UAVIDS-2025: A Benchmark Dataset for Intrusion Detection in UAV Networks Using Machine Learning Techniques
Qingli Zeng, Abdalrahman Bashir, Farid Naït‐Abdesselam · 2025
Unmanned Aerial Vehicle (UAV) networks have be-come increasingly prevalent across various sectors, including mil-itary operations, disaster management, and commercial services. However, these networks face unique security challenges due to their mobility, resource constraints, and wireless communication vulnerabilities. Despite the growing importance of UAV security, there exists a significant gap in specialized datasets for UAV network intrusion detection research, forcing researchers to rely on general network security datasets that fail to capture the unique characteristics of aerial networks. In this paper, we present UAVIDS-2025, a comprehensive UAV Network Intrusion Detection dataset specifically designed to address this critical gap. Using NS3 simulation, we modeled realistic UAV network environments under both normal operations and various attack scenarios. We evaluated multiple machine learning algorithms on UAVIDS-2025, achieving promising performance metrics that demonstrate the dataset's utility for developing specialized intrusion detection systems. To promote reproducibility and advance research in this domain, we have made UAVIDS-2025 publicly available. This novel dataset provides researchers with a dedi-cated resource for developing and evaluating security solutions tailored to the unique requirements of UAV networks.