Measurement and statistical analysis of asymmetric multipoint videoconference traffic in IP networks

The paper contributes results on the modeling of videoconferencing traffic over IP networks. The study is based on extensive data, gathered by tracing the actual packet exchange during a comprehensive set of realistic teleconferencing sessions over an asymmetric platform, in which commercial H.261-compliant terminal clients were communicating, through a Multipoint Control Unit (MCU) at 'continuous presence' mode. Analysis of the data suggests that the video traffic from the client terminals can always be represented at the frame level as a stationary stochastic process with an autocorrelation function of exponentially fast decay and a marginal frame size distribution of approximately Gamma form. The video traffic from the MCU to the clients is again stationary and with exponentially decaying correlations, while the corresponding marginal frame-size PDF has the form of an appropriately weighted sum of Gamma components, the number of terms in the sum always being equal to the number of conferring terminals. The paper discusses methods for correctly matching the parameters of the modeling components to the data and for combining these components into complete traffic models that have been proposed in the literature.