Trajectory and Phase Shift Design for Maritime UAV-ARIS System Based on Deep Reinforcement Learning

In maritime wireless communication networks, the link between unmanned aerial vehicle (UAV) temporary base stations and buoy sensors may be obstructed by passing ships and small islands et al, which lead to reduce or interrupt data transmission rates. To address this problem, this paper introduces an aerial reconfigurable intelligent surface (ARIS) to assist maritime UA V communication systems. In our study, we are considering the use of deep reinforcement learning algorithm DDQN to simultaneously optimize the 3D trajectory of the UA V and the passive reflection phase shift of ARIS to enhance the UA V ‘sdata transmission rate and reduce its overall Energy Consumption (EC), ultimately maximizing the UA V's Energy Efficiency (EE). Extensive simulation results demonstrate that with the assistance of DDQN algorithms, the EE of the maritime UA V communication system with optimal ARIS phase shift assistance surpasses that of systems with random ARIS phase shift assistance or without ARIS assistance.