Fast simulation for self-similar trac in ATM networks

ABSTRACT Recently self-similar (or fractal) stochastic pro-cesses were proposed as more accurate models of certain cat-egories of trac (e.g., Ethernet trac, variable-bit-rate video) which will be transported in ATM networks. Existing analytical results for the tail distribution of the wait-ing time in a single server queue based on Fractional Gaussian Noise and large deviation theory, are valid under a steady-state regime and for asymptotically large buer size. However, pre-dicted performance based on steady-state regimes may be overly pessimistic for practical applications. Theoretical approaches to obtain transient queueing behavior and queueing distributions for small buer size become quickly intractable. The approach we followed in this paper was based on fast simulation techniques for the study of certain rare vents such as cell losses with very small probability of occurrence. Our simu-lation experiments provide insight on transient behavior that is not possible to predict using current analytical results. Finally, they show good agreement with existing results when approach-ing steady-state. I