Low-Cost Maximum Entropy Covariance Matrix Reconstruction Algorithm for Robust Adaptive Beamforming

TL;DR

This paper introduces a low-complexity adaptive beamforming method that uses maximum entropy power spectrum techniques to efficiently estimate covariance matrices and improve signal steering vector estimation, reducing computational load.

Contribution

It proposes a novel stochastic gradient-based algorithm leveraging MEPS for covariance matrix reconstruction, enhancing adaptive beamforming performance with lower complexity.

Findings

Outperforms previous beamformers in simulations

Reduces computational complexity by avoiding matrix inversions

Improves accuracy of steering vector estimation

Abstract

In this letter, we present a novel low-complexity adaptive beamforming technique using a stochastic gradient algorithm to avoid matrix inversions. The proposed method exploits algorithms based on the maximum entropy power spectrum (MEPS) to estimate the noise-plus-interference covariance matrix (MEPS-NPIC) so that the beamforming weights are updated adaptively, thus greatly reducing the computational complexity. MEPS is further used to reconstruct the desired signal covariance matrix and to improve the estimate of the desired signals's steering vector (SV). Simulations show the superiority of the proposed MEPS-NPIC approach over previously proposed beamformers.