Estimation of loss probability and the admissible number of users in a network link considering a lognormal multifractal traffic model with exponential variance

Abstract

In this work, we propose an analytical expression for estimating the byte loss probability at a single server queue with multifractal traffic arrivals. Initially, we address the theory concerning multifractal processes, especially the Hölder exponents of the multifractal traffic traces. Next we focus our attention on the second order statistics of multifractal traffic processes. More specifically, we assume that an exponential model is adequate for representing the variance of the traffic process under different time scale aggregation. Then, we compare the performance of the proposed approach to other relevant approaches. In addition, based on the results of the analysis, we propose a new admission control strategy that takes into account the multifractal traffic characteristics. We compare the proposed admission control strategy to some other widely used admission control methods. The simulation results show that the proposed loss probability estimation method is accurate, and the proposed admission control strategy is robust and efficient.