Malicious User Detection with Local Outlier Factor during Spectrum Sensing in Cognitive Radio Network

Abstract

In collaborative sensing, multiple secondary users (SUs) cooperate for a more accurate sensing decision to detect spectrum holes in cognitive radio networks (CRNs). This technique, however, can be adversely affected by malicious users (MUs) who route falsified spectrum sensing data to the fusion centre (FC). This attack is known as the spectrum sensing data falsification (SSDF) attack. The task of the FC is to aggregate local sensing reports and is thereby responsible for making the final sensing decision. In this paper, we propose a detection and isolation scheme based on local outlier factor (LOF) to detect and reduce the unfavourable effects of SSDF attack. The key feature of this scheme is that for each SU a metric is calculated, which is called the LOF. Based on the LOF, a decision is made about whether an SU is an attacker or not. We support the validity of the proposed scheme through extensive simulation results.