A QoS-Aware DRL-Based Service Switching Approach for Vehicular Edge Computing

To meet the requirements of low latency and high reliability of vehicular computing, vehicular services need to be offloaded to multi-access edge computing (MEC) servers. However, due to the mobility of connected vehicles and the limited coverage of roadside units (RSUs), vehicular services need to be adaptively switched. In order to ensure the quality of service (QoS) in the process of vehicular service switching, a service switching approach based on deep reinforcement learning is proposed in this paper. Firstly, the optimization problem model is established by considering the QoS, service migration cost, and latency. Secondly, the problem is decomposed into several subproblems and transformed into Markov decision processes (MDPs). Finally, a deep reinforcement learning-based algorithm is employed to find the optimal service migration strategy. Simulation results indicate that the proposed approach can effectively ensure the QoS compared to the baseline approaches.