Degree weighted recurrence networks for the analysis of time series data

TL;DR

This paper introduces a novel weighted recurrence network method for analyzing time series, revealing fractal structures and distinguishing chaotic dynamics from noise, especially useful for short, noisy data.

Contribution

The paper proposes a new weighted recurrence network construction and generalized network measures, enhancing the analysis of chaotic attractors and noise discrimination.

Findings

Node strength distribution follows a power law with exponential tail.

Measures effectively discriminate chaos from white and colored noise.

Networks from all standard chaotic attractors belong to a new class.

Abstract

Recurrence networks are powerful tools used effectively in the nonlinear analysis of time series data. The analysis in this context is done mostly with unweighted and undirected complex networks constructed with specific criteria from the time series. In this work, we propose a novel method to construct "weighted recurrence network"(WRN) from a time series and show how it can reveal useful information regarding the structure of a chaotic attractor, which the usual unweighted recurrence network cannot provide. Especially, we find the node strength distribution of the WRN, from every chaotic attractor follows a power law (with exponential tail) with the index characteristic to the fractal structure of the attractor. This leads to a new class among complex networks, to which networks from all standard chaotic attractors are found to belong. In addition, we present generalized definitions for clustering coefficient and characteristic path length and show that these measures can effectively discriminate chaotic dynamics from white noise and $1/f$ colored noise. Our results indicate that the WRN and the associated measures can become potentially important tools for the analysis of short and noisy time series from the real world systems as they are clearly demarked from that of noisy or stochastic systems.