Impact Of Self-Similar Input Processes In The Initial Transient Period In Communication Networks

Self-similar stochastic processes with long-range dependence occur in many natural and man-made systems. In particular, they have become the subject of numerous research investigations of their nature and consequences since such processes were discovered in the Internet and other multimedia telecommunication networks a decade ago. This paper shows that, assuming self-similar inter-event processes, many more numbers of mean observations are required in the initial transient period, as H increases, than when assuming Poisson models, exhibiting short-range dependence. Two queueing systems are simulated in a fully automated simulation package, Akaroa-2 in methods of sequential simulation output data analysis of mean values based on spectral analysis and batch means.