Multiple Features-Aided Malicious Users Detection in OFDM-based Cognitive Radio-Internet of Things

Cooperative Spectrum Sensing (CSS) is a beneficial approach to alleviate the spectrum scarcity issue by making it possible for the unlicensed users including normal and Malicious Users (MUs) to share the allocated available spectrum in a Cognitive Radio-based Internet of Things (CR-IoT). Nevertheless, the detection performance has been compromised due to this approach being responsive to malicious attacks where each MU sends their fake sensing data to a Fusion Center (FC). In this paper, we propose multiple features-aided Machine Learning (ML) to accurately detect MUs in Orthogonal Frequency Division Multiplexing (OFDM)-based CR-IoT networks. Moreover, we solved the sum rate maximization problem for both primary and secondary networks using the golden section search method. Comparing the proposed approach to other conventional schemes, the numerical results demonstrate that it achieves superior detection gains, lower error rate, better system sum rate, and enhanced security where MUs are accurately detected.