Joint Task Offloading and Resource Allocation for Multi-Task Multi-Server NOMA-MEC Networks

Jianbin Xue,Yaning An

doi:10.1109/access.2021.3049883

Abstract

By offloading computationally intensive tasks of smart end devices to edge servers deployed at the edge of the network, mobile edge computing (MEC) has become a promising technology to provide computing services for Internet of Things (IoT) devices. In order to further improve the access capability of MEC and increase the spectrum utilization efficiency, in this article, Non-Orthogonal Multiple Access (NOMA) technology is introduced into MEC systems and we study the computing offloading problem of multi-user, multi-task and multi-server through joint optimization of task offloading and resource allocation, we intend to maximize the system’s processing capability as an optimization goal. To solve the proposed mixed integer nonlinear programming (MINLP) problem, the objective optimization problem is firstly decoupled into two sub-problems of resource allocation and task allocation. Secondly the resource allocation problem is further decomposed into computation resource optimization and communication resource allocation. For the communication resource allocation, it first fixed power allocation, then the sub-channel allocation problem is regarded as a many-to-one matching problem between sub-channels and users. In addition, we propose a low-complexity sub-optimal matching algorithm for sub-channel allocation to maximize the offloading efficiency. Based on our proposed sub-channel allocation scheme, the transmission power allocation is regarded as a convex optimization problem, which is tackled by Lagrangian multiplier method. Finally, under the condition of resource allocation, the tasks of all end devices (EDs) are allocated. Experimental numerical results show that the proposed scheme can effectively decrease latency and energy consumption of networks, improve system processing capability, and further improve MEC system performance.

Highlights

The process of social industrialization puts forward highquality requirements for fast and effective data services and the application of 5G network provides a basic platform for this demand
The cloud center is rich in computing and storage resources, the main problem is resource centralization and the distance between mobile terminals (MTs) and the cloud is longer, which will lead to the large network delay, high energy consumption and task execution overhead, while sensitive applications require low delay and small energy consumption [4]
To confront such a real-time challenge, as a distributed computing paradigm, mobile edge computing (MEC) was proposed to solve the above problems by bringing computation and storage close to edge network [5]

Summary

Introduction

The process of social industrialization puts forward highquality requirements for fast and effective data services and the application of 5G network provides a basic platform for this demand. The explosive growth of mobile Internet services has generated various emerging mobile applications with huge amount of computation, such as virtual reality (VR), human-computer interaction and big data analysis, which often demand stringent delay and processing requirements. This will bring challenges to MTs with limited battery capacity and computing resources [1]-[3]. In order to further improve the efficiency and flexibility of computing offloading by MEC, a new multi-

Objectives

Results

Conclusion