Structured Channel Covariance Estimation from Limited Samples in Massive\n MIMO

Abstract

Obtaining channel covariance knowledge is of great importance in various\nMultiple-Input Multiple-Output MIMO communication applications, including\nchannel estimation and covariance-based user grouping. In a massive MIMO\nsystem, covariance estimation proves to be challenging due to the large number\nof antennas ($M\\gg 1$) employed in the base station and hence, a high signal\ndimension. In this case, the number of pilot transmissions $N$ becomes\ncomparable to the number of antennas and standard estimators, such as the\nsample covariance, yield a poor estimate of the true covariance and are\nundesirable. In this paper, we propose a Maximum-Likelihood (ML) massive MIMO\ncovariance estimator, based on a parametric representation of the channel\nangular spread function (ASF). The parametric representation emerges from\nsuper-resolving discrete ASF components via the well-known MUltiple SIgnal\nClassification (MUSIC) method plus approximating its continuous component using\nsuitable limited-support density function. We maximize the likelihood function\nusing a concave-convex procedure, which is initialized via a non-negative\nleast-squares optimization problem. Our simulation results show that the\nproposed method outperforms the state of the art in various estimation quality\nmetrics and for different sample size to signal dimension ($N/M$) ratios.\n