Real Time Malicious Drone Detection Using Deep Learning on FANETs
Cengizhan Yapıcıoğlu, Mehmet Demirci, M. Ali Akcayol · 2024
Unmanned aerial vehicles, especially drones, are increasingly used for transportation, communication and military purposes. Using only one drone to accomplish a mission leads to a solution that is costly and has low error tolerance. For this reason, a network structure called Flying Ad-Hoc Networks (FANET) organize drones with a task sharing mechanism to lower cost. However, these networks remain vulnerable to various attacks due to vulnerabilities such as the use of civilian drones, use of unencrypted GPS signals, physical attacks using malicious drones, etc. Solutions to these attacks should consider the limited memory and calculation capabilities of drones. In this study, drone detection and subsequent classification of malicious drones, which are potential sources of man-in-the-middle or physical attacks, were implemented based on real images that can be captured by the drone camera. You Only Look Once (YOLO) detection algorithm was used in the drone detection phase and a Convolutional Neural Network model was used in the classification phase. In the study, a dataset consisting of 4 classes (Yuneec Typhon, DJI Tello, DJI Phantom 4 and Other) was created using internet resources and YouTube videos, and the classification success was measured as 88.78 %.