Shaping and policing of fractal /spl alpha/-stable broadband traffic

We study the effect on shaping on self-similar non-Gaussian (Stable) traffic. Although there are several self-similar Stable models in the literature, to the authors knowledge, this is the first time the effectiveness of shaping in the inherent fractality and heavy-tailness of broadband traffic has been examined. The goal of this effort is to increase network operating points dealing with the difficult traffic engineering problems, in a more efficient way. Using simulation and statistical estimation techniques, we show that, (a) the Hurst parameter, which describes the asymptotic self-similarity of the traffic, is unaffected; and (b) traffic can be made less stable, even Gaussian, with the cost of performance degradation.