APPROXIMATION OF VIDEO CELL TRAFFIC BY AR(1)+IPP MODEL

For a network design and a cell traffic control in an asynchronous tra nsfer mode (ATM) system, it is significant to construct a stochastic m odel to represent bursty cell streams. As models of bursty cell stream s, a Markov modulated Poisson process (MMPP) model and an autoregressi ve process of order 1 (AR(1)) model have been studied. The MMPP is a p rocess which is based on the Poisson processes, so the MMPP allows the exact queueing theoretic analysis to provide some performance measure s. However, it is not capable of expressing the autocorrelation of the cell stream directly. Moreover, the MMPP requires considerable calcul ation time to achieve the exact analysis. The AR(1) process has advant ages of a simple form and the capability of directly expressing the au tocorrelation of the cell stream. However, it cannot follow a rapid ch ange of the cell stream. This paper adopts the AR(1) as a model of the basic cell stream and combines the interrupted Poisson process (IPP) to follow a rapid change. The AR(1) + IPP is proposed to model a video cell traffic, which is the most important source in ATM networks. By applying an approximation technique, the cell loss probability of the system is analyzed. With respect to the accuracy and the calculation t ime, the proposed model is compared with other conventional models to confirm that the proposed model is superior to the others.