The primary goal of this research is to investigate resource management and traffic-related issues that will enable efficient transport of compressed video streams over an ATM network. These issues include the determination of network resources to be allocated for a given traffic stream, the prioritization of compressed video streams, the design and management of scheduling techniques that the network can implement to provide differential service to various priority levels, and the effect of resource sharing on the effective bandwidth and buffer size allocated to a stream. In order to study the aforementioned issues, an accurate characterization of compressed video traffic is required. Accordingly, we will first capture, digitize, and compress (via compression standards) several video streams (e.g., movies), whereby the picture dynamics (in terms of scene changes, panning, zooming) varies from one movie to another. These captured traces will be used to develop traffic models for single and aggregate (statistically multiplexed) compressed video streams. The traffic models developed will extend beyond the entry node of an ATM network to include compressed streams at intermediate nodes inside the ATM network. Performance studies (queueing-based) will be done of interfering traffic mixed with compressed streams (MPEG-2). The ATM testbed at Michigan State University will be used for the proposed research. Refer to our URL address (www.cps.msu.edu//~hughes/mpeg-2.html) for additional information about this project.
