RIS-Aided Near-Field Channel Estimation under Mutual Coupling and Spatial Correlation

TL;DR

This paper improves near-field channel estimation in RIS-aided MIMO systems by incorporating mutual coupling effects into the RS-LS estimator, significantly enhancing accuracy and performance in complex propagation environments.

Contribution

It advances the RS-LS estimation method by explicitly modeling mutual coupling effects in near-field RIS-aided MIMO channels, addressing high-dimensional estimation challenges.

Findings

Incorporating mutual coupling improves estimation accuracy by about 5 dB at 5 dB SNR.

The analysis shows mutual coupling affects the spatial degrees of freedom significantly.

The proposed method outperforms conventional estimators that ignore mutual coupling.

Abstract

The integration of reconfigurable intelligent surfaces (RIS) with extremely large multiple-input multiple-output (MIMO) arrays at the base station has emerged as a key enabler for enhancing wireless network performance. However, this setup introduces high-dimensional channel matrices, leading to increased computational complexity and pilot overhead in channel estimation. Mutual coupling (MC) effects among densely packed unit cells, spatial correlation, and near-field propagation conditions further complicate the estimation process. Conventional estimators, such as linear minimum mean square error (MMSE), require channel statistics that are challenging to acquire for high-dimensional arrays, while least squares (LS) estimators suffer from performance limitations. To address these challenges, the reduced-subspace least squares (RS-LS) estimator leverages array geometry to enhance estimation accuracy. This work advances the promising RS-LS estimation algorithm by explicitly incorporating MC effects into the more realistic and challenging near-field propagation environment within the increasingly relevant generalized RIS-aided MIMO framework. Additionally, we investigate the impact of MC on the spatial degrees of freedom (DoF). Our analysis reveals that accounting for MC effects provides a significant performance gain of approximately 5 dB at an SNR of 5 dB, compared to conventional methods that ignore MC.