Mitigating malicious attacks using Bayesian nonparametric clustering in collaborative cognitive radio networks

Reliable detection of primary users is an important task in cognitive radio. It becomes challenging in the presence of malicious users' sharing false sensing data in a collaborative spectrum sensing. In this paper, we propose a Bayesian nonparametric clustering approach to estimate the primary user's channel behavior and identify malicious users' collaborative spectrum sensing. The proposed scheme clusters malicious attacks in a Bayesian nonparametric way and identifies malicious users. From the simulation results, we demonstrate the effectiveness of the proposed approach by using real wireless traces and comparing with the nonparametric mean-shift clustering approach.