Mobile edge computing (MEC) is an emerging technology that provides cloud services at the edge of network to enable latency-critical and resource-intensive E-healthcare applications. User mobility is common in MEC. User mobility can result in an interruption of ongoing edge services and a dramatic drop in quality of service. Service migration has a great potential to address the issues and brings inevitable cost for the system. In this paper, we propose a service migration solution based on migration zone and formulate service migration cost with a comprehensive model that captures the key challenges. Then, we formulate service migration problem into Markov decision process to obtain optimal service migration policies that decide where to migrate in a limited area. We propose three algorithms to resolve the optimization problem given by the formulated model. Finally, we demonstrate the performance of our proposed algorithms by carrying out extensive experiments. We show that the proposed service migration approach reduces the total cost by up to 3 times compared to no migration and outperforms the general solution in terms of the total expected reward.
Copyright © 2021 Xuhui Zhao et al.