Connection-Level Modeling of Network Traffic
Xin Wang, Shriram Sarvotham, Rudolf H. Riedi, Richard G. Baraniuk · 2002
Aggregate network traffic exhibits strong burstiness and non-Gaussian marginals, which popular models like fractional Gaussian noise (fGn) fail to capture. To better understand the cause of traffic burstiness, we look into connection-level information of traffic trace. A care- ful study reveals that traffic burstiness is directly related to the heterogeneity in connection bandwidths (and round-trip times), and that a small number of high-bandwidth connec- tions are solely responsible for burstiness. This leads to a new model for network traffic: the mixture of fGn and sta- ble Lnoise, which captures both long-range dependence and burstiness. This decomposition has far-reaching impli- cations on network control.