Joint Resource Allocation and Trajectory Optimization Algorithm in UAV-Assisted MEC System Based on DDPG

Due to the advantages of maneuverability and the small size of Unmanned Aerial Vehicles (UAVs), mobile edge computing (MEC) servers can be deployed for them to provide computing services. Because the computing resources and energy are limited and the tasks have strict time constraints, tasks are prioritized and their age of information (AoI) is recorded. To solve the issues, we propose a deep deterministic policy gradient (DDPG)-based joint resource allocation and trajectory optimization algorithm in UAV-assisted MEC system. We formulate the resource allocation and trajectory problems as a Markov Decision Process (MDP). Then, we propose a DDPG algorithm to solve the formulated problem and minimize the total energy consumption of the system. Simulation results show that the proposed DDPG-based joint optimization algorithm significantly reduces the energy consumption of the system and has a fast convergence rate.