Cable companies -- looking for ways to lower the cost of delivering new interactive entertainment and streamlining the integration of web content into the service mix -- are embracing IP and GigE, which already has gained excellent traction among the telecom set.
Multiple system operators appear to be committed to retaining the last-mile components of their MPEG-based digital TV and DOCSIS-based high-speed data techniques for some time to come. But their need to make more efficient use of bandwidth over fiber and to add flexibility to the way services are shifted across the metro region has inspired a search for solutions that rely on IP and GigE technologies for transport and traffic management between master headends, primary hubs and secondary hubs in the metro architecture. The new thinking came to light at the recent Society of Cable Telecommunications Engineers' Emerging Technologies conference in San Jose, Calif., where a number of speakers representing leading vendors described solutions not previously broached by mainstream suppliers in the cable market.
James Chiddix, president of the new Interactive Personal Video Group at AOL Time Warner Inc., makes clear why new thinking about network architecture is vital to industry efforts to transform customers' viewing options. Speaking of next-generation versions of the industry's high-speed data technology, Chiddix says, "DOCSIS [Data Over Cable System Interface Specification] 1.1 and 2.0 open doors to new services, but I think we need to be thinking beyond that. True broadband services are about two-way audio and video, maybe including TV over IP. Even DOCSIS 2.0 will run out of steam to support our requirements."
The idea isn't to undo what's been done. Chiddix explains the hybrid fiber/coax design and the MPEG and DOCSIS platforms are ideal starting points for migration to an environment where interactive, personalized channel streaming to each user can be accommodated easily on a pay-as-you-go upgrade path that moves fiber termination nodes ever closer to the home. "With our ability to deliver six to seven bits per Hertz, we could deliver 4 to 5gbps in total payload over the fiber to every node, whatever the size of the service area that node covers," he says.
Chiddix says the looming re-entry of telcos into the video market along with new competitive pressures from DBS suppliers offering new personalized services involving caching and replay of programming at the set-top are motivating forces in cable's move to interactive services. While AOL Time Warner has launched video on demand (VOD) services in several major markets, Chiddix says the network topologies employed in these early efforts are too limited to support an environment where every subscriber can access any movie, time-stored program or other application over a dedicated link back to the core servers and switches. "The scale of storage and routing mechanisms and the scale of transport of bits to hubs and nodes are orders of magnitude beyond where we are now," Chiddix says. "We need to be in a position to cost-effectively stream tens or hundreds of thousands of streams simultaneously at peak usage periods."
All of which points to an off-the-shelf IP-based backbone architecture to handle the switching and load balancing requirements. "If it turns out we want to deliver both MPEG and IP (entertainment services), we'd probably want to do it with a great big switch operating in IP mode," Chiddix says. "If we want that kind of infrastructure, maybe we want to build on an IP infrastructure and at the hub employ an (as-yet-undeveloped) DOCSIS 3.0 CMTS (Cable Modem Termination System) along with MPEG QAM (quadrature amplitude modulation) modulators, using MPEG as the transport mode for IP as well as MPEG at the edge."
This is radical new thinking for cable, but it's clear many minds are focused on similar approaches. "The driver for a new solution is the cost per stream, which at today's utilization rate in VOD systems of 5 to 10 percent is commercially viable, but which must be reduced when we move to a high rate of usage for personalized services," says Nimrod Ben-Natan, director of narrowcast service solutions at Harmonic Inc. He adds, the solution involves treating video and IP streams over the backbone "as if the network were all-IP, and, at the edge, go to QAM and MPEG to the set-top."
The transport mode of choice is GigE, where any type of packet-based service can be transmitted from any distribution point in the backbone to any node at the edge. "You can start with point-to-point GigE and migrate to high-end metro GigE switching with switches at every distribution point," Ben-Natan says. "By doing this you're optimizing efficient use of your network capacity and benefiting from the cost advantages of using commoditized equipment."
Yvette Gordon-Kanouff, vice president for interactive technologies at SeaChange International Inc., a leading supplier of VOD technology, and Joachim Vanhoecke, director of technology at Scientific-Atlanta Inc. offer similar visions.
"My topic, IP switched video network for VOD applications, sounds familiar," Vanhoecke told the conference audience. "That means everybody is looking in the same direction."
"I believe this approach is going to have a major impact on our future network architectures," Gordon-Kanouff says. "Everyone is figuring out how to construct the backbone in order to make extreme narrowcast possible."
A representative from AT&T., which is selling its cable properties to Comcast Corp., went even further, suggesting Ethernet could be on a migration path all the way to the home. That would raise the IP data rate per end user without requiring creation of new, higher-capacity channels beyond the traditional 6MHz currently used in HFC networks. Sheryl Woodward, technology consultant for AT&T Labs, suggested this could be done by dividing the FastEthernet 100mbps feed from the outer network into four traditional QAM-modulated cable channels using IP tunneling technology at the edge and at the subscriber premises. That, along with increasing throughput per user, would allow distribution of signals in the home over a 100baseT LAN.
"A quad-FastChannel system can serve 39 percent more users with the same equivalent circuit rate as four DOCSIS channels, or the same quad-FastChannel system can serve the same number of users as DOCSIS with four times the equivalent circuit rate performance," Woodward says, using ECR as a way of characterizing the actual data rates any end user gets over a shared access system. More users are able to tap off the combined channels delivering FastEthernet at 100mbps as a virtual single data stream than is possible if each set of users can only access four separate DOCSIS channels one at a time, Woodward explained.
With such thinking driving future network development in cable, it may be only a matter of time before the unique distinctions that have separated the telecom vendor community from the cable vendor group begin to go away. Commoditization of the cable network along these lines could go a long way toward keeping cable competitive as telcos and other new players move into the mass consumer market for broadband services.
