Can QOS Guarantees be Supported for Live Video Over ATM Networks?

Emerging B-ISDN networks which use asynchronous transfer mode (ATM) as a backbone network, need to provide quality of service (QoS) to various multimedia traffic classes such as live video, high resolution images, video on demand, etc. Since live video is expected to dominate multimedia traffic, guaranteeing QoS for live video is complex because of no a priori knowledge of the video stream and hence requires a good understanding of such video sources. This paper investigates a complete framework for transporting live video over ATM networks with QoS guarantees. The proposed framework comprises of a source model, call admission controller and a scheduler. The source model uses entropy as a measure to characterize the live video traffic. A Markov chain model is developed based on this measure. The source model aids in reserving resources for the admitted sessions. The algorithm developed for call admission guarantees a minimum bandwidth throughout the duration of the session. Two new scheduling algorithms viz guaranteed bandwidth schedule and guaranteed delay schedule optimize the bandwidth utilization of all active sessions. Using this framework for transporting live video a cell loss probability of $4.87\times10^{-7}$ and a best effort bandwidth utilization of 80% is obtained