Periodic Data Traffic Modeling and Predicition-Based Bandwith Allocation

For the purpose of provisioning bandwidth for Internet access, we need to model the traffic at large time scales, over which the traffic shows evident periodicity, long correlation and a non-Gaussian marginal distribution. To capture these characteristics simultaneously, in this paper we use a periodicity transform-to identify the most significant periods of the traffic and use an autoregressive time series to capture the autocorrelation and apply the G-and-H distribution to model the marginal distribution. A prediction-based bandwidth provisioning scheme is proposed and many experimental results on real Internet traces are also provided