Discrete Wavelet Spectrum Analysis for Multifractal Network Traffic

The aggregate traffic of rapid Internet traffic have shown to be multifractal because of the self-similarity behavior at all time scales. In this paper, we study the stochastic properties and construction of multifractal products of processes using wavelet analysis. Following the lead of Fourier spectrum analysis, we define a spectrum function, based on the discrete wavelet analysis, to represent statistic properties of multifractal process. Based on large deviation theorem, the existence proof of the spectrum is given as well as its characterization properties and the corresponding conditions. A simple method of spectrum estimating is given at last