Quantifying long-range correlations with a multiscale ordinal pattern approach

Abstract

In this paper we use the ordinal patterns probabilities associated with fractional Brownian motions for estimating the Hurst exponent of artificially generated and experimentally measured data. Numerical analysis show a reliable estimation of this scaling parameter, even when data with low resolution are analysed. Robustness to observational noise is also obtained. Several experimental applications allow us to confirm the practical utility of the proposed approach. We contrast results obtained by implementing this multiscale symbolic tool with those obtained from the classical detrended fluctuation analysis.