SMCoEdge: Simultaneous Multi-server Offloading for Collaborative Mobile Edge Computing

Changfu Xu, Yupeng Li, Xiaowen Chu, Haodong Zou, Weijia Jia, Tian Wang*

*Corresponding author for this work

Research output: Chapter in book/report/conference proceedingConference proceedingpeer-review

Abstract

Collaborative Mobile Edge Computing (MEC) has emerged as a promising solution for low service delay in computation-intensive Internet of Things (IoT) applications. However, current approaches typically perform offline task partitioning and offload each subtask to an Edge Server (ES) for processing. This leads to varying delays in subtask processing across different ESs, resulting in a high make-span of task offloading. To address this issue, we propose a novel approach called SMCoEdge, which utilizes simultaneous multi-ES offloading to minimize the make-span of task offloading for computation-intensive IoT applications. Specifically, we formulate our problem as a mixed integer non-linear programming problem and prove its NP-hardness. We then decompose our problem into two sub-problems of multi-ES selection and task allocation, and propose a Deep Reinforcement Learning-based Simultaneous Multi-ES Offloading (DRL-SMO) algorithm to effectively solve it. Additionally, we analyze the computation complexity of DRL-SMO. Our extensive simulation results demonstrate that SMCoEdge outperforms state-of-the-art approaches by reducing make-span by 18.93% while maintaining a low offloading failure rate.

Original languageEnglish
Title of host publicationAlgorithms and Architectures for Parallel Processing
Subtitle of host publication23rd International Conference, ICA3PP 2023, Tianjin, China, October 20–22, 2023, Proceedings, Part V
EditorsZahir Tari, Keqiu Li, Hongyi Wu
PublisherSpringer Singapore
Pages73-91
Number of pages19
ISBN (Electronic)9789819708086
ISBN (Print)9789819708079
DOIs
Publication statusPublished - 27 Feb 2024
Event23rd International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2023 - Tianjin, China
Duration: 20 Oct 202322 Oct 2023

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume14491 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference23rd International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2023
Country/TerritoryChina
CityTianjin
Period20/10/2322/10/23

Scopus Subject Areas

  • Theoretical Computer Science
  • Computer Science(all)

User-Defined Keywords

  • Edge-edge collaboration
  • Mobile edge computing
  • Simultaneous multi-server offloading
  • Task allocation

Cite this