Time-aware MDP-based Service Migration in 5G Mobile Edge Computing

With the advent of edge-centric computing paradigms, computation capabilities have been transferred from cloud data centers toward the network edge to support the requirements of highly demanding and real-time services. Hence, in edge computing models, servers located on the network edge can host cloud-based services to reduce network overhead and latency in user services. However, due to user mobility in a dynamic and non-deterministic environment and the limited coverage of the edge servers along with the intrinsic limitations of the Mobile Edge Computing (MEC) paradigm, service migration is inescapable to ensure end-users Quality of Service (QoS). In this study, we formulate the service migration problem as a Markov Decision Process (MDP) to determine when to migrate the service between edge servers by considering the path loss effect in wireless MEC networks. To this end, we propose a value-iteration method to tackle the problem and find the optimal policy for all MEC servers. Experimental results reveal that our proposed scheme outperforms two baseline schemes in terms of the discounted sum cost.