A Dynamic Service Migration Method in Mobile Edge Computing Based on Reinforcement Learning

With the popularization and rapid development of the Internet of Vehicles and intelligent networked vehicles, the number of mobile edge computing (MEC) devices and data requests are growing rapidly. The quality of service (QoS) of intelligent connected vehicles (ICVs) under the traditional data processing mode is greatly restricted. This paper studied the dynamic service migration method of MEC node based on reinforcement learning. First, the edge migration model of urban 2-D intersection is established, and the model is described quantitatively; Secondly, aiming at the optimization indexes such as the complexity of edge migration and delay sensitivity in the network, a reinforcement learning algorithm integrating attention mechanism is proposed to optimize the dynamic migration process of edge services; Finally, a series of simulation experiments verify the performance of the proposed algorithm.