User-Centric Cooperative MEC Service Offloading

TL;DR

This paper presents a distributed, cooperative service caching strategy for user-centric mobile edge computing that minimizes delay and caching costs in wireless networks with dynamic service demands.

Contribution

It introduces a novel online distributed caching algorithm based on ADMM for long-term delay minimization in MEC networks.

Findings

Achieves optimal average completion delay in simulations.

Reduces caching costs while maintaining low latency.

Validates effectiveness of the distributed algorithm.

Abstract

Mobile edge computing provides users with a cloud environment close to the edge of the wireless network, supporting the computing intensive applications that have low latency requirements. The combination of offloading with the wireless communication brings new challenges. This paper investigates the service caching problem during the long-term service offloading in the user-centric wireless network. To meet the time-varying service demands of a typical user, a cooperative service caching strategy in the unit of the base station (BS) cluster is proposed. We formulate the caching problem as a time-averaged completion delay minimization problem and transform it into time-decoupled instantaneous problems with a virtual caching cost queue at first. Then we propose a distributed algorithm which is based on the consensus-sharing alternating direction method of multipliers to solve each instantaneous problem. The simulations validate that the proposed online distributed service caching algorithm can achieve the optimal time-averaged completion delay of offloading tasks with the smallest caching cost in the unit of a BS cluster.