Detection Strategy Against Restricted SSDF Attack With Potential Interaction Assistance

Abstract

This paper investigates the malicious users (MUs) detection problem against restricted spectrum sensing data falsification (RSSDF) attack in cooperative spectrum sensing (CSS) network. We consider a CSS network with non-ideal reporting channels, in which there are two kinds of secondary users (SUs): interactive SUs (ISUs) and non-interactive SUs (NSUs). We simplify the interaction process among ISUs with proper assumptions and formulate the impact of non-ideal reporting channels on falsified data. Based on the traditional probability SSDF attack on M-ary quantized data, we generalize the attack model by considering the restriction on attack magnitude. To detect ISUs, we construct a detection architecture with ISUs partitioned into groups of two and perform distributed detection by proposed interaction-based detection algorithm. For the detection of NSUs and undetected ISUs, we further propose a semi-supervised fuzzy c-means-based (SSFCM-based) detection algorithm with potential interaction assistance. Simulation results demonstrate the validity of the proposed theorems. In addition, we illustrate the detection performance of the proposed strategy in different conditions and show the superiority of the proposed strategy over other detection algorithms.