A.S. Bugaev, O.I. Sheluhin

Results of numerous researches show that measurements of queuing performance of a fractal traffic can essentially differ that are predicted by appropriate systems with traditional traffic patterns. The paper shows that the characteristic of queue distribution in case of multifractal traffic input is completely characterized by the scaling function and scaling coefficient of input traffic. The paper proposed a method for estimating these parameters on the results of statistical analysis of telecommunications traffic samples. Analytical relations and algorithms for the numerical evaluation of multifractal traffic charasteristics are presented. Analytical relations for estimating the queue tail probability distribution of multifractal traffic of an arbitrary form at the entrance. From these characteristics it is concluded that multifractal nature of traffic at the input buffer device has a significant influence on the characteristics of queuing. The greatest influence is caused by a component of the multifractal traffic with moment coefficient of two and more.