Service Deployment for Latency Sensitive Applications in Mobile Edge Computing

Mobile edge computing (MEC) emerges as a promising computing paradigm to meet the increasing demand for compute-intensive latency-sensitive applications. By offloading tasks to the edge MEC can achieve many benefits such as ultra-low latency and saving backbone bandwidth, etc. Many studies have concentrated on the high efficient task offloading. However, the existing studies assume that tasks can be executed on an arbitrary edge server, but in reality different application's tasks require varied services to provide the execution environment and an edge server can host a limited number of services. How to determine the set of services deployed on each server in a MEC system is challenging. To address this challenge, in this paper, we focus on the problem of service deployment, and formulate the problem as a multi-slot latency minimization by leveraging the Lyapunov optimization. An efficient online algorithm is designed to minimize the average latency under a tight budget, and the supremum on algorithm performance is also presented. Experimental results indicate that our algorithm can reduce average latency for end users, and guarantee a low expenditure for the edge service provider.