Fight Drones with Drones: Detecting Aerial Perimeter Intrusion using Drone-mounted Microphones

While in recent years there has been a growing interest in de- tecting drones via their characteristic sound signature, especially through ground-based sensory capture, the usage of drones to detect the presence of other drones has shown little support and, even more so, the usage of drones to detect perimeter violation by other drones. In this paper, we are the first to propose, to the best of our knowledge, a framework for deploying a drone swarm as an adaptive acoustic detection barrier. One application of this solution is to detect opponents’ drone-perimeter violation of critical infrastructure such as airports, energy plants, and national borders, to cite a few. Mounting microphones on patrolling de- fender drones turns a static detection footprint into an adaptive coverage layer. In detail, starting from an acoustic model calibrated on real DJI flight data for the Mavic Air 2 and Mini Pro 3, we show that monitoring a 500 × 200 m boundary with a static grid without any gap requires over 400 surveillance drones. Whereas, leveraging mobility and adopting an optimal patrol speed, we reduce the number of needed drones by more than 90%, while still having a 99% detection probability to detect the intruder. We detail the parameters that influence the detection and show that, among others, a 6dB reduction in intruder acoustic signature doubles the required fleet, while any improvement in ego-noise cancellation yields defense a disproportionate advantage. Finally, we carry out a complete analysis covering energy budgets, battery-swap logistics, and collision-safe spacing, providing sizing tables directly applicable to real-life deployments such as the protection of oil refineries, airports, or harbors.