Prediction of Variable Bit Rate Tra c Using Earthquake

In this paper, we use prediction algorithms to estimate the behavior of VBR video traac based on statistical short-term earthquake prediction methods. The probabilistic estimation of future inter-arrival times of peak values of MPEG-VBR video stream is considered in terms of Bayes theorem, and the Bayesian analysis of peak rate sample is evaluated on the basis of the proportion of successes rather than in the number of successes. The estimation results of aftershock peak values are satisfactory. The distribution of VBR video data magnitudes based on maximum entropy distribution method will also be investigated. As an application , the method will be implemented to adapt the leaky bucket rate in order to police VBR video streams in ATM networks without using very high bucket sizes.