Alternating Channel Estimation and Prediction for Cell-Free mMIMO with Channel Aging: A Deep Learning Based Scheme

TL;DR

This paper introduces a deep learning-based hybrid channel estimation and prediction scheme for cell-free mMIMO systems, significantly reducing overhead and improving spectral efficiency amid user mobility and channel aging.

Contribution

It proposes a novel CEP scheme combining MMSE estimation and DNN prediction to lower overhead and enhance performance in dynamic wireless networks.

Findings

Reduces CE overhead by at least 50%

Maintains negligible CE error with practical user mobility

Improves spectral efficiency compared to conventional methods

Abstract

In large scale dynamic wireless networks, the amount of overhead caused by channel estimation (CE) is becoming one of the main performance bottlenecks. This is due to the large number users whose channels should be estimated, the user mobility, and the rapid channel change caused by the usage of the high-frequency spectrum (e.g. millimeter wave). In this work, we propose a new hybrid channel estimation/prediction (CEP) scheme to reduce overhead in time-division duplex (TDD) wireless cell-free massive multiple-input-multiple-output (mMIMO) systems. The scheme proposes sending a pilot signal from each user only once in a given number (window) of coherence intervals (CIs). Then minimum mean-square error (MMSE) estimation is used to estimate the channel of this CI, while a deep neural network (DNN) is used to predict the channels of the remaining CIs in the window. The DNN exploits the temporal correlation between the consecutive CIs and the received pilot signals to improve the channel prediction accuracy. By doing so, CE overhead is reduced by at least 50 percent at the expense of negligible CE error for practical user mobility settings. Consequently, the proposed CEP scheme improves the spectral efficiency compared to the conventional MMSE CE approach, especially when the number of users is large, which is demonstrated numerically.