Statistical comparison of three analysis methods of fBm sequences

Mandelbrot and Van Ness' fractional Brownian motion (fBm) has been considered a suitable model to describe the long-range correlations exhibited by a wide variety of physiological signals, such as the heart rate. The fBm is a temporal fractal whose properties of self-affinity and infinite inter-dependency of its increments are characterized by a parameter H, referred to as the Hurst exponent. This parameter is quite useful in analysing biomedical signals and determining certain pathologies or physiological states these signals may exhibit. Over the years, several algorithms for the computation of H have been published. In this study we compare the performance of the well-known Higuchi's method, with two other relatively recent ones, DePetrillo's and Chang's methods. The results reveal that the last one seems to be the most accurate in analysing these sequences.