August 17, 2009 – Arris, Inc., having demonstrated end-to-end IP video connectivity over DOCSIS 3.0 at the recent CableLabs summer conference, is telling MSOs they can proceed with delivering heavy concentrations of video content for TV Everywhere-type services without requiring a new version of DOCSIS to overcome perceived cost issues.
The company has gone a long way toward persuading operators that what some industry players are calling “DOCSIS Lite” techniques will obviate any need for a change to new architectural approaches that bypass the CMTS (cable modem termination system), says Arris CTO Ken Wright. “There’s a little bit of mind share out there that DOCSIS is too expensive for delivering IP video,” Wright acknowledges. “In our efforts with larger operators I think we’ve gotten past those doubts.”
The long-debated issue of the suitability of DOCSIS for high volumes of IP video streams is now front and center as MSOs prepare to move ahead with full-scale implementation of TV Everywhere or, as Comcast calls it, On Demand Online. The largest cable company, now in a 5,000-home trial of On Demand Online involving some 23 programming networks, plans to go commercial with the service by fall with no apparent trepidations about the suitability of the DOCSIS 3.0 units it has been using in the field.
But there’s no question some degree of modification in how DOCSIS 3.0 CMTSs operate now will be required to cost effectively support delivery of thousands of long-form program titles over IP unicast streams to tens of thousands of subscribers served by any given CMTS. As previously reported, CableLabs is looking at what changes might be needed in specifications to accommodate these new video initiatives and, by extension, the long-term possibility that all cable content will eventually migrate to IP.
“We’re getting proposals that would move to a next generation of DOCSIS where a lot of the CMTS functionality would be moved to the edge,” says a CableLabs source, speaking on background. “But we’re also hearing from some vendors that they think they can make accommodations to overcome cost issues with some fairly minor tweaks in the existing CMTSs. It’s unclear at this point whether there will be any need for new specifications.”
Wright is confident the industry will stay with the existing mode. “Operators have looked at the DOCSIS bypass approach,” he says. “They’ve said that rather than switch to a brand new platform they’d prefer to go with a field-proven DOCSIS platform that they know scales and works.”
The key to working successfully in the existing centralized CMTS framework, where all video packets flow through the CMTS rather than bypassing it for encapsulation into DOCSIS frames at the network edge, involves exploiting the processing savings associated with distribution of long-form video from a trusted source, Wright says. “What we’re finding going forward is that the use of DOCSIS for delivering trusted content is much less costly,” he asserts. “When you look at what the next generation of the CMTS looks like to accommodate IP video, you find the cost per downstream is drastically reduced.”
This next generation is not a major revolution in design but simply a software-based refinement that ensures the CMTS is not applying unnecessary DOCSIS processes where they’re not needed, Wright says, declining to go into details of what Arris has planned in this regard. “Going forward we can take advantage of those efficiencies in future revs of code and the evolution of the platform in place,” he says. “It’s not a brand new platform. It’s more about a few different factors driving cost reduction.”
One of these has to do with the reduction in processing required by virtue of the long packet lengths in video streams. Tricks employed to compile individual packets within a single DOCSIS frame so as to increase the bit ratio of payload to Media Access Control headers aren’t needed if the payload ratio of individual packets is large relative to the MAC frame.
And if the streamed video file is from a trusted source, such as the cable operator’s TV Everywhere servers, the need for authentication, session set up and other processes is reduced, Wright notes. “If you’re doing a continuous stream from one source there are a lot fewer steps along the way,” he says. “You only need to authenticate once every time the user hits enter.”
It’s not necessary to set up separate processing centers in the CMTS to handle video apart from everything else, Wright adds. “The CMTS has the intelligence to look at the streams packet by packet and to act accordingly,” he says. “So within the same CMTS you can do levels of processing as required without isolating the flows from each other.”
At the CableLabs conference Arris demonstrated how existing technology can be used not only to handle the traffic load but to perform the many other functions associated with TV Everywhere, including dynamic ad placements in targeted advertising applications. While the vendor showed its own products in the end-to-end demo, Wright stresses the open standards approach taken by Arris allows operators to mix different vendor products to accomplish what was shown.
“Our goal wasn’t to show you need to buy everything from us,” he says. “If you’re using brand X VOD servers and our CMTS, the end-to-end solution works just fine. If you’re using our VOD servers and someone else’s CMTS, the same holds true. The demonstration exemplifies our ability to take several industry-leading technologies and integrate them seamlessly into a solution that solves MSOs’ next-generation architectural and advanced advertising requirements.”
The demonstration streamed on-demand IP content and interactive advertising from a single server and a single on-demand/advertising management system, which interfaced with various advanced advertising collaboration partners, including Sigma Systems, Black Arrow and OpenTV. Ads were dynamically placed to fit the profiles of end users as they ordered the movies. In Arris’s case a single server can be used to stream VOD content and to stream the right ads into each piece of content.
The IPTV streams were distributed from the headend using bonded DOCSIS(r) 3.0 channels over a multi-wavelength RF over Glass (RFoG) optical network, terminating on an IP set-top, all monitored with the Arris Service Assurance network surveillance platform. While the RFoG segment showed off the new Arris RFoG optical network unit, Wright made clear the same capabilities could have been shown on a traditional HFC connection.
He also noted the demo served to emphasize the bandwidth efficiencies made possible with use of statistical multiplexing across the four bonded DOCSIS channels. With MPEG-2 video sources from the vendor’s VOD server and pre-multiplexed MPEG-4 movies from a separate source all feeding into the CMTS, Wright says the use of stat muxing in conjunction with variable bit rate transmissions produced a bandwidth savings of 20 to 40 percent in comparison to the bandwidth that would be consumed were all these individual unicast streams delivered in the constant bit rate mode used in today’s VOD operations.
“We’ve seen bandwidth savings of up to 60 percent with more bonded channels,” Wright adds. “When you bond eight versus four channels the efficiencies increase.”
While the demo didn’t include support for transcoding, Wright indicated placement of transcoding units in the network would allow operators to format specific streams for delivery to handhelds and other devices with smaller screens and lower resolution requirements. This would result in lower bandwidth requirements for a TV Everywhere service that targets all devices.
The same efficiencies could be achieved by storing files preformatted for different device form factors. “One of the intriguing things about IP video is it gives you the ability to use same technology to deliver to whatever screens are out there,” Wright says. “Operators might have different resolutions or file sizes depending on which terminal is watching or they might do transcoding on the fly. Whatever they decide to do, this platform is CPE agnostic.”