Off-Grid Aware Spatial Covariance Estimation in mmWave Communications

TL;DR

This paper presents a novel off-grid aware spatial covariance estimation method for mmWave MIMO systems that improves accuracy by addressing basis mismatch issues inherent in traditional compressed sensing approaches.

Contribution

It introduces a parameter perturbed framework combined with OMP to effectively mitigate basis mismatch without increasing discretization complexity.

Findings

Enhanced covariance estimation accuracy

Improved efficiency over conventional CS algorithms

Effective handling of off-grid effects

Abstract

This work investigates the problem of spatial covariance matrix estimation in a millimeter-wave (mmWave) hybrid multiple-input multiple-output (MIMO) system with an emphasis on the basis-mismatch effect. The basis mismatch is prevalent in the compressed sensing (CS) schemes which adopt discretization procedure. In such an approach, the algorithm yields a finite discrete point which is an approximation to the continuous parametric space. The quality of this approximation depends on the number of discretized points in the dictionary. Instead of increasing the number of discretized points to combat this off-grid effect, we propose an efficient parameter perturbed framework which uses a controlled perturbation mechanism in conjunction with the orthogonal matching pursuit (OMP) algorithm. Numerical results verify the performance improvement through our proposed algorithm in terms of relative efficiency metric, which is basically due to taking care of the off-grid effect carefully that is ignored in the conventional CS algorithms.