The ability to access real-time video and even more complicated multimedia data (e.g., realistic, immersive 3D environments) at at any time from any location promises to be a next step in the ongoing impact of the Internet in arenas such as education, electronic commerce, and entertainment. Key problems to be confronted, though, result from the enormous server, network, and client receive bandwidth requirements such media data streams imply. This seminar will cover papers from the recent literature that provide some of the most promising solutions to date for the design of scalable, cost-effective video-on-demand systems. The selected papers cover a variety of topics including the state of the art in server disk scheduling, cpu scheduling for latency-sensitive applications, source (compression) and channel (error correction) coding methods, broadcast and multicast delivery techniques, regional proxy caching, and support for complex objects and interactive services.
Students enrolled in the course will be expected to read the papers prior to the class in which they are covered (one or two papers per week) and to lead the discussion of one of the papers that will be covered during the semester. This is a graduate level seminar. There will be no exams or any other graded material.
All students are welcome. The course has no prerequisites. Everyone should be able to learn a lot from reading and discussing the papers. However, everyone who plans to participate in the class meetings should also enroll in the class.