Reinforcement Learning-Based UAVs Resource Allocation for Radar-Communication Integrated System

Due to the limited mission accomplished by a single unmanned aerial vehicle (UAV), radar-communication integrated system for group UAVs has attracted more attention in military and civilian fields. In this paper, the resource allocation problem is addressed for the group UAVs using the radar-communication integrated system. Large number of UAVs leads to the complexity of the resource allocation problem, so we propose a reinforcement learning-based methods. A novel reward combining both the mutual information (MI) and the communication rate (CR) is defined, where the MI is used to describe the detection performance and the CR is for that of the wireless communication. The implement of reinforcement learning for radar-communication integrated system for group UAVs includes deep Q network (DQN) and dueling-DQN algorithms. Simulation results show that the resource utilization in the radar-communication integrated system is improved effectively for the group UAVs.