The NCTA technical session "Technical Screaming Streaming" on Monday, May 8, took a deeper look at the changing definition of convergence and how cable operators will integrate interactive content and IP content into their offerings.
"Where convergence used to mean telephony and video, it now means IP and video and how the cable infrastructures and technologies are going to handle them globally," says Paul Harr, vice president, marketing, for Wegener Communications. "It's not just a small aspect of a particular architecture that has to change."
Harr, who spoke on the issue of IP data over satellite to cable headends, believes that the evolution to a feature-rich multimedia experience will outgrow the "unicast" infrastructures that send the same information to all customers.
"The model is going to require the storing and forwarding of content to the edge servers," he says, "because the costs of disk-based servers is less than creating a backbone that can handle the demands.
"The unicast model doesn't work with high-bandwidth content and increased demand for that content," Harr added. "What makes most sense is taking that content closer to the edge of the network. And that model follows for video-on-demand as well as IP."
Dr. Bill Wall, Scientific-Atlanta technical director, subscriber networks, added that, for all the talk of technology, it will be services, not set-tops, that will be bought by consumers. And these services will define the capabilities of tomorrow's set-top boxes. Wall said set-tops will meet a number of demands.
"They'll support both analog and digital video and will also be an Internet appliance," he explained. "They'll also be a communications device for voice and video messaging, and will be a home gateway to the home network."
Dr. Mukta Kar, CableLabs senior member of the technical staff, addressed the issue of whether streaming should be accomplished with MPEG-2, IP or both. His conclusion? "Both will co-exist for a while until one precludes the other," he said. "IP is preferable for subscriber-to-subscriber demands, because it has a variable bit rate and is optimized for non-real-time delivery. In terms of broadcast demands, MPEG-2 is the best because of its fixed packet size and that it's optimized for the synchronization of video."
Dr. Kar also mentioned RTP, but added that it has a large overhead.
Chia Li, Lucent Technologies technical manager, cable networks planning and model, says there will be no end to the demand for bandwidth and applications. The cable plant is evolving to a two-way, fiber-deep infrastructure, but quality will be heavily dependent on delay, jitter and packet loss. "Those factors are not independent, and they shouldn't be treated separately," he said.
Another factor in flexibility in future plants could be the use of MPEG-4 compression, according to Jan van der Meer, Philips consumer electronics technology manager. Van der Meer said MPEG-4 is suitable for streaming and is not a replacement for MPEG-2. "It's object-oriented, and that allows for sophisticated multimedia applications."
That's something many of today's interactive services lack. Cable services are primarily focused on video-on-demand, but there are companies like Wink that are engaged in offering icons that the user can click to get more information or buy products.
Wall added that video will increasingly be distributed on demand, some of it over IP and some of it over MPEG-2. There will also be the ability to play video games through the set-top box. "The bandwidth explosion is being driven in a couple of directions," he said.
But Van der Meer says that MPEG-4, which shares some commonality with MPEG-2, offers high coding efficiencies that could allow for object-oriented interactive services to be integrated into programming globally delivered over MPEG-2 or via IP for personal delivery.