Priority-Aware Task Offloading in UAV-Assisted Vehicular Network Via Deep Reinforcement Learning

Vehicular Edge Computing (VEC) has been regarded as a promising mechanism that can increase the computational capacity of vehicles to support the execution of tasks with various priorities. However, the ground base stations can be destroyed when disaster occur, and vehicle users may not compute tasks timely. How to design an efficient task offloading method is a main challenge. In this paper, we propose a VEC scheme with an Unmanned Aerial Vehicles (UAV) as a MEC server providing computing services. We present the Deep reinforcement leArning based offloading deCision and rEsourcement Management (DACEM) algorithm, where leverages deep deterministic policy gradient (DDPG) method to optimize the computation offloading and resource allocation for minimizing the total energy consumption of all tasks with different priorities, subject to the maximum tolerated latency constraints. Simulation results demonstrate that the DACEM outperforms conventional methods.