Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

A General Framework for Scalable UE-AP Association in User-Centric Cell-Free Massive MIMO based on Recurrent Neural Networks

Giovanni Di Gennaro, Amedeo Buonanno, Gianmarco Romano, Stefano Buzzi, Francesco A. N. Palmieri

TL;DR

The paper tackles AP-UE association in CF-mMIMO, a problem made intractable by scaling to large numbers of UEs and APs and by pilot contamination. It proposes a BiLSTM-based, master-centric deep learning framework with a probabilistic training scheme (Bernoulli sampling) to dynamically form UE-centric AP clusters, yielding scalability with respect to both UEs and APs. It introduces multi-objective loss functions (SUM, BALANCE, MIN) and a pilot-contamination robust variant that augments inputs with pilot indicators, and demonstrates superior performance over heuristic baselines through extensive simulations. The approach enables distributed, parallelizable learning and can adapt to different network goals, making it well suited for large-scale, next-generation CF-mMIMO deployments.

Abstract

This study addresses the challenge of access point (AP) and user equipment (UE) association in cell-free massive MIMO networks. It introduces a deep learning algorithm leveraging Bidirectional Long Short-Term Memory cells and a hybrid probabilistic methodology for weight updating. This approach enhances scalability by adapting to variations in the number of UEs without requiring retraining. Additionally, the study presents a training methodology that improves scalability not only with respect to the number of UEs but also to the number of APs. Furthermore, a variant of the proposed AP-UE algorithm ensures robustness against pilot contamination effects, a critical issue arising from pilot reuse in channel estimation. Extensive numerical results validate the effectiveness and adaptability of the proposed methods, demonstrating their superiority over widely used heuristic alternatives.

A General Framework for Scalable UE-AP Association in User-Centric Cell-Free Massive MIMO based on Recurrent Neural Networks

TL;DR

The paper tackles AP-UE association in CF-mMIMO, a problem made intractable by scaling to large numbers of UEs and APs and by pilot contamination. It proposes a BiLSTM-based, master-centric deep learning framework with a probabilistic training scheme (Bernoulli sampling) to dynamically form UE-centric AP clusters, yielding scalability with respect to both UEs and APs. It introduces multi-objective loss functions (SUM, BALANCE, MIN) and a pilot-contamination robust variant that augments inputs with pilot indicators, and demonstrates superior performance over heuristic baselines through extensive simulations. The approach enables distributed, parallelizable learning and can adapt to different network goals, making it well suited for large-scale, next-generation CF-mMIMO deployments.

Abstract

This study addresses the challenge of access point (AP) and user equipment (UE) association in cell-free massive MIMO networks. It introduces a deep learning algorithm leveraging Bidirectional Long Short-Term Memory cells and a hybrid probabilistic methodology for weight updating. This approach enhances scalability by adapting to variations in the number of UEs without requiring retraining. Additionally, the study presents a training methodology that improves scalability not only with respect to the number of UEs but also to the number of APs. Furthermore, a variant of the proposed AP-UE algorithm ensures robustness against pilot contamination effects, a critical issue arising from pilot reuse in channel estimation. Extensive numerical results validate the effectiveness and adaptability of the proposed methods, demonstrating their superiority over widely used heuristic alternatives.

Paper Structure

This paper contains 24 sections, 32 equations, 10 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. State of the art
  3. Contributions
  4. Preliminaries on LSTM networks
  5. User-Centric CF-mMIMO model and transceiver processing
  6. Master selection
  7. Channel estimation
  8. Downlink operation
  9. Problem formulation
  10. Cumulative Weighted Spectral Efficiency
  11. Spectral Efficiency vs. Number of Connections
  12. Minimum Spectral Efficiency
  13. Scalable AP-UE association based on DL
  14. Scalable Solution for UEs
  15. Model Inference
  16. ...and 9 more sections

Figures (10)

  • Figure 1: A user-centric CF-mMIMO paradigm illustrating multiple low-complexity APs serving a set of UEs within the same time-frequency slot, all interconnected through fronthaul links to a central processing unit (CPU).
  • Figure 2: Conceptual scheme of an LSTM cell.
  • Figure 3: Conceptual scheme of a BiLSTM network.
  • Figure 4: Deep learning framework using UE ordering by master APs.
  • Figure 5: Statistics for non-interfering pilot scenario: (a) CDFs of SE sums; (b) Distributions of activated connection counts; (c) CDFs of SE minima.
  • ...and 5 more figures