Performance analysis of self-similar traffic for future service-oriented internet

In this paper, we present a new method to analyze the performance of automatic repeat request (ARQ) schemes in self-similar traffic for the future service-oriented Internet. Taking into account the self-similar traffic of a massive-scale wireless multimedia service, we build a batch arrival queueing model and suppose the batch size is a random variable following Pareto (c, alpha) distribution. Considering the delay in the setting up procedure of a data link, we introduce a setup strategy in this queueing model. Thus a batch arrival Geomxi/G/1 queueing system with setup is built in this paper. We give the formula for performance measures in terms of setup ratio and offered load for different ARQ schemes. Numerical results are given to evaluate the performance of the system and to show influence of the self-similar degree and the delay of the setup procedure on these system performance.