A Novel Discrete Distribution and Process to Model Self-Similar Traffic

Network traffic is modeled as Poisson process for analytic simplicity, even though a number of traffic studies have shown that packet interarrivals are not exponentially distributed. Packet arrivals deviate considerably from Poissonity and follow statistical self-similar and long-range dependent behavior (LRD). We propose a novel discrete distribution for the counting process and show that for such a process the interarrivals are Pareto distributed. This new process is self-similar and exhibits long-range dependent behavior.