Performance Limits of Massive MIMO Systems Based on Bayes-Optimal Inference

TL;DR

This paper analyzes the fundamental performance limits of massive MIMO systems using Bayesian inference, revealing that interference can be effectively separated and MSE decreases with system size, without residual interference.

Contribution

It provides a replica analysis of the MMSE in massive MIMO using Bayesian inference, showing interference separation and MSE reduction as system dimensions grow.

Findings

Interference from adjacent cells can be separated without pilot signals.

MSE decreases with the number of antennas and data symbols.

No residual interference remains as system size increases.

Abstract

This paper gives a replica analysis for the minimum mean square error (MSE) of a massive multiple-input multiple-output (MIMO) system by using Bayesian inference. The Bayes-optimal estimator is adopted to estimate the data symbols and the channels from a block of received signals in the spatial-temporal domain. We show that using the Bayes-optimal estimator, the interfering signals from adjacent cells can be separated from the received signals without pilot information. In addition, the MSEs with respect to the data symbols and the channels of the desired users decrease with the number of receive antennas and the number of data symbols, respectively. There are no residual interference terms that remain bounded away from zero as the numbers of receive antennas and data symbols approach infinity.