Study Of Spectrum Sensing Based On Brain Storm Optimization Algorithm

Cognitive radio technology (CR) is that the cognitive users (CUs) can dynamically perceive idle spectrum and opportunistic access the spectrum without interference to primary users (PUs). Spectrum sensing technology is the basis and precondition of CR and has important significance for improving the spectral efficiency. The combination of spectrum sensing and swarm intelligence algorithm can improve the detection probability of spectrum sensing greatly. Brainstorming algorithm is a new and effective intelligent algorithm by mimicking human problem-solving process which is suitable on solving high-dimensional multi-peak function optimization problems. The algorithm has strong global search capability with clustering operation and increase the diversity of the solution with creating operation.But the study of the algorithm is in its infancy, there are some problems to be solved, for example, slow speed and poor precision of convergence. Based on the above shortcomings and with a detailed analysis about brainstorming algorithm,this paper proposes someimprovements.Firstly, analysis the evolution of differential evolution (DE) algorithm, which is used to the creation operation of the brainstorming algorithm, and propose three different update strategies, so that the improved algorithm can have effective search in the global scope .Secondly, with the double evolutionary mechanism of cultural algorithm (CA), embed the brainstorming algorithm into cultural algorithm framework and propose cultural brainstorming algorithms. In this proposed algorithm, the knowledge of belief space (BS)guides the evolution of the population in the main population space (MPS), thereby ensuring evolution of the algorithm to the direction of optimal solution.Finally, test the brainstorm algorithm and two improved algorithms through four benchmark functions and compare with other intelligent algorithms; the improved algorithms are applied to the linear model of cooperative spectrum sensing, and spectrum sensing model is transformed into a fitness function to be optimized. The simulation results show that:compared with other intelligent algorithms,two improved brainstorming algorithms have a strong ability to exploration, and improve the performance of brainstorming algorithm significantly; get a high probability of detection in spectrum sensing problem, and improve the detection performance of spectrum sensing problem greatly.