Weighted Deep Learning Implementation for Cognitive Radio Attack Detection

The marked rise of wireless devices and the Internet of Things devices has increased the demand for wireless spectrum, leading to spectrum scarcity. Cognitive Radio Networks (CRNs) enable secondary users (SUs) to use the underutilized and unused frequency bands but are vulnerable to security threats like Byzantine and Primary User Emulation (PUE) attacks, disrupting spectrum sensing. This paper investigates the use of a novel integrated spectrum sensing algorithm combining weighted sensing with Deep Q-Networks (DQN) and Generative Adversarial Networks (GAN) to optimize spectrum utilization and counteract these attacks. Implemented in Google Colab and validated in MATLAB, the experimental analysis demonstrates that the integrated algorithm significantly outperforms the basic weighted algorithm. The integrated approach achieved 99.03% accuracy and reduced false alarm and miss detection rates by 83.85% and 87.10%, respectively. The implications of incorporating these methods in CRNs are profound, offering potential advancements in reliable and efficient spectrum sensing. Overall, key findings suggest that integrating these methods can lead to enhanced robustness against security threats, contributing to the advancement of CRN technologies. The implications of incorporating these methods in CRNs are profound, offering potential advancements in reliable and efficient spectrum sensing by effectively mitigating Byzantine and PUE attacks. The proposed approach significantly enhances decision-making and robustness against security threats.