A Multi-Agent Service Migration Algorithm for Mobile Edge Computing with Diversified Services

In mobile edge computing (MEC), servers are deployed at the location near to users to provide a better service for mobile devices (MDs). Compared to cloud computing, it’s more common for MDs to move between different servers in MEC. As a result, a good service migration strategy should proactively migrate services on current servers to a closer server to minimize service delays or interruptions. While in previous works, users’ individual needs for service quality based on their conditions are ignored. Besides, distributed migration algorithms are rarely studied. To this end, we build a MEC system with different quality level of services. The MD can notify the server to dynamically adjust the quality of the remaining services when a user enters the coverage area of a new server. We formulate the service migration process as a service profit maximization problem where the user’s diverse requirements are considered as a quality of experience (QoE) factor. Finally, a Multi-Agent deep deterministic policy gradient based Service Migration algorithm (MASM) is proposed to acquire collaborative migration decisions. Simulation results show that our method significantly reduces the service delay and improves the service quality score, achieving the best QoE among six migration schemes.