Long-Term Optimization-Based Data Scheduling and Trajectory Planning for UAV-Assisted Systems

In recent years, unmanned aerial vehicles (UAVs) have been experienced rapid development and have been widely used in many fields. UAVs can serve as aerial base stations to alleviate base station load in cellular systems. In this paper, we study data scheduling and UAV trajectory planing problem in UAV-assisted systems. Stressing the service sensitiveness on queuing delay, we examine the user queue length and formulate joint data scheduling and UAV trajectory planning problem as a longest user queue length minimization problem. To tackle this problem, we first model the problem as a Markov decision process (MDP), and then we propose two deep reinforcement learning (DRL)-based algorithms to determine the joint data scheduling and UAV trajectory planing strategy. Simulation results demonstrate the effectiveness and superiority of the proposed algorithms.