LMMSE Receivers in Uplink Massive MIMO Systems with Correlated Rician Fading

Ikram Boukhedimi, Abla Kammoun, Mohamed-Slim Alouini

Abstract

We carry out a theoretical analysis of the uplink (UL) of a massive MIMO system with per-user channel correlation and Rician fading, using two processing approaches. Firstly, we examine the linear minimum-mean-square-error receiver under training-based imperfect channel estimates. Secondly, we propose a statistical combining technique that is more suitable in environments with strong Line-of-Sight (LoS) components. We derive closed-form asymptotic approximations of the UL spectral efficiency (SE) attained by each combining scheme in single and multi-cell settings, as a function of the system parameters. These expressions are insightful in how different factors such as LoS propagation conditions and pilot contamination impact the overall system performance. Furthermore, they are exploited to determine the optimal number of training symbols which is shown to be of significant interest at low Rician factors. The study and numerical results substantiate that stronger LoS signals lead to better performances, and under such conditions, the statistical combining entails higher SE gains than the conventional receiver.

Introduction

In the evolution toward 5G and 6G networks, Massive MIMO (Multiple-Input Multiple-Output) remains a cornerstone technology. While much of the foundational research in this field assumes Rayleigh fading—where signals reach the receiver only through indirect scattering—real-world deployments, particularly in millimeter-wave (mmWave) and small-cell environments, often feature a dominant Line-of-Sight (LoS) component. This physical reality is best modeled by Rician fading.

This paper addresses the performance of uplink Massive MIMO systems under Rician fading, specifically focusing on the interplay between channel estimation overhead, spatial correlation, and pilot contamination. We contrast the performance of the standard Linear Minimum Mean Square Error (LMMSE) receiver with a proposed statistical combining approach designed to exploit the deterministic nature of the LoS component.

The Challenge: Estimation and Contamination

Conventional Massive MIMO receivers rely on an uplink training phase where users transmit pilot sequences to allow the Base Station (BS) to estimate the instantaneous Channel State Information (CSI). However, this approach faces two significant hurdles in Rician environments:

  1. Estimation Overhead: In high-LoS scenarios, the channel is dominated by a stable, deterministic component. Spending significant time and power to estimate the rapidly changing scattered component may yield diminishing returns.
  2. Pilot Contamination: In multi-cell networks, users in adjacent cells often reuse the same pilot sequences. This creates interference that does not vanish even as the number of antennas ($N$) grows, setting a fundamental ceiling on spectral efficiency (SE).

Proposed Methodology

To analyze these challenges, we utilize Random Matrix Theory (RMT) to derive closed-form asymptotic approximations for the SE in both single-cell and multi-cell settings. The analysis considers the large-antenna limit ($N \to \infty$) while keeping the number of users ($K$) fixed. We compare two receiver architectures:

  • Conventional LMMSE Receiver: This receiver uses training-based imperfect channel estimates. While it is the industry standard, its performance is limited by the quality of the pilot-based estimates and is highly susceptible to pilot contamination.
  • Statistical Combiner: We propose a receiver that utilizes long-term statistics—specifically the deterministic LoS component and spatial correlation matrices—rather than instantaneous estimates of the scattered path. Because this method does not rely on pilot sequences for the fluctuating part of the signal, it is inherently immune to pilot contamination.

Key Insights and the Training Paradox

Our theoretical analysis reveals several critical insights into the behavior of Rician Massive MIMO systems:

The Training Paradox

In environments with a high Rician factor (strong LoS), we find that increasing the training sequence length ($\tau$) is often counterproductive. The analysis proves that the optimal training length $\tau^*$ approaches the minimum possible value (the number of users, $K$) as the Rician factor increases. In these cases, the resources are better spent on data transmission rather than refining estimates of a negligible scattered component.

Statistical "Bypassing"

The proposed statistical receiver effectively "bypasses" the need for instantaneous CSI. When the direct path is dominant, the errors introduced by trying to estimate a noisy, weak scattered path can actually degrade performance. By focusing on the stable LoS path, the statistical combiner provides a more robust signal recovery mechanism.

Resilience to Contamination

The most significant advantage of the statistical combiner appears in multi-cell environments. Since it does not utilize pilots to estimate the scattered components, the "contamination" effect from neighboring cells' pilots is entirely mitigated, allowing for significantly higher SE in interference-limited scenarios.

Numerical Results

We validated our asymptotic derivations using Monte Carlo simulations (1,000 realizations) based on standard pathloss and correlation models (e.g., the "one-ring" model). Key findings include:

  • Performance Thresholds: In single-cell scenarios, the statistical receiver begins to outperform the conventional LMMSE when the Rician factor $\kappa$ exceeds a specific threshold (typically $\kappa > 4$).
  • Multi-Cell Gains: In a 3-cell setup with users at the cell edge, the statistical combiner achieved a 33% Spectral Efficiency gain over the conventional LMMSE at a Rician factor of 2. This highlights the receiver's efficacy in mitigating inter-cell interference.
  • Asymptotic Accuracy: The derived closed-form expressions proved to be highly accurate even for finite system dimensions, such as $N=150$ antennas, making them reliable tools for system design.

Conclusion and Future Work

This study demonstrates that in the presence of strong LoS components, traditional training-based receivers may not be the optimal choice for Massive MIMO. Statistical combining offers a path to higher spectral efficiency and immunity to pilot contamination.

Future research should address the limitations of this study, including the assumption of static LoS components and the fixed power allocation. Extending the analysis to high-mobility scenarios and investigating joint power control and receiver optimization will be essential for the practical implementation of these insights in 6G architectures.

Summary updated

Abstract

We carry out a theoretical analysis of the uplink (UL) of a massive MIMO system with per-user channel correlation and Rician fading, using two processing approaches. Firstly, we examine the linear minimum-mean-square-error receiver under training-based imperfect channel estimates. Secondly, we propose a statistical combining technique that is more suitable in environments with strong Line-of-Sight (LoS) components. We derive closed-form asymptotic approximations of the UL spectral efficiency (SE) attained by each combining scheme in single and multi-cell settings, as a function of the system parameters. These expressions are insightful in how different factors such as LoS propagation conditions and pilot contamination impact the overall system performance. Furthermore, they are exploited to determine the optimal number of training symbols which is shown to be of significant interest at low Rician factors. The study and numerical results substantiate that stronger LoS signals lead to better performances, and under such conditions, the statistical combining entails higher SE gains than the conventional receiver.
BibTeX
@article{boukhedimi2019lmmse,
  author = {Ikram Boukhedimi and Abla Kammoun and Mohamed-Slim Alouini},
  title = {{LMMSE Receivers in Uplink Massive MIMO Systems with Correlated Rician Fading}},
  journal = {IEEE Trans. Communications},
  year = {2019},
  url = {https://openreview.net/forum?id=wimE78B_8Sd},
}
Figure