9357685 Rangan Technological advances in hardware are revolutionizing computers and networks so as to support digital multimedia. By effectively combining the programmability of computers with the pervasiveness of mass-market information networks, these advances promise to herald a new epoch in media communication. In this research, we focus on the following areas arising out of the synergies between computing and media communication: (1) Multimedia on-demand information servers that can support a wide variety of services, ranging from video on-demand for home entertainment to distance learning and digital libraries for education; (2) Media synchronization techniques, communication architectures and admission control strategies for supporting high-performance multimedia applications efficiently over future integrated networks; and (3) Emerging paradigms and software architectures for structured multimedia tele-collaborations. For designing interactive multimedia on-demand information services, we are developing a constrained block allocation model that guarantees continuous media retrieval, and algorithms to service an optimal number of user simultaneously without violating the real-time retrieval, rates of any of the users. Multimedia services generally involve playback of multiple media strands (continuously recorded video frames of audio samples) over high- speed networks onto user display sites (e.g., HDTVs). We propose to develop adaptive, rate-based techniques for synchronous playback of multimedia strands at display sites over networks. In high-performance multimedia applications (e.g., Tele- orchestra), media data form all the user must be continuously mixed together and displayed synchronously. We are proposing optimal hierarchical mixing architectures for supporting scalable tele-collaborations and conferences with minimum real time end- to-end delays and are developing a Packet train protocol in which routing nodes themselves of a network can b e used as mixers, thereby integrating mixing with routing in networks such as the Internet. We expect to experimentally demonstrate these ideas at the Multimedia Laboratory at UCSD. Our experience will prove valuable in designing multimedia services for future integrated networks in this rapidly approaching age of tele-personal interaction.
