Investigation on the IP Flow Inter-Arrival Time in Large-Scale Network

This paper analyzes the IP flow inter-arrival time distribution and the inter-arrival time autocorrelations of the traces collected in different large-scale networks. It is found that the IP flow inter-arrival time distribution follows the Weibull distribution which has the parameter alpha close to 1.0 as the scale of network becomes larger, so the Weibull distribution degrades to the exponential distribution. It is also discovered that the autocorrelation decreases as the IP flows´ length increases. According to the studies above, this paper deduces the conclusion as following: The IP flow inter-arrival time distribution fits to the Weibull distribution generally, when the IP flow length exceed a certain threshold, the IP flow inter-arrival time fits to the Poisson distribution.