Traffic and queue fractality in queuing systems

Traffic in modern multiservice networks is usually correlated, therefore, the methods of classic theory do not work properly. In this paper the results of simulation and comparative analysis of traffic correlation influence for a fractal requests ceiling and requests flow in a case of real multiservice telecommunication networks are given. We examine Pollaczek-Khinchine formula generalization for an average queue length of stationary requests flow with a random correlation as well as opportunities for its application. It’s shown that a real random value of requests number that is proceed at the interval of one request service has an expectation value lesser than 1. It is shown that a process of queueing has passive time intervals that interrupt a correlation link with previous and following process parts. It is shown that correlation dependencies between separate queues values spread only at intervals of employment of queuing network for self-similar processes that have an infinitely big correlation interval.