Comparison of information criteria for detection of useful signal in noisy environment

TL;DR

This paper evaluates various information criteria, including entropy and complexity measures, for detecting useful acoustic signals in noisy environments, demonstrating the effectiveness of statistical complexity especially at high noise levels.

Contribution

It introduces analytical formulas for spectral complexity and disequilibrium and compares their effectiveness with other information metrics in noisy conditions.

Findings

Statistical complexity outperforms other metrics at high noise levels.

Analytical formulas for spectral complexity are derived.

Proposed approach is effective across different acoustic signals.

Abstract

This paper considers the problem of appearance indication of useful acoustic signal in the signal/noise mixture. Various information characteristics (information entropy, Jensen-Shannon divergence, spectral information divergence and statistical complexity) are investigated in the context of solving this problem. Both time and frequency domain are studied for information entropy calculation. The effectiveness of statistical complexity is shown in comparison with other information metrics for different levels of added white noise. In addition analytical formulas for complexity and disequilibrium are obtained using entropy variation in the cases of one- and two-dimensional spectral distributions. The effectiveness of the proposed approach is shown for different types of acoustic signals and noise level especially when additional noise characteristics estimation is impossible.