ReD-FAN-MiD: Generating a Reference Dataset for Misbehavior Detection in FANETs

Unmanned Aerial Vehicles (UAVs) have become a major enabling technology for various fields. Their versatile and cost-effective nature have revolutionized the way we approach critical and life-threatening missions. The use of Flying Ad Hoc Networking (FANET) was a logical response to the growing demand for enhanced operational capabilities in cooperative and autonomous missions. However, despite the rapid evolution of critical FANET applications and the need for precise and accurate data exchange, security and privacy aspects are often overlooked. Consequently, FANET may carry malicious nodes, potentially leading to substantial damage to the communication system and the overall FANET. Therefore, in order to ensure the proper use of the system by the authenticated UAVs, a misbehavior detection mechanism becomes indispensable. The role of this latter is to monitor and analyze the behavior and the contextual information to detect sophisticated insider attacks and enable vehicle revocation. So far, several studies have aimed to solve this problem, yet, in most cases, the authors give very little information about the experiment scenarios, as a result, it is difficult to compare, reproduce, or validate the findings due to the lack of a reference dataset. For this matter, a well-structured and representative dataset that provides a complete vision of entity activities within the FANET is of paramount importance, especially since datasets have become an increasingly valuable resource for the training and the validation of learning-based misbehavior detection models. In this paper, we propose a Reference Dataset for FANET Misbehavior Detection (ReD-FAN-MiD) as a synthetic preprocessed dataset to compensate for the lack of data in the area of FANET misbehavior detection. The ReD-FAN-MiD is the first publicly available dataset of its kind that it is also extensible allowing anyone to add new attacks and reproduce the generation process.