A Truncated alpha-Stable Long-Range Dependent Process Model for Modeling Network Traffic

Abstract

The burstiness and long-range dependence constitutes important characteristics of high-speed communication network traffic, which have been extensively studied in the past decade. Though several existing models can account for the long-range dependence, the burstiness has not been captured as well. In this paper, we propose a truncated α-stable longrange dependent process model (TSLRDP), which incorporates the long-range dependence and burstiness precisely into one model. Furthermore, it demonstrates considerable flexibility in adapting real traffic. A novel algorithmic procedure to estimate the model parameters from real traffic is presented. Simulation results show that the proposed TSLRDP model fits the real traffic accurately.