Research On Spectrum Sensing In Cognitive Radio Networks With Adaptive Threshold Control And New Cooperative Sensing Algorithm

Nowadays the mainstream of spectrum allocation strategy is static allocation, using licensing system mostly.Many authorized users, however, leave the spectrum idle, which directly led to the low spectrum utilization rate.In such cases, Joseph Mitola in 1999 proposed the concept ofCognitive Radio(CR) on the basis of Software Radio, describing the opportunity-accessmethod.In order to sense the spectrumdynamically, the first problem for secondary user(SU) to be solved is how to effectively detect spectrum hole, while avoidinginterference to the primary user(PU).In this paper, we focus on the spectrum sensingin cognitive radio networks, and proposed the algorithm to maximize utilization rate of spectrum with the interference limitedrespectivelyin the case of single-userand multi-user.With the deep study of the mathematical models, we put forward the innovative research results as follows:1. Proposed a power detection algorithm adopting the adaptive threshold control(ATC) technique.Although the traditional energy detection algorithm using fixed decision threshold can guarantee the interference to the primary users in a certain range, but do seldom with the utilization of spectrum.The adaptive threshold controlalgorithm can maximize the spectrum utilizationby dynamically adjusting energy decision threshold to improve the detection performance.2. Proposed a joint detection algrithm based on the power set belief.The traditional joint detection techniquesuch as theK/N joint detection algorithm, lackstheknowledge of the accuracy ofeach user.While the power set belief,which develops from the Dempster-Shafer theory, fuses the decisions of all users effectively according to the belief of them.3. Proposed a joint detection strategy based on Adaboost algorithm.Adaboost works in the region of image recognition, artificial intelligence, etc. The joint detection combined with Adaboost can achieve a much better detection performanceby trainingwith samples.