Multi-Agent Reinforcement Learning Based Trajectory Design for UAV-Assisted Emergency Wireless Communication

Communication gaps in specific areas, caused by natural disasters and other factors, create an urgent need for emergency communication solutions. Unmanned aerial vehicles (UAVs), with their low cost and easy deployment, offer a promising solution for such needs by providing flexible wireless coverage. However, due to the limited coverage range and energy capacity of UAVs, an effective trajectory design method is crucial to maximize the user coverage ratio. In this paper, we formulate the trajectory design problem as a constrained Markov decision process (MDP) problem. A multi-agent reinforcement learning (MARL)-based algorithm, focused on maximizing the user coverage ratio while accounting for energy consumption, is proposed to solve this problem. In particular, we integrate turning control factors into the immediate reward and penalty constraints to mitigate frequent turning, a factor often neglected in existing algorithms. Simulation results demonstrate that the proposed algorithm not only improves user coverage but also enhances communication efficiency.