Joint optimization task offloading and trajectory control for unmanned-aerial-vehicle-assisted mobile edge computing

The appearance of Mobile Edge Computing (MEC) and Unmanned Aerial Vehicle (UAV) is significant for the future progress of the Internet of Things (IoT). Since the system model with a continuous action space and high-dimensional state space, the joint optimization of UAV trajectory and the computational offloading problem is non-convex, and traditional algorithms for instance ant colony algorithm, genetic algorithm, Actor Critic (AC) algorithm, and Deep Deterministic Policy Gradient (DDPG) algorithm are difficult to cope with. Reasonably formulating the computational task offloading strategy and the trajectory control of the UAV is crucial for the high-efficiency completion of the task. In this paper, a computational offloading and trajectory control system model for UAV-assisted MEC is proposed. We seek to maximize the user ratio of coverage by jointly optimizing computing offload scheduling and UAV trajectories. We propose an improved DDPG algorithm to optimize the objective function and achieve the optimal solution. Meanwhile, our algorithm can achieve an improvement in the user rate of coverage while avoiding obstacles as compared with baseline algorithms, AC, and DDPG.