Overload control in SIP networks based on statistical analysis of signaling traffic

Latest papers studying traffi c in multi-service networks have shown that pattern of that traffi c is somewhat different from the one we used to see in traditional telephone networks which assumed Poisson as an inter-call arrival distribution process. Nowadays, packet switched networks are better described by a self-similar traffi c models which require new approaches based on long-tailed inter-call arrival distribution processes (Pareto, log-normal, Weibull etc.). Current paper studies SIP traffi c collected on a real VoIP network and shows that it has all features of a self-similar process. This fact is later used to enhance SIP overload control mechanism to eliminate shortcoming of the existing methods.