Deep Q-Network Based Dynamic Trajectory Design for UAV-Aided Emergency Communications

Wang, Liang, Wang, Kezhi, Cunhua Pan, Cunhua Pan, Chen, Xiaomin and Aslam, Nauman (2020) Deep Q-Network Based Dynamic Trajectory Design for UAV-Aided Emergency Communications. Journal of Communications and Information Networks. ISSN 2096-1081 (In Press)

[img]
Preview
Text
JCIN18.pdf - Accepted Version

Download (766kB) | Preview

Abstract

In this paper, an unmanned aerial vehicle (UAV)-aided wireless emergence communication system is studied, where an UAV is deployed to support ground user equipments (UEs) for emergence communications. We aim to maximize the number of the UEs served, the fairness, and the overall uplink data rate via optimizing the trajectory of UAV and the transmission power of UEs. We propose a Deep Q-Network (DQN) based algorithm, which involves the well-known Deep Neural Network (DNN) and Q-Learning, to solve the UAV trajectory problem. Then, based on the optimized UAV trajectory, we further propose a successive convex approximation (SCA) based algorithm to tackle the power control problem for each UE. Numerical simulations demonstrate that the proposed DQN based algorithm can achieve considerable performance gain over the existing benchmark algorithms in terms of fairness, the number of UEs served and overall uplink data rate via optimizing UAV’s trajectory and power optimization.

Item Type: Article
Uncontrolled Keywords: Deep Reinforcement Learning, Deep Q-Network (DQN), Successive Convex Approximation (SCA), UAV, Power Control
Subjects: G400 Computer Science
H600 Electronic and Electrical Engineering
Department: Faculties > Engineering and Environment > Computer and Information Sciences
Depositing User: John Coen
Date Deposited: 01 Dec 2020 14:54
Last Modified: 01 Dec 2020 15:00
URI: http://nrl.northumbria.ac.uk/id/eprint/44884

Actions (login required)

View Item View Item

Downloads

Downloads per month over past year

View more statistics