Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Generative Diffusion Model Driven Massive Random Access in Massive MIMO Systems

Keke Ying, Zhen Gao, Sheng Chen, Tony Q. S. Quek, H. Vincent Poor

TL;DR

<3-5 sentence high-level summary> The paper tackles massive random access in massive MIMO by separating initial user activity detection from joint channel and data recovery. It introduces VPL-AUDNet, a Transformer-based AUD that works across variable pilot lengths and leverages a spatial-correlation module and pilot-length awareness for robustness. For JCEDD, it develops a diffusion-model framework with predictor-corrector sampling to iteratively estimate channels and data, using a learned prior and closed-form likelihood components, and it operates asynchronously to improve convergence. Empirical results show substantial improvements over traditional CS/LS-based methods and prior diffusion approaches in both Rayleigh and 3GPP channel settings, highlighting the approach’s potential for scalable, high-density wireless access and ultra-massive connectivity.

Abstract

Massive random access is an important technology for achieving ultra-massive connectivity in next-generation wireless communication systems. It aims to address key challenges during the initial access phase, including active user detection (AUD), channel estimation (CE), and data detection (DD). This paper examines massive access in massive multiple-input multiple-output (MIMO) systems, where deep learning is used to tackle the challenging AUD, CE, and DD functions. First, we introduce a Transformer-AUD scheme tailored for variable pilot-length access. This approach integrates pilot length information and a spatial correlation module into a Transformer-based detector, enabling a single model to generalize across various pilot lengths and antenna numbers. Next, we propose a generative diffusion model (GDM)-driven iterative CE and DD framework. The GDM employs a score function to capture the posterior distributions of massive MIMO channels and data symbols. Part of the score function is learned from the channel dataset via neural networks, while the remaining score component is derived in a closed form by applying the symbol prior constellation distribution and known transmission model. Utilizing these posterior scores, we design an asynchronous alternating CE and DD framework that employs a predictor-corrector sampling technique to iteratively generate channel estimation and data detection results during the reverse diffusion process. Simulation results demonstrate that our proposed approaches significantly outperform baseline methods with respect to AUD, CE, and DD.

Generative Diffusion Model Driven Massive Random Access in Massive MIMO Systems

TL;DR

<3-5 sentence high-level summary> The paper tackles massive random access in massive MIMO by separating initial user activity detection from joint channel and data recovery. It introduces VPL-AUDNet, a Transformer-based AUD that works across variable pilot lengths and leverages a spatial-correlation module and pilot-length awareness for robustness. For JCEDD, it develops a diffusion-model framework with predictor-corrector sampling to iteratively estimate channels and data, using a learned prior and closed-form likelihood components, and it operates asynchronously to improve convergence. Empirical results show substantial improvements over traditional CS/LS-based methods and prior diffusion approaches in both Rayleigh and 3GPP channel settings, highlighting the approach’s potential for scalable, high-density wireless access and ultra-massive connectivity.

Abstract

Massive random access is an important technology for achieving ultra-massive connectivity in next-generation wireless communication systems. It aims to address key challenges during the initial access phase, including active user detection (AUD), channel estimation (CE), and data detection (DD). This paper examines massive access in massive multiple-input multiple-output (MIMO) systems, where deep learning is used to tackle the challenging AUD, CE, and DD functions. First, we introduce a Transformer-AUD scheme tailored for variable pilot-length access. This approach integrates pilot length information and a spatial correlation module into a Transformer-based detector, enabling a single model to generalize across various pilot lengths and antenna numbers. Next, we propose a generative diffusion model (GDM)-driven iterative CE and DD framework. The GDM employs a score function to capture the posterior distributions of massive MIMO channels and data symbols. Part of the score function is learned from the channel dataset via neural networks, while the remaining score component is derived in a closed form by applying the symbol prior constellation distribution and known transmission model. Utilizing these posterior scores, we design an asynchronous alternating CE and DD framework that employs a predictor-corrector sampling technique to iteratively generate channel estimation and data detection results during the reverse diffusion process. Simulation results demonstrate that our proposed approaches significantly outperform baseline methods with respect to AUD, CE, and DD.
Paper Structure (32 sections, 30 equations, 17 figures, 3 tables, 2 algorithms)

This paper contains 32 sections, 30 equations, 17 figures, 3 tables, 2 algorithms.

Table of Contents

  1. Introduction
  2. Related Work
  3. Our Contributions
  4. System Model
  5. Problem Formulation
  6. Stage 1-AUD
  7. Stage 2-JCEDD
  8. Transformer-Based AUD for Variable Pilot Lengths
  9. Preprocessing
  10. Heterogeneous Transformer and Activity Decoder
  11. Pilot Length Adaptive Module
  12. Loss Function
  13. Generative Diffusion Models Driven JCEDD Using Predictor-Corrector Sampler
  14. Preliminary: Diffusion Process and SDE
  15. Channel Estimation
  16. ...and 17 more sections

Figures (17)

  • Figure 1: System diagram of the massive random access with variable-length pilot transmission and corresponding VPL-AUDNet at the receiver.
  • Figure 2: Block diagram of the proposed spatial correlation module.
  • Figure 3: Block diagram of the proposed pilot length adaptive module.
  • Figure 4: Block diagram of the proposed PC sampler-based CE algorithm.
  • Figure 5: Block diagram of the proposed PC sampler-based JCEDD algorithm.
  • ...and 12 more figures