Cost-effective task offloading and trajectory optimization in UAV assisted edge networks with DDPG

• Date: 12 September 2024
• Pages: 494-519
• DOI: https://doi.org/10.1108/IJWIS-05-2024-0132
• Published date: 12 September 2024
• Subject Matter: Information & knowledge management,Information & communications technology,Information systems,Library & information science,Information behaviour & retrieval,Metadata,Internet
• Author: Jiaqing Shen,Xu Bai,Xiaoguang Tu,Jianhua Liu

Cost-effective task off‌loading and

trajectory optimization in UAV

assisted edge networks with DDPG

Jiaqing Shen,Xu Bai,Xiaoguang Tu and Jianhua Liu

Institute of Electronic and Electrical Engineering,

Civil Aviation Flight University of China, Guanghan, China

Abstract

Purpose –Unmanned aerial vehicles (UAVs), known for their exceptional f‌lexibility and maneuverability,

have become an integral part of mobile edge computing systems in edge networks. This paper aims to

minimize system costs withina communication cycle. To this end, this paper has developeda model for task

off‌loading in UAV-assisted edge networks under dynamic channelconditions. This study seeks to eff‌iciently

execute task off‌loading while satisfying UAV energy constraints, and validates the effectiveness of the

proposedmethod through performance comparisons with other similar algorithms.

Design/methodology/approach –To address this issue, this paper proposes a task off‌loadingand trajectory

optimization algorithm using deep deterministic policy gradient,which jointly optimizes Internet of Things (IoT)

device scheduling, power distribution, task off‌loadingand UAV f‌light trajectory to minimizesystem costs.

Findings –The analysis of simulation results indicates that this algorithm achieves lower redundancy

compared to others, alongwith reductions in task size by 22.8%, f‌light time by 34.5%, number of IoT devices

by 11.8%,UAV computingpower by 25.35% and the required cycle for per-bit tasks by 33.6%.

Originality/value –A multi-objective optimization problem is established under dynamic channel

conditions,and the effectiveness of this approach is validated.

Keywords Mobile edge computing (MEC), UAV, DDPG, Task off‌loading, Flight trajectory

Paper type Research paper

1. Introduction

Given the swift progression of the Internetof Things (IoT) in recent years, many devices and

sensors have been connected to the internet, leading to the formation of large and

decentralized IoT systems. These devices generate vast amounts of raw data, necessitating

real-time and eff‌icient computation to provide intelligent services and decision support.

Currently, although IoT devices are equipped with processors, their computational

capabilities are relativelylow, resulting in high latency and energy consumptionissues while

meeting network performance requirements. To tackle this challenge, mobile edge

computing (MEC) has arisen as an innovative technology grounded on distributed

computing architecture (Liu et al., 2023). MEC uses MEC methods for processing tasks on

IoT devices, enabling computationaltasks of IoT devices to be off‌loaded to the network edge

for processing. Thisgreatly reduces latency and energy consumption(Hu et al., 2022b,2021,

Funding: This article is supported in part by the China Postdoctoral Science Foundation under project

(2022M722248), the Central University Basic Research Funds (J2023-027), and the Open Fund of the

Key Laboratory of Flight Technology and Flight Safety of the Civil Aviation Administration of China

(FZ2022KF06).

IJWIS

20,5

494

Received 9 May 2024

Revised20 July 2024

19August 2024

Accepted20 August 2024

InternationalJournal of Web

InformationSystems

Vol.20 No. 5, 2024

pp. 494-519

© Emerald Publishing Limited

1744-0084

DOI 10.1108/IJWIS-05-2024-0132

The current issue and full text archive of this journal is available on Emerald Insight at:

https://www.emerald.com/insight/1744-0084.htm

2022a;Liu et al.,2023). However, traditional edge computing servers are typicallydeployed

in f‌ixed locations, whichcannot meet the needs of wireless network scenarios with limitedor

no available infrastructure, such as disaster response, emergency rescue and military

exercises. To address these challenges, unmanned aerial vehicle (UAV)-assisted MEC has

been proposed as a potential technology due to its adaptability and rapid deployment

capabilities. Compared to infrastructure-based MEC, UAVs establish line of sight (LoS)

links with ground users,enabling edge computing services to be provided at any locationand

time (Wang et al., 2021b), thus overcoming the limitations of traditional edge computing

servers in remote areasor during emergency rescue operations.

Despite the advantages, UAV-assisted edge computing faces several challenges in practical

applications. First, environmental obstacles such as trees and buildings can cause signal

attenuation, which affects the data transmission rate of communication links. Second,

maintaining high-quality communication links requires real-time and precise adjustment of the

transmit power of IoT devices. Finally, optimizing the eff‌icientuseoflimitedenergybyUAVs

during task computations to ensure long-term stable operation is also a critical issue that needs

addressing. Based on the above analysis, this paper aims to address the following core issues:

Q1. How to dynamically select suitable communication links to address signal

attenuation caused by environmentalfactors?

Q2. How to precisely and in real-time adjust the transmit power of IoT devices to

maintain the quality of communicationlinks?

Q3. How to eff‌iciently use the limited energy of UAVs to optimize their performance in

task computations?

To address the aforementioned issues, this paper constructs a model for a single UAV-

assisted edge computing system, consisting of a UAV and IoT devices. Building on this

model, we propose a task off‌loading and trajectory optimization algorithm based on deep

deterministic policy gradient(DDPG). This algorithm aims to minimize the total systemcost

by jointly optimizing the schedulingof IoT devices, UAV f‌light trajectories,power allocation

and task off‌loading. The algorithmspecif‌ically focuses on dynamically selecting the optimal

communication links and real-time decision-making on task off‌loading ratios, as well as

timely adjusting the transmission power of IoT devices and the f‌light paths of UAVs. The

optimization strategy not only enhances the data transmission rate but also improves the

system’s robustnessand energy eff‌iciency, enabling UAV-assisted MEC systems to maintain

high eff‌iciency even undervarying environmental conditions.

Therefore, the maincontributions of this paper are as follows:

•Based on the communication and computing capabilities of the edge nodes,

considering the presence of obstacles affecting the transmission rate, a multi-

objective optimization problem of task off‌loading of UAV-assisted edge computing

system under dynamic channel is established. The aim is to minimize the cost of the

system under the condition of UAV energy constraints by optimizing the f‌light

trajectory of the UAV.

•We construct a task off‌loading and trajectory optimization algorithm for UAV-

assisted edge computing using DDPG. To facilitate more effective training of the

deep neural network (DNN), various state variables are normalized in this study.

Dividing each input state by the difference between the maximum and minimum

values of the variable serves as a scaling factor, which is subsequently used to

address discrepancies in input variable magnitudes.

International

Journal of Web

Information

Systems

495