Asymptotic Moments for Interference Mitigation in Correlated Fading Channels

TL;DR

This paper derives asymptotic moments for large random matrices with correlated entries, enabling efficient interference mitigation in wireless systems with large antenna arrays.

Contribution

It provides deterministic approximations of moments for a class of large correlated random matrices and applies these results to design a low-complexity polynomial detector.

Findings

Deterministic moment approximations improve interference mitigation.

Polynomial expansion detector achieves near-optimal performance.

Simulation confirms accuracy of asymptotic analysis for finite systems.

Abstract

We consider a certain class of large random matrices, composed of independent column vectors with zero mean and different covariance matrices, and derive asymptotically tight deterministic approximations of their moments. This random matrix model arises in several wireless communication systems of recent interest, such as distributed antenna systems or large antenna arrays. Computing the linear minimum mean square error (LMMSE) detector in such systems requires the inversion of a large covariance matrix which becomes prohibitively complex as the number of antennas and users grows. We apply the derived moment results to the design of a low-complexity polynomial expansion detector which approximates the matrix inverse by a matrix polynomial and study its asymptotic performance. Simulation results corroborate the analysis and evaluate the performance for finite system dimensions.