Abstract
The rapid development and deployment of Unmanned Aerial Vehicles (UAV)-enabled wireless networks, while promising capabilities in providing ubiquitous connectivity and extended coverage, introduces critical security vulnerabilities that demand robust countermeasures. This paper presents a comprehensive framework for Physical Layer Authentication (PLA) in UAV communications, leveraging machine learning techniques to enhance authentication and threat detection mechanisms. We propose a novel approach that integrates Support Vector Machine (SVM) classification with traditional PLA methods to identify and mitigate security threats in real time. Our framework exploits the inherent properties of wireless channels and signal characteristics in UAV communications to establish reliable authentication protocols. Experimental results demonstrate that our proposed system achieves a 97% detection accuracy for various attack vectors, including eavesdropping and attacking attempts, while maintaining low computational overhead. The framework's effectiveness is validated through extensive simulations in diverse UAV communication scenarios, showcasing its potential for the advancement of secure UAV communications by establishing a robust and adaptive data-driven security paradigm.