RMT Estimator with Adaptive Decision Criteria for Estimating the Number of Signals Based on Random Matrix Theory

TL;DR

This paper introduces an adaptive RMT-based estimator for accurately determining the number of signals in noise, effectively addressing bias issues and outperforming existing methods through simulation validation.

Contribution

It proposes an RMT estimator with adaptive decision criteria that incorporate bias correction, improving signal number estimation accuracy in finite sample scenarios.

Findings

Significantly reduces under-estimation of signals.

Adapts decision criteria based on eigenvalue bias analysis.

Outperforms existing estimators in simulations.

Abstract

Estimating the number of signals embedded in noise is a fundamental problem in signal processing. As a classic estimator based on random matrix theory (RMT), the RMT estimator estimates the number of signals via sequentially testing the likelihood of an eigenvalue as arising from a signal or noise for a given over-detection probability. However, it tends to under-estimate the number of signals as weak signal eigenvalues may be immersed in the non-negligible bias term among eigenvalues for finite sample size. In order to solve this problem, we propose an RMT estimator with adaptive decision criterion (termed as RMT-ADC estimator) by adaptively incorporating the bias term into the decision criterion of the RMT estimator. Firstly, we analyze the effect of this bias term among eigenvalues on the estimation performance of the RMT estimator. Then, we derive both the decreased over-estimation probability and the increased under-estimation probability of the RMT estimator incurred by the bias term when assuming the eigenvalue being tested is arising from a signal, and also derive the increased under-estimation probability of the RMT estimator incurred by the bias term when assuming the eigenvalue being tested is arising from noise. Based on these results, the RMT-ADC estimator can adaptively determine whether the noise variance should be estimated under the assumption that the eigenvalue being tested is arising from a signal or from noise, and thus can adaptively select its decision criterion. Moreover, the RMT-ADC estimator can adaptively determine whether the bias term among eigenvalues should be incorporated into the selected decision criterion or not. Therefore, the RMT-ADC estimator can avoid the higher under-estimation probability of the RMT estimator. Finally, simulation results are presented to show that the proposed RMT-ADC estimator significantly outperforms the existing estimators.