Joint Channel Estimation and User Grouping for Massive MIMO Systems

TL;DR

This paper proposes a new Bayesian framework for joint downlink channel estimation and user grouping in massive MIMO systems, effectively handling realistic sparsity patterns including outliers to improve performance.

Contribution

It introduces a general sparsity model with shared and individual components and develops a Bayesian method to better exploit sparsity in practical scenarios.

Findings

Significant performance improvements over existing methods.

Effective handling of outliers in user sparsity patterns.

Enhanced channel estimation accuracy in massive MIMO.

Abstract

This paper addresses the problem of joint downlink channel estimation and user grouping in massive multiple-input multiple-output (MIMO) systems, where the motivation comes from the fact that the channel estimation performance can be improved if we exploit additional common sparsity among nearby users. In the literature, a commonly used group sparsity model assumes that users in each group share a uniform sparsity pattern. In practice, however, this oversimplified assumption usually fails to hold, even for physically close users. Outliers deviated from the uniform sparsity pattern in each group may significantly degrade the effectiveness of common sparsity, and hence bring limited (or negative) gain for channel estimation. To better capture the group sparse structure in practice, we provide a general model having two sparsity components: commonly shared sparsity and individual sparsity, where the additional individual sparsity accounts for any outliers. Then, we propose a novel sparse Bayesian learning (SBL)-based framework to address the joint channel estimation and user grouping problem under the general sparsity model. The framework can fully exploit the common sparsity among nearby users and exclude the harmful effect from outliers simultaneously. Simulation results reveal substantial performance gains over the existing state-of-the-art baselines.