An Optimized Cyber-Physical Dataset of Attack Detection for Internet of Drones

This research addresses the growing need for advanced security mechanisms in Unmanned Aerial Vehicle (UAV) environments, particularly within the Internet of Drones (IoD) framework. As UAVs become central to critical applications such as surveillance, logistics, and emergency response, they are increasingly vulnerable to both cyber and physical attacks. Traditional Intrusion Detection Systems (IDS) often rely on either cyber-only datasets , which are simulated and lack physical context, or real-time physical datasets, which are incomplete and lack attack variety. This project proposes the use of the \u201cCyber-Physical Dataset for UAVs Under Normal Operation and Attacks (2023)\u201d\u2014a dataset that uniquely simulates realistic cyberattacks (such as DoS, MITM, and GPS spoofing) alongside physical UAV telemetry. The study conducts a comprehensive comparison of available datasets, identifies key limitations such as lack of attack evolution and risk classification, and proposes enhancements using adaptive labeling, synthetic data augmentation, and risk-impact modeling. By integrating cyber and physical data dimensions, the proposed approach enables the development of intelligent IDS models that can detect and respond to complex multi-dimensional threats. The outcome contributes toward robust, real-time UAV defense mechanisms and sets a foundation for future research in secure drone systems.