Quality of Service Optimization in UAV-Assisted Edge Computing Based on Deep Reinforcement Learning

‍ ‍Aiming at the Quality of Service （QoS） problem when UAV （Unmanned Aerial Vehicle， UAV） is equipped with a mobile edge server to serve ground users， an optimization scheme based on deep reinforcement learning is proposed to optimize the UAV flight trajectory and offloading scheme to maximize the QoS when UAV serves users. First， the task delay is defined to characterize the freshness of tasks， and a new type of QoS evaluation index is proposed based on the task delay. Second， the problem of maximizing QoS is modeled as a Markov decision process without transition probability， and defines the state space， action space and reward function of the process. Finally， UAV trains through the proposed algorithm and optimizes the task offloading scheme and finds the optimal flight trajectory to serve the ground users to improve QoS. The simulation results show that the proposed algorithm can effectively improve the QoS and the freshness of tasks in the process of UAV serving ground users compared with other algorithms.