Stochastic modeling and performance analysis of video streaming system over IP networks

摘要

In this paper, we develop a novel stochastic model, i.e., an extended two-dimensional stochastic fluid model (2D-SFM), to describe the video streaming system over the IP network dynamics. We first derive two Laplace-Stieltjes transform (LST) matrices of the 2D-SFM, based on which two key performance metrics of the video streaming system, i.e., the video freeze probability and the moment of the buffering delay are derived, in particular, the expectation and the variance of the buffering delay are obtained. As an application, we develop an algorithm to find the optimal initial buffering level with the constraint that the freeze probability is below a given tolerable probability. Finally, numerical results and simulations are provided to corroborate the theoretical findings, and the effects of the system parameters in different scenarios on system performance metrics are further studied numerically.

关键词