Online Trajectory Optimization for UAV-Assisted Hybrid FSO/RF Network With QoS-Guarantee

This letter investigates a hybrid FSO/RF wireless network supported by the unnamed aerial vehicle (UAV), providing coverage for mobile vehicles. Albeit mobility and signal-propagation impede quality-of-service (QoS), existing research, however, ignores the impact of practical mobility that reshapes the behaviour pattern UAVs would learn to conduct. This letter proposes a deep reinforcement learning (DRL) algorithm with proximal policy optimization (PPO) to guarantee the UAV-supported QoS through trajectory optimization online. Under various setups of speed limitation of vehicles and QoS requirements, our numerical results demonstrate the effectiveness and robustness of the herein proposed algorithm.