Genetic Algorithm Based Cooperative Spectrum Sensing Optimization in the Presence of Malicious Users in Cognitive Radio Networks

Abstract

The spectrum scarcity problem is rectified in cognitive radio network by allowing opportunistic secondary users (SUs) to utilize primary user spectrum with minimum disturbance. However, multipath effects degrade the sensing capability of an individual user. Therefore, more precise sensing is obtained by collaborating multiple sensing users. In the centralized Cooperative Spectrum Sensing (CSS), fusion center (FC) collects sensing information of all individual users for a global decision. The problem in CSS is the presence of inaccurate sensing information received by the FC from the multipath affected SUs and malicious users. A Genetic algorithm-based scheme proposed in this paper is able to determine optimum weighting coefficient vector against the SUs sensing. The vector is further utilized in the soft decision schemes that assign appropriate weight to the reports of cooperative users to take a global decision. Low weights are assigned to the sensing information of compromised users with false spectrum sensing data as compared to the normal cooperative users. Simulation results illustrate the minimum error probabilities for the proposed GA based technique at different levels of signal-to-noise ratios (SNRs) against the Kullback-Leibler (KL) divergence, count decision scheme and maximum gain combination (MGC) schemes.