Two-level stochastic fluid tandem queuing model for burst impact analysis

Queuing analysis is important in providing guiding principles for packet network analysis. Stochastic fluid queueing models have been widely used as burst scale models for high speed communication networks. In this paper, we propose a novel two-level Markov on-off source model to model the burstiness of a packet stream at different time scales. Analytical results are obtained to reveal the impact of traffic burstiness at two levels on the queue lengths in a tandem queue system. Our method combines the modeling power of the Poisson processes with that of stochastic differential equations to handle the complex interactions between the packet arrivals and the queue content. Our results for the tandem queuing network could be used to further justify the packet spacing scheme in helping deploying small buffer routers.