Coordinated UAV Trajectory Planning and Resource Allocation for Target Pursuit and Communication

In this paper, we develop a learning-based framework for an unmanned aerial vehicle (UAV) to sense, track and pursue mobile users as well as communicate with them. UAV is assumed to be equipped with a radar unit, and radar measurements are incorporated into the extended Kalman filter (EKF) algorithm to track the position of a mobile user. An integrated sensing and communication (ISAC) approach is considered, and resource management for the allocation of radar dwell time for target tracking and the communication time for data transfer is addressed. Based on the estimated location, the UAV's task is to select actions to approach and pursue the target users and also to perform time management between radar sensing and data communication. A bootstrapped deep Q-network (DQN) based reinforcement learning model is utilized to generate a set of actions for directing UAV movement and allocating time within the ISAC system. The eventual goal for the UAV is to optimize the data communication by tracking and pursuing mobile users. The numerical results demonstrate that with the designed approach, UAV is able to approach the target user and communicate efficiently. When applied to a multi-target scenario, the proposed approach enables the UAV to pursue all targets and perform data communication effectively.