Hybrid Beamforming for mmWave Massive MIMO Systems Using Conditional Generative Adversarial Networks

Massive multiple-input multiple-output (MIMO) systems operating in millimeter wave (mmWave) frequency bands are considered to be one of the key enablers of beyond-fifth-generation cellular systems. Although the highest spectral efficiency in such systems would be achieved using fully-digital precodi...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on vehicular technology Vol. 73; no. 10; pp. 15803 - 15808
Main Authors: Banerjee, Bitan, Elliott, Robert C., Krzymien, Witold A., Medra, Mostafa
Format: Journal Article
Language:English
Published: New York IEEE 01.10.2024
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Antenna arrays
Antennas
Array signal processing
Beamforming
Cellular radio
conditional generative adversarial network (CGAN)
Generative adversarial networks
hybrid beamforming
Massive MIMO
Millimeter wave communication
Millimeter waves
MIMO communication
mmWave
Precoding
Radio frequency
Wasserstein conditional generative adversarial network (WCGAN)
ISSN:0018-9545, 1939-9359
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Massive multiple-input multiple-output (MIMO) systems operating in millimeter wave (mmWave) frequency bands are considered to be one of the key enablers of beyond-fifth-generation cellular systems. Although the highest spectral efficiency in such systems would be achieved using fully-digital precoding, the large number of antennas in massive MIMO systems makes using a radio frequency (RF) chain for each antenna expensive and currently infeasible in practice. A common alternative solution is hybrid beamforming, which combines analog beamforming and digital precoding and reduces the required number of RF chains. The primary goal of hybrid beamforming is to provide precoding performance as close as possible to that of a fully-digital precoder. In this work, we consider two variants of a generative adversarial network (GAN), namely a conditional GAN (CGAN) and Wasserstein CGAN (WCGAN) to develop the hybrid precoder. The CGAN is used to implement the (partially-connected) analog beamformer and the WCGAN is used for the digital precoder. Our simulation results demonstrate the proposed method yields an improvement in spectral efficiency of about 12-19% over some existing hybrid beamforming schemes and achieves up to 87% of the performance of fully-digital precoding.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-9545
1939-9359
DOI:10.1109/TVT.2024.3409192