Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Distributed Graph Neural Network Design for Sum Ergodic Spectral Efficiency Maximization in Cell-Free Massive MIMO

Nguyen Xuan Tung, Trinh Van Chien, Hien Quoc Ngo, Won Joo Hwang

TL;DR

Numerical results show that the proposed distributed learning-based approach achieves a sum ergodic rate close to that of centralized learning while outperforming the model-based optimization.

Abstract

This paper proposes a distributed learning-based framework to tackle the sum ergodic rate maximization problem in cell-free massive multiple-input multiple-output (MIMO) systems by utilizing the graph neural network (GNN). Different from centralized schemes, which gather all the channel state information (CSI) at the central processing unit (CPU) for calculating the resource allocation, the local resource of access points (APs) is exploited in the proposed distributed GNN-based framework to allocate transmit powers. Specifically, APs can use a unique GNN model to allocate their power based on the local CSI. The GNN model is trained at the CPU using the local CSI of one AP, with partially exchanged information from other APs to calculate the loss function to reflect system characteristics, capturing comprehensive network information while avoiding computation burden. Numerical results show that the proposed distributed learning-based approach achieves a sum ergodic rate close to that of centralized learning while outperforming the model-based optimization.

Distributed Graph Neural Network Design for Sum Ergodic Spectral Efficiency Maximization in Cell-Free Massive MIMO

TL;DR

Numerical results show that the proposed distributed learning-based approach achieves a sum ergodic rate close to that of centralized learning while outperforming the model-based optimization.

Abstract

This paper proposes a distributed learning-based framework to tackle the sum ergodic rate maximization problem in cell-free massive multiple-input multiple-output (MIMO) systems by utilizing the graph neural network (GNN). Different from centralized schemes, which gather all the channel state information (CSI) at the central processing unit (CPU) for calculating the resource allocation, the local resource of access points (APs) is exploited in the proposed distributed GNN-based framework to allocate transmit powers. Specifically, APs can use a unique GNN model to allocate their power based on the local CSI. The GNN model is trained at the CPU using the local CSI of one AP, with partially exchanged information from other APs to calculate the loss function to reflect system characteristics, capturing comprehensive network information while avoiding computation burden. Numerical results show that the proposed distributed learning-based approach achieves a sum ergodic rate close to that of centralized learning while outperforming the model-based optimization.

Paper Structure

This paper contains 11 sections, 9 equations, 4 figures, 4 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Downlink Cell-free Massive MIMO
  3. System and Signal Models
  4. Problem Formulation
  5. Distributed GNN-Based Learning Algorithm
  6. Cell-Free Massive MIMO as Multiple Graphs
  7. GNN Designed based on the Message Passing Method
  8. Distributed GNN-based Learning
  9. Data Exchange Analysis
  10. Performance Evaluation
  11. Conclusion

Figures (4)

  • Figure 1: The centralized versus distributed fashion.
  • Figure 2: The cell-free massive MIMO as multiple graphs.
  • Figure 3: The general flow chart of the training phase.
  • Figure 4: Total ergodic rate versus the number of APs when the system serves $N=5$ or $N=15$ UEs with $4$ antennas per AP.

Theorems & Definitions (2)

  • Remark 1
  • Remark 2