Enhanced Reinforcement Learning Based Multi-Node Cooperative Deployment Strategy for UAV Monitoring

The proliferation of unmanned aerial vehicles (UAVs) has brought more uncertainties to urban electromagnetic safety, necessitating enhanced UAV monitoring for effective urban management and security. As one effective solution, the deployment strategy of ground monitoring nodes is studied in this paper to achieve a three-dimensional (3D) wide range of electromagnetic coverage monitoring of UAVs in densely distributed urban areas with buildings. We first predict the path loss between ground monitoring nodes and UAVs through various environmental characteristics based on a machine learning algorithm, which serves as the basis for assessing connectivity to evaluate monitoring performance. Then considering the changes in monitoring performance, the proposed algorithm adapts the transformer framework to jointly optimize the deployment locations of ground monitoring nodes within the constraints of the monitoring area, thereby achieving 3D monitoring coverage of UAVs. Simulation results show that the algorithm achieves better stability and improves coverage performance by 8.7% and 4.08% compared with the genetic algorithm (GA) and the deep deterministic policy gradient (DDPG) algorithm under the same channel model and the same transmitting power of 25 dBm.