Wavelet packet based Hurst parameter estimation

There are several methods to estimate parameter H, which measure the degree of self-similar of network traffic, but researches have indicated that they have limitation either on accuracy or on computation. The advantages of wavelet analysis based method inspire us that we proposed a new estimating method based on the discrete wavelet packet transform (DWPT). In this paper, we introduce the DWPT and infer the power law between the energy distribution of decomposed coefficients and the index. In order to verify the accuracy and applicability of DWPT based method, we apply it to data from two kinds of scenario: data with known H and real Internet data. We also discuss the parameter H of abnormal traffic influenced by different Internet worms.