An Adaptive Learning-Based Attack Detection Technique for Mitigating Primary User Emulation in Cognitive Radio Networks

Abstract

Cognitive radio (CR) technology is designed to improve reliability in communication between users through efficient and dynamic spectrum exploitation. CRs address the problems in spectrum allocation and channel access and improve the rate of radio resource utilization. The flexibility of the CR networks (CRN) and communication medium exposes it to a variety of threats; primary user emulation attack (PUEA) is a malicious and denial-of-service kind of adversary that defaces CRN performance. This manuscript proposes an adaptive learning-based attack detection in CRN for detecting and mitigating PUEA by analyzing the received power of the transmitter. The learning process endorses some beneficial features by distinguishing low spectrum legitimate PU from an adversary. The learning process adopts cyclostationary feature analysis for distinguishing adversaries and low power PU in CR communications. The process of learning is further enhanced by estimating distance variance and communication time-based analysis for improving the rate of signal classification and SU communication rate. The experimental analysis proves the stability of the proposed detection method by improving the SU throughput, with lesser signal classification time and misdetection probability.