A Stochastic Differential Equation Model for Queues on Network Video Traffic

Although a lot of study and research have been focused on buffer control and traffic smoothing on server side, which is one of the critical issues in network video traffic area, the problem has not been solved satisfactorily because of lacking of basic theory and the complexity of its internal characteristic. In this paper, we introduce PDSDE (Poisson driven stochastic differential equation) to model the traffic fluid queueing properly and show that ltd item, which was neglected in normal Riemann integral, has great impact on the statistical queueing length both theoretically and experimentally. We construct a well-defined module for real network video traffic environment and give the explicit statistical length of the dynamic buffer precisely, which lays the foundation of studying buffer control and rate smoothing quantitatively for further work.