Validation of Gaussian Traffic Modeling using Standard IP Measurement

Abstract:
In the past decade, the characteristics of Internet traffic has been in-vestigated by many measurement and modeling approaches. Most of them are focused on the variability of traffic over time and its autocorrelation function. On the other hand, Gaussian traffic models can be used to bound the probability of overload on links and other network resources, which only take the station-ary distribution of the traffic rate into account. This paper compares properties of Gaussian traffic modeling with standard measurement obtained from routers in IP backbone networks and summarizes the conclusions for network dimen-sioning.