Multidimensional Swarm Flight Approach for Chasing Unauthorized UAVs Leveraging Asynchronous Deep Learning

Abstract

This article introduces a novel uncrewed aerial vehicles (UAV) chasing system designed to track and chase unauthorized UAVs, significantly enhancing their neutralization effectiveness. The system utilizes a multidimensional swarm flight strategy, employing deep reinforcement learning (DRL) to dynamically adapt the tracking unit's movements based on the received signal strength indicators emitted by unauthorized UAVs. Asynchronous learning techniques involving multiple agents are implemented to expedite the system's learning process. A key feature of our approach is the coordinated use of a swarm of UAVs, which circumvents the considerable size burden associated with mounting multiple antennas on a single UAV. We further refine the asynchronous DRL framework by integrating advanced channel modeling techniques, such as spatial correlation and Doppler shift, to augment the robustness and adaptability of the system. Performance evaluations confirm the system's efficacy under varying channel conditions and operational scenarios. Key contributions include the integration of tracking and chasing functionalities into a unified system, the employment of realistic channel models to enhance system adaptability, and a comprehensive analysis of the relationship between channel sampling frequency and chasing performance. This research advances the field of UAV regulation and control, offering an effective solution to the escalating security challenges posed by unauthorized UAVs.