Leakage-Based Precoder Optimization and Deep Unfolding for User-Centric Cell-Free Massive MIMO

This article investigates the leakage-based precoder design for user-centric cell-free massive MIMO systems. Given the clustered APs for each UE, the problem is formulated as a group sparse optimization to maximize the leakage rate under per-AP power constraints. Its equivalent formulation is derive...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on vehicular technology Vol. 74; no. 5; pp. 8369 - 8373
Main Authors: Xu, Chunmei, Zhang, Cheng, Huang, Yongming, Wu, Yi, Lu, Zhaohua
Format: Journal Article
Language:English
Published: New York IEEE 01.05.2025
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
Cell-free
Clustering algorithms
deep unfolding
Interference
Iterative methods
Leakage
Massive MIMO
Measurement
Neural networks
Optimization
Optimization methods
precoder
Receivers
Resource management
Signal to noise ratio
Simulation
SLNR
Vectors
ISSN:0018-9545, 1939-9359
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This article investigates the leakage-based precoder design for user-centric cell-free massive MIMO systems. Given the clustered APs for each UE, the problem is formulated as a group sparse optimization to maximize the leakage rate under per-AP power constraints. Its equivalent formulation is derived by introducing two auxiliary variables. An iterative optimization method (IOM) is proposed based on block coordinate descent and bisection search techniques, which converges to a stationary solution. To reduce the required computations, a deep unfolding-based method (DUBM) is developed to unfold the iterative procedure of IOM via the proposed neural network layer. Simulation results demonstrate the competitive performance of the proposed methods, with IOM achieving higher sum-rates while DUBM providing higher computational efficiency compared to the reference methods.
Bibliography:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:0018-9545
1939-9359
DOI:10.1109/TVT.2024.3524182