Malicious Recognition and Exploitation of Byzantine Attack for Cooperative Spectrum Sensing in Cognitive Wireless Sensor Networks

To alleviate the problem of spectrum scarcity, cognitive radio (CR) technology enables multiple secondary sensors (SSs) exploiting cooperative spectrum sensing (CSS) to opportunistically utilize the licensed spectrum of primary users (PUs). However, the open nature of CSS makes it highly vulnerable to Byzantine attacks launched by malicious sensors (MSs), which can falsify sensing reports to deliberately degrade critical CSS performance metrics such as detection accuracy and reliability—posing a significant barrier to the practical deployment of CR networks. To address these challenges, a defense paradigm that encompasses attack pattern recognition and attack exploitation is proposed in this paper. At the beginning, we propose a probabilistic Byzantine attack model in cognitive wireless sensor networks (CWSNs) and extract features from both the historical reports and historical reputation values of all SSs. Based on these features, the support vector machine (SVM) is then employed to identify MSs and classify its attack patterns. As a consequence, leveraging the identified attack patterns, the malicious reports from these MSs are utilized to further enhance the CSS performance. At last, a series of numerical simulation results demonstrated the proposed method performs well in both the identification and exploitation of Byzantine attack.