Onboard Audio and Video Processing for Secure Detection, Localization, and Tracking in Counter-UAV Applications

Abstract

Nowadays, UAVs are of fundamental importance in numerous civil applications like search and rescue and military applications like monitoring and patrolling or counter-UAV where the remote UAV nodes collect sensor data. In the last case, flying UAVs collect environmental data to be used to contrast external attacks launched by adversary drones. However, due to the limited computing resources on board of the acquisition UAVs, most of the signal processing is still performed on a ground central unit where the sensor data is sent wirelessly. This poses serious security problems from malicious entities such as cyber attacks that exploit vulnerabilities at the application level. One possibility to reduce the risk is to concentrate part of the computing onboard of the remote nodes. In this context, we propose a framework where detection of nearby drones and their localization and tracking can be performed in real-time on the small computing devices mounted on board of the drones. Background subtraction is applied to the video frames for pre-processing with the objective of an on-board UAV detection using machine-vision algorithms. For the localization and tracking of the detected UAV, multi-channel acoustic signals are instead considered and DOA estimations are obtained through the MUSIC algorithm. In this work, the proposed idea is described in detail along with some experiments and, then, methods of effective implementation are provided.