A new model for VBR videoconferencing traffic in ATM networks

This paper studies VBR videoconferencing traffic in ATM networks. A first-order autoregressive model with residuals of Rayleigh distribution is proposed instead of Gaussian noise. It is easy to generate and practical. The results show that our model can generate many frames of large cells. This is important for estimating cell loss rate. Moreover, we assume that VBR videoconferencing traffic is distributed with a Rayleigh distribution. The numerical results for statistical multiplexing show that it is also a better fit to real data under many sources of statistical multiplexing. For this situation, we point out that it is the best compared with the Gaussian and gamma distributions currently used.