Estimation of heterogeneous traffic parameters

We consider the problem of parameter estimation of input flow of telecommunication system. It is assumed that the flow is the composition of several independent asymptotically selfsimilar gaussian processes (for example, fractional brownian motions). Very important practical problem is the estimation of Hurst parameters of the components of the process (equivalently, the parameters of long-range dependence). The main idea is the using the set of wavelet coefficients of the process and then applying the regression analysis.