Fitting stochastic model into network traffic

This paper presents the results of fitting stochastic model into real network traffic. Accurate modeling of network traffic is the first step in optimizing resource allocation and Quality of Service requirements. Because measurements reveals presence of self-similarity and long-range dependence, unlike the models based on Poisson or Markov processes, fractional stochastic model seems to be a good approximation of network traffic, since it can capture both short-range and long-range dependence. A methods of generation as well as the model order selection and parameter estimation techniques will be presented and discussed.