Earlier this week at NAB, Verizon unveiled a multi-screen video solution dubbed "Verizon Digital Media Services" (VDMS) to deliver digital media to virtually any device or platform on a large scale. (For the announcement, click here).
Verizon said its VDMS ultimately will enable high-quality and reliable “unicast" individualized video streams. But the carrier’s talk of unicast video streams raises this question for cable operators: What’s the best IP-video delivery method?
According to Dan Dodson, Fellow at Interactive Broadband Consulting Group (IBB), "For on-demand content, the technology is pretty well advanced. Most everybody is using some form of adaptive, unicast delivery method – Flash or Microsoft. But that starts to break down with linear video; the capability to do a multicast of an adaptive video stream doesn’t exist too far out of the science lab."
Dodson delivered a paper – "IP Video Delivery is Ready for Prime Time” – at the SCTE Canadian Summit 2011 in March, which hypothesizes an architecture that combines IP unicast and IP multicast. (For more from the SCTE Summit, click here).
IBB’s proposed IP architecture prepares the video for adaptive streaming with a three-step process. First, the video is encoded or transcoded, then the resultant transport streams are fragmented, and finally the fragments are placed into the Content Delivery Network (CDN).
The fragmentation process is key to the architecture. It involves passing each transport stream through a fragmenter that reads the stream and divides it into a series of two-second or three-second MPEG-4 fragments with multiple wrappers for the various adaptive streaming formats like Microsoft Smooth Streaming or Adobe Dynamic Streaming.
It doesn’t matter if the fragments are destined for linear video or for on-demand video.
The fragmented content available in the CDN can be used to transport unicast video streams for VoD content, and they also can be multicast for linear video, which improves bandwidth efficiency.
"With multicast, a single source sends data to multiple destinations at one time," according to IBB’s paper. "Each TV channel has its own multicast. If the multicast is not already joined in a service group, then the first join from any subscriber in that service group starts it, similar to how SDV works."
"With any kind of penetration below 10 percent, for a cable operator or a telco, they can use unicast," adds Dodson. "But if they’re going to enable streaming video in a big way, they need to have it be multicast. The right answer is: Let’s make multicast adaptive technology work."