Signal representation and segmentation based on multifractal stationarity

We present a new scheme for signal representation which is well suited for the study of multifractal features. In particular, our approach, which is based on the use of Weakly Self-Affine functions, allows to segment a signal into parts which are "multifractally homogeneous". Furthermore, it opens the possibility of estimating non-concave multifractal spectra, a valuable improvement for many practical applications such as Internet traffic modeling.