UW's "Internet HDTV" project provides a foundation for answering the question "How would the world change if you could send studio-quality HDTV over a general purpose Internet?" More fundamentally, it is intended to explore the intersection of network, video, and server technologies where near real-time distribution of extremely high-quality images is required.
For some time, UW's office of Computing & Communications has been involved in developing and integrating technology and techniques for the distribution of high-quality video over the Internet. These efforts have been in support of the UW-led ResearchChannel consortium. Over time, however, the definition of "high-quality", and the corresponding requirements for network capacity, have dramatically increased.
Early in 1999 we began looking for ways we could transmit HDTV video over the high-capacity Internet2 "Abilene" network. With generous support from Sony Electronics, these efforts culminated in a very successful demonstration at the Fall 1999 Internet2 members' meeting in Seattle, wherein two HDTV streams were sent simultaneously from Stanford University to the University of Washington, one of them at 40 Mbps and the other at 200+ Mbps. Both experiments relied on software developed by UW/C&C engineers.
Building on the I2 meeting success, the next challenge was to send five concurrent 200+ Mbps HDTV streams. The Supercomputing '99 conference provided the venue for this ambitious demonstration. In early 2000, the boundaries of research networks were pushed one step further with a cross-country five stream demonstration across a DARPA-funded Supernet. At NAB2000, the convergence of Broadcast and the Internet was demonstrated when multiple streams from KING5 Television in Seattle were streamed to the convention in Las Vegas, switched on the show floor, and streamed back to Seattle to be broadcast directly off the Internet.
One hallmark of this work is the unusual coincidence of experts in networking, computer systems, and video. The project draws talent from nearly every C&C division. Another key aspect is the reliance on industry standards, both for networking and video. This means that the results are more widely applicable than might otherwise be the case, and it also means that members of the broadcast television industry can easily see direct implications for their future.
UW's Internet HDTV project is pushing the envelope on several fronts. Multiple system integration challenges, from PC performance problems to Gigabit Ethernet incompatibilities, have been encountered. As the project continues, we hope to gain insights that may help guide current network QoS debates, questions of quality vs. latency, limitations of current PC architectures, etc. We also hope to explore other applications besides HD television (e.g. collaboration, telemedicine, interactive visualization) which may be able to benefit from Internet HDTV technology.