Machine Learning Based Primary User Emulation Attack Detection

The rapidly growing demand for IoT applications requires the widespread use of cognitive radio technologies. However, modern wireless communication systems have a large number of vulnerabilities. Malicious nodes can cause heavy performance degradation by DoS attacks. Thus, the problem of developing effective protection mechanisms is quite relevant. In this paper, we consider one of the most destructive DoS attacks in cognitive radio networks called the primary user emulation attack. We offer an effective approach to intrusion detection based on machine learning, suitable for deployment on low-resource network nodes. Moreover, the proposed scheme is compared with several baselines methods by using the metrics of accuracy, precision, recall, and F1 score, where the proposed method achieved the best results.