Statistical Analysis and Simulation Study of Video Teleconference Traffic in

This paper presents results of our study of some of the source modeling and performance issues related to providing video teleconference services over asynchronous transfer mode (ATM) networks. Unlike most previous studies that have used very short sequences, our statistical analysis and simulation study uses a long (30 min) sequence of real video teleconference data. Our major results are: 1) Under certain circumstances, traffic periodicity (due to the constant video frame rate) can cause different sources with identical statistical characteristics to experience cell-loss rates that can differ from each other by several orders of magnitude. For a single-stage multiplexer model, some of this source-periodicity effect can be mitigated by appropriate buffer scheduling and we present one effective scheduling policy. 2) Unlike some previous studies, for the video teleconference sequence that we analyzed (without scene changes orwene cuts and with moderate motion), the number of cells per frame is not normally distributed. Instead, it follows a gamma (or negative binomial) distribution. Also, the number of cells per frame is a stationary stochastic process. 3) For traffic studies, neither an autoregressive model of order 2 nor a two- state Markov chain model is good because they do not model correctly (either underestimate or overestimate) the occurrence of frames with a large number of cells and these frames with a large number of cells are a primary factor in determining cell-loss rates. The order 2 autoregressive model, however, fits the data well in a statistical sense. 4) A multistate Markov chain model that can be derived from three traffic parameters (mean, correla- tion, and variance) is sufficiently accurate for use in traffic studies.