| Abstrakt: |
In response to the increasing issues associated with unmanned aerial vehicles (UAVs), this paper proposes a novel anti-drone system designed to efficiently track and neutralize illegal drones. Unlike previous studies focusing on predicting the location and flight paths of illegal UAVs, this system aims to quickly approach these drones, thereby minimizing potential side effects during the neutralization process. The system utilizes multiple small tracking drones equipped with receiving antennas to intercept communication signals emitted by illegal UAVs. Using these signal data, a deep reinforcement learning network continuously predicts the location of the illegal drones and controls the positions and movements of the tracking drones to facilitate close approach. The proposed deep reinforcement learning model was trained considering various channel conditions, and performance evaluations show that our method reduces the number of movements by 5.41% to 6.92% and the travel distance by 11.5% to 15.8%, compared to existing methods. [ABSTRACT FROM AUTHOR] |
