Wavelet entropy of stochastic processes

L. Zunino,D. G. Perez,M. Garavaglia,O. A. Rosso

DOI: https://doi.org/10.1016/j.physa.2006.12.057

2006-03-18

Abstract:We compare two different definitions for the wavelet entropy associated to stochastic processes. The first one, the Normalized Total Wavelet Entropy (NTWS) family [Phys. Rev. E 57 (1998) 932; J. Neuroscience Method 105 (2001) 65; Physica A (2005) in press] and a second introduced by Tavares and Lucena [Physica A 357 (2005)~71]. In order to understand their advantages and disadvantages, exact results obtained for fractional Gaussian noise (-1<alpha< 1) and the fractional Brownian motion (1 < alpha < 3) are assessed. We find out that NTWS family performs better as a characterization method for these stochastic processes.

Data Analysis, Statistics and Probability,Computational Physics

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to evaluate and compare the performance of two differently - defined wavelet entropies in characterizing random processes, especially fractional Brownian motion (fBm) and fractional Gaussian noise (fGn). Specifically:

1. **Problem Background**:
- Wavelet Analysis is a powerful tool for processing data samples, especially for non - stationary signals.
- Wavelet transform can provide time - frequency representation and has the localization property, being able to detect dynamic changes.
- Entropy, as a measure, can quantify the degree of order - disorder in multi - frequency signal responses.

2. **Research Objectives**:
- Compare two different definitions of wavelet entropy: the Normalized Total Wavelet Entropy (NTWS) family and the Tavares - Lucena Wavelet Entropy (TLWS).
- Evaluate the advantages and disadvantages of these two definitions by analyzing fractional Brownian motion and fractional Gaussian noise.

3. **Specific Problems**:
- For fractional Brownian motion ($1 < \alpha < 3$) and fractional Gaussian noise ($- 1 < \alpha < 1$), which definition of wavelet entropy can better characterize these random processes?
- Analyze the influence of different resolution levels ($N$) on the two entropy definitions.

4. **Main Conclusions**:
- The NTWS family performs better in characterizing fractional Brownian motion and fractional Gaussian noise.
- The NTWS family covers almost the entire range from 0 to 1, while the TLWS only covers about 25% of the range.
- The NTWS family shows better performance in distinguishing short - memory and long - memory processes.

Through these analyses, the author hopes to find a more effective definition of wavelet entropy to better understand and characterize complex random processes.