Analytical load transformations of video streams: Validation using measured traffic

Load characteristics as induced by instances of communicating applications incur strong modifications during the processing of the corresponding requests. In order to increase the accuracy of load models at a given interface, in this paper, we use analytic load transformation to account for such modifications. We apply load transformations on BMAP (Batch Markovian Arrival Process) models to model load as induced by video transmissions over UDP/IP. The obtained models are compared to measured packet traces derived from experiments in a test-bed network. We show that the presented load transformations modelling fragmentation, header generation and Token Bucket regulation are able to model the influence of the corresponding real-world network components with good accuracy. The presented studies demonstrate the validity of the obtained analytic models.