Joint Optimization on Computation Offloading and Resource Allocation in Mobile Edge Computing

Kaiyuan Zhang,Xiaolin Gui,Dewang Ren

DOI: https://doi.org/10.1109/WCNC.2019.8886130

2019-01-01

Abstract:We consider a general multi-user mobile edge computing (MEC) system with multiple MEC servers. For each user, a MEC serves both as the network access point and a computation service provider, where users can offload part of their tasks. We formulate the sum cost of time delay and energy consumption for all mobile users as our optimization objective. This problem is NP-hard in general. In this paper, we jointly optimize the offloading decisions of all users tasks as well as the allocation of computation and communication resources, pursing the minimal sum cost for all users. We proposed an efficient two-stage algorithm comprising of one-dimensional search (ODS) and alternating optimization (AO). The first stage is responsible for obtaining the optimal offloading decision, and the second stage is in charge of computing a locally optimal solution for resource allocation. It is shown to give nearly optimal performance under a wide range of parameter settings. Through evaluating the performance of different combinations of the two stages of ODS-AO algorithm, we provide insights into their roles and contributions in the overall solution. Our simulation results show that the proposed scheme achieves significant reduction on the average delay and sum cost compared to other baselines.