Estimation of Hurst parameter by variance-time plots

Related research showed that traditional mathematical models have failed to simulate the traffic on the information highway. In contrast to traditional models, the traffic is self-similar or fractal inherently. Hurst parameter is one very important parameter to capture this behavior. Tianjin University is currently collaborating with U. of Ottawa in the research of self-similarity, This paper summarizes the work done using variance-time plots to estimate the Hurst parameter of real network traffic. Apart from the ordinary conclusion that the network traffic is self-similar, we also made some interesting observations of multi-fractal behaviour and the value of Hurst parameter when traffic is merged. WE noted that variance-time plot are not reliable for empirical records with small sample sizes.