Regression Video Traffic Models in Broadband Networks

Recently, there has been significant progress towards the deployment of broadband integrated services digital networks capable of flexibly supporting digital video technologies. Digital video such as video on demand, video teleconferencing, video telephony, and broadcast HDTV will constitute a major traffic component of these networks. Statistical analysis and performance modeling of various types of video traffic are required to estimate network resources and predict the behavior of network under various conditions. For over a decade, there has been an enormous amount of interest and research in traffic modeling of compressed variable-bit-rate (VBR) video. An important class of video models that has received much attention lately is regression models. This paper presents a survey of state of the art of regression traffic studies that have been proposed in the literature to model a variety of video applications. The video models will be classified into two classes: teleconference video, and full motion video. In each class, we will highlight the main features, describe the underlying model, and examine the advantages and limitations. Where appropriate, we also show some results of the statistical properties and the ability of the model to predict the queuing performance of a single and multiplexed video sequence