Year: 2026
Type: article
Abstract: This paper proposes an entropy-consistent temporal segmentation method for the formation of complex signal ensembles, which is based on the use of an entropy measure simultaneously as an integral indicator of the signal structural complexity and as a control variable for the coordinated selection of temporal decomposition parameters. A distinctive feature of the proposed method is the enhanced sensitivity to local structural transformations of the signal and the improved noise robustness of ensemble characteristics under nonlinear dynamics and intensive interference conditions. Unlike classical entropy-based approaches, including ordinal permutation entropy and weighted ordinal permutation entropy, which are computed for a fixed segmentation scale, the proposed method provides coordination of the temporal segmentation scale with the local structural heterogeneity of the signal. Within the proposed framework, the entropy measure is employed as a feedback mechanism that enables adaptive modification of temporal decomposition during transitions between irregular and regularized dynamical regimes. Experimental verification of the method was performed using nonlinear signals with controlled dynamics in the presence of additive white Gaussian noise at different signal-to-noise ratio (SNR) levels. The obtained results demonstrate that the proposed entropy-consistent method ensures enhanced robustness of entropy estimation to noise, reducing metric degradation to 2,2% at SNR = −10 dB, whereas the corresponding reductions for the classical PE and WPE methods amount to 3,9% and 4,1%, respectively. In addition, the method provides improved contrast between different dynamical regimes of the signal and reduced variability of entropy estimates under repeated simulations. Thus, the proposed entropy-consistent temporal segmentation method enables efficient and noise-robust formation of complex signal ensembles.
Source: Science-based technologies
Authors: Lyubomyr Нoleychuk, Volodymyr Lysechko
Language: English
Subfield: Signal Processing
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7157009339
FWCI:
Cites:
Cited by:
Related to: