Covariance Matrix Estimation for Massive MIMO

TL;DR

This paper introduces a new pilot structure for massive MIMO systems that improves covariance matrix estimation by using two pilot sequences with a random phase shift, enhancing flexibility and accuracy.

Contribution

The paper proposes a novel pilot scheme with staggered pilots and random phase shifts, enabling better covariance estimation without requiring all users to transmit pilots simultaneously.

Findings

Improved covariance matrix estimation accuracy.

Enhanced achievable rate with the proposed pilot structure.

Better mean-squared error performance compared to existing methods.

Abstract

We propose a novel pilot structure for covariance matrix estimation in massive multiple-input multiple-output (MIMO) systems in which each user transmits two pilot sequences, with the second pilot sequence multiplied by a random phase-shift. The covariance matrix of a particular user is obtained by computing the sample cross-correlation of the channel estimates obtained from the two pilot sequences. This approach relaxes the requirement that all the users transmit their uplink pilots over the same set of symbols. We derive expressions for the achievable rate and the mean-squared error of the covariance matrix estimate when the proposed method is used with staggered pilots. The performance of the proposed method is compared with existing methods through simulations.