Blind User Activity Detection for Grant-Free Random Access in Cell-Free mMIMO Networks

TL;DR

This paper introduces a fully data-driven deep learning approach for user activity detection in cell-free massive MIMO networks, eliminating the need for prior large-scale fading estimation and outperforming existing methods.

Contribution

It proposes a novel blind deep learning-based method for user activity detection in CF-mMIMO networks that does not require large-scale fading coefficients.

Findings

The DL-based method outperforms covariance-based approaches.

It effectively merges distributed antenna information for activity detection.

The approach is fully data-driven and does not rely on prior channel knowledge.

Abstract

Cell-free massive MIMO (CF-mMIMO) networks have recently emerged as a promising solution to tackle the challenges arising from next-generation massive machine-type communications. In this paper, a fully grant-free deep learning (DL)-based method for user activity detection in CF-mMIMO networks is proposed. Initially, the known non-orthogonal pilot sequences are used to estimate the channel coefficients between each user and the access points. Then, a deep convolutional neural network is used to estimate the activity status of the users. The proposed method is "blind", i.e., it is fully data-driven and does not require prior large-scale fading coefficients estimation. Numerical results show how the proposed DL-based algorithm is able to merge the information gathered by the distributed antennas to estimate the user activity status, yet outperforming a state-of-the-art covariance-based method.