3D-Trajectory and Phase-Shift Design for RIS-Assisted UAV Systems Using Deep Reinforcement Learning

Unmanned aerial vehicle (UAV) can effectively work as temporary base station or access point in the air to transfer/receive data to/from ground terminals (GTs). However, UAV-GT links might be blocked by ground obstacles, like buildings in urban area, leading to a poor performance on data transferrin...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on vehicular technology Vol. 71; no. 3; pp. 3020 - 3029
Main Authors: Mei, Haibo, Yang, Kun, Liu, Qiang, Wang, Kezhi
Format: Journal Article
Language:English
Published: New York IEEE 01.03.2022
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
3D-trajectory
Algorithms
Approximation
Array signal processing
Autonomous aerial vehicles
Base stations
Decision making
Deep learning
deep reinforcement learning (DRL)
intelligent reflecting surface (IRS)
Machine learning
Optimization
Phase shift
Propulsion
Reconfigurable intelligent surface (RIS)
Resource management
Rotors
System effectiveness
unmanned aerial vehicle (UAV)
Unmanned aerial vehicles
Urban areas
Wireless networks
ISSN:0018-9545, 1939-9359
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Unmanned aerial vehicle (UAV) can effectively work as temporary base station or access point in the air to transfer/receive data to/from ground terminals (GTs). However, UAV-GT links might be blocked by ground obstacles, like buildings in urban area, leading to a poor performance on data transferring rate. To address this problem, reconfigurable intelligent surface (RIS), as a promising technique, can intelligently reflect the received signals between UAV and GT to significantly enhance the communication quality. Under this deployment of RIS-assisted UAV, we intend to jointly optimize the 3D-space of the UAV and the phase-shift of the RIS to maximize the data transferring rate of the UAV, while minimizing the UAV propulsion energy. The joint problem is non-convex in its original form and difficult to be timely solved by using traditional method, like successive convex approximation (SCA). Therefore, to facilitate the online decision making to this joint problem, we leverage deep reinforcement learning (DRL) to learn the near-optimal solution, and the well known Double Deep Q-Network (DDQN) and Deep Deterministic Policy Gradient (DDPG) algorithms are ultilized. Numerical results show that DRL can effectively improve the energy-efficiency performance of the RIS-Assisted UAV system, compared with benchmark solutions.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-9545
1939-9359
DOI:10.1109/TVT.2022.3143839