Joint task offloading and resource allocation for multi-user collaborative mobile edge computing

The exponential growth of Internet of Things (IoT) terminal devices in recent years has caused overload when offloading their computational tasks solely to edge servers. Partial offloading and collaborative mobile edge computing (MEC) are thus proposed to segment the tasks and offload a portion of t...

Full description

Saved in:
Bibliographic Details
Published in:Computer networks (Amsterdam, Netherlands : 1999) Vol. 250; p. 110604
Main Authors: An, Xiaobei, Li, Yanjun, Chen, Yuzhe, Li, Tingting
Format: Journal Article
Language:English
Published: Elsevier B.V 01.08.2024
Subjects:
Ant colony optimization
Collaborative mobile edge computing
Deep deterministic policy gradient
Non-orthogonal multiple access
Offloading decision
Partial offload
Resource allocation
Ant colony optimization
Offloading decision
Resource allocation
Deep deterministic policy gradient
Collaborative mobile edge computing
Partial offload
Non-orthogonal multiple access
ISSN:1389-1286, 1872-7069
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The exponential growth of Internet of Things (IoT) terminal devices in recent years has caused overload when offloading their computational tasks solely to edge servers. Partial offloading and collaborative mobile edge computing (MEC) are thus proposed to segment the tasks and offload a portion of the tasks to neighboring devices via device-to-device (D2D) communication, which alleviates the server load and fully explores the idle resources among end-devices. Consider that the end-devices face the limitation in energy provision, this paper aims to minimize the energy consumption of the entire end-devices while ensuring the task completion delay, by optimizing the devices’ offloading decisions, corresponding offloading ratios and transmit powers. This problem is formulated as a mixed-integer nonlinear programming (MINLP) problem. To solve the problem, a two-level alternating optimization framework is proposed. At the upper level, ant colony optimization (ACO) is used to obtain a population of offloading decisions, while deep deterministic policy gradient (DDPG) algorithm is employed at the lower level to obtain the optimal offloading decision, corresponding offloading ratios and transmit powers. Simulation results demonstrate that our proposed ACO-DDPG algorithm converges well. Further more, it consumes the lowest energy compared with all the baseline algorithms and can achieve 100% task completion ratio under various simulation settings.
ISSN:1389-1286
1872-7069
DOI:10.1016/j.comnet.2024.110604