Robust estimation of the parameters of a disturbed non-stationary Gaussian process

TL;DR

This paper introduces a robust method for estimating Gaussian process parameters in non-stationary, noisy gravitational wave data, addressing impulsive disturbances with a matched filter applied to AR histograms.

Contribution

The paper presents a novel parameter estimation technique that improves robustness in non-stationary Gaussian noise with impulsive disturbances in gravitational wave data.

Findings

Enhanced parameter estimation accuracy in disturbed non-stationary noise

Effective suppression of impulsive disturbances

Improved detection sensitivity for gravitational waves

Abstract

A typical problem in the detection of the gravitational waves in the data of gravitational antennas is the non-stationarity of the Gaussian noise (and so the varying sensitivity) and the presence of big impulsive disturbances. In such conditions the estimation of the standard deviation of the Gaussian process done with a classical estimator applied after a "rough" cleaning of the big pulses often gives poor results. We propose a method based on a matched filter applied to an AR histogram of the absolute value of the data