MODELING VIDEO TRAFFIC USING M G/INFINITY INPUT PROCESSES - A COMPROMISE BETWEEN MARKOVIAN AND LRD MODELS/

Statistical evidence suggests that the autocorrelation function rho(ka ppa) (kappa = 0, 1, ...) of a compressed-video sequence is better capt ured by rho(kappa) = e(-beta root kappa) than by rho(kappa) = kappa(-b eta) = e(-beta log kappa) (long-range dependence) or rho(kappa) = e(-b eta kappa) (Markovian). A video model with such a correlation structur e is introduced based on the so-called M/G/infinity input processes. I n essence, the M/G/infinity process is a stationary version of the bus y-server process of a discrete-time M/G/infinity queue. By varying G, many forms of time dependence can be displayed, which makes the class of M/G/infinity input models a good candidate for modeling many types of correlated traffic in computer networks. For video traffic, we deri ve the appropriate G that gives the desired correlation function rho(k appa) = e(-beta root kappa). Though not Markovian, this model is shown to exhibit short-range dependence, Poisson variates of the M/G/infini ty model are appropriately transformed to capture the marginal distrib ution of a video sequence. Using the performance of a real video strea m as a reference, we study via simulations the queueing performance un der three video models: our M/G/infinity model, the fractional ARIMA m odel [9] (which exhibits LRD), and the DAR(1) model (which exhibits a Markovian structure). Our results indicate that only the M/G/infinity model is capable of consistently providing acceptable predictions of t he actual queueing performance. Furthermore, only O(n) computations ar e required to generate an M/G/infinity trace of length n; compared to O(n(2)) for an F-ARIMA trace.