One-to-Many Relationship Based Kullback Leibler Divergence against Malicious Users in Cooperative Spectrum Sensing

Abstract

The centralized cooperative spectrum sensing (CSS) allows unlicensed users to share their local sensing observations with the fusion center (FC) for sensing the licensed user spectrum. Although collaboration leads to better sensing, malicious user (MU) participation in CSS results in performance degradation. The proposed technique is based on Kullback Leibler Divergence (KLD) algorithm for mitigating the MUs attack in CSS. The secondary users (SUs) inform FC about the primary user (PU) spectrum availability by sending received energy statistics. Unlike the previous KLD algorithm where the individual SU sensing information is utilized for measuring the KLD, in this work MUs are identified and separated based on the individual SU decision and the average sensing statistics received from all other users. The proposed KLD assigns lower weights to the sensing information of MUs, while the normal SUs information receives higher weights. The proposed method has been tested in the presence of always yes, always no, opposite, and random opposite MUs. Simulations confirm that the proposed KLD scheme has surpassed the existing soft combination schemes in estimating the PU activity.