme can almost be implemented in every situation, as long as the SNR is not very low, and the method possesses high practical value.