Time Warner Cable claims the technology will provide “virtually unlimited” high-definition capacity. Cablevision Systems is relying on it to have the ability to deliver 500 HD channels by the end of this year.
The magic elixir in question is switched digital video, a powerful tool cable is looking to use to pack in more high-definition content and niche-oriented programming to counter satellite and telco threats.
But unlike ultra-fast cable modems or high-definition TV, switched digital video technology should be totally invisible to subscribers — it’s supposed to look and act exactly like regular broadcast cable channels. The only thing Joe Channel Surfer should know is that there are suddenly a bunch of additional choices on the dial.
“Switched digital will not be widely marketed,” said Bruce Bradley, Motorola’s director of product management for SDV. “Subscribers will see the benefits, but if switched digital video is properly deployed, they shouldn’t even know it’s there.”
The technology works on the theory that not every single channel in a cable system’s subgroups will be watched at any given time. Instead of broadcasting channels to every subscriber, a switched digital video system will send a given channel just to those who choose to view it at a given time.
That lets a cable operator deliver twice as many live TV channels — or even three times or more — over the same amount of bandwidth.
After a handful of trials in the past few years, switched digital video is now on the cusp of becoming widely deployed. But getting from point A to point B with switched digital video requires careful planning and exhaustive testing, as some early movers can attest, like Time Warner’s Austin division (see “Throwing the Switch in Austin,” page 26).
Operators should not “underestimate the operational complexity that is associated with this,” said Harmonic director of cable strategy and business development Tom Kennedy.
“If you need to make modifications in a lab environment, it’s not a big deal,” he said. “But if you roll switched digital video into a large metropolitan area like New York City, you better be darn sure you did your homework.”
Pick the right channels: A month of viewership data should reveal patterns indicating the optimal set. Note that national ratings won’t correspond exactly to local tastes.
Make sure networks are ready: The cable plant should provide reliable two-way capabilities (so set-tops can communicate back to the headend); IP infrastructure may need upgrades.
Design for 100% availability: Subscribers might not fuss over VOD outages, but they expect linear TV channels to always be on.
Monitor channel mix: As new channels get added, the least-watched programming will change.
WHICH TO SWITCH?
Among the first steps in rolling out the technology is determining the set of channels that should go into the switched groups.
As a starting point, most vendors of switched digital video systems provide monitoring and analysis features that will automatically identify the least-watched set of channels.
It’s an exercise in statistical probability. Cablevision vice president of technology Bob Clyne, on a panel last month at the Society of Cable Telecommunications Engineers’ Cable-Tec Expo, described how the operator selected its core group of channels to switch.
Cablevision polled usage data from 4,000 set-top boxes for 24 hours a day for four weeks. Clyne and his team found that over every 24-hour period, the same 92 channels on the network were viewed by no more than 52 of those 4,000 people at any given time.
In addition, Cablevision is using switched digital video — using equipment from BigBand Networks — to deliver about 50 niche-oriented international-language channels, like Channel 1 Russia and the Chinese Channel.
Meanwhile, Comcast, which is kicking the tires on the technology in Denver and New Jersey, has compiled “The Easy 80”: an internal list of channels that may be good candidates for individual systems to switch. (None of the executives interviewed for this article would identify which channels are typically the least-watched ones in a system.)
But it’s critical to do the actual analysis of viewing patterns, according to Joe Matarese, senior vice president of global advanced technology for C-COR. “It’s the old carpenter’s adage: Measure twice, cut once,” he said.
Depending on local demographics, what’s watched in one neighborhood may be widely different from the popular networks across town. Operators are often surprised what they find when they dig into the numbers, said BigBand vice president and general manager of cable video products for North America Biren Sood.
“You think you can guess from a channel lineup which ones should be switched and which should stay in broadcast — but there are always exceptions,” he said.
Even Nielsen Media Research’s TV-viewer ratings “don’t necessarily offer a guide to your particular situation,” Motorola’s Bradley said.
This analysis, by the way, is not a one-time event. Operators must keep tabs on the system to make sure the least-watched channels still make sense to switch. That’s particularly true if you’ve added a slew of new broadcast channels into the switched digital video system.
“If you add another 250 channels, you’ve then changed the balance,” Bradley said. “You have to ask, 'Is the content still niche?’ ”
Next, in designing and configuring the switched-digital-video network, high availability must be paramount, said Tandberg Television VP of system architecture Michael Adams. “You have to really, really worry about the things that go wrong — what if a server fails?” he said.
Adams imagines a telco or direct-broadcast satellite operator capitalizing on a botched SDV rollout: “Those guys would run commercials saying, 'With cable sometimes you get it and sometimes you don’t.’ ”
To BigBand’s Sood, one goal in designing an SDV system should be to minimize the dependency of the switched digital video components on the existing infrastructure, such as network-control systems, to cut the chance of outages. “Switched digital video is linear TV, which subscribers tend to think of as being available 99.999% of the time,” he said. “There’s always going to be some dependency [on existing elements], like the set-top box software, but you want to minimize that.”
Networks may need to be reconfigured or upgraded to support switched digital video, said Scientific Atlanta VP of business development Greg Hardy.
For example, operators need to make sure there’s a reliable reverse path from set-top boxes to the headend, to allow set-tops to send the channel-change requests to the SDV system. “A lot of it is fixing the problems of the reverse plant,” he said, “making sure the upstream is clean.”
Hardy added that some cable systems will need to expand their IP network connectivity to be able to deliver switched-video streams from headends to individual hubs. Operators may have to drop in another Ethernet switch, and they should make sure that there are enough IP addresses available. “These services are going to ride on existing IP networks, which may not have enough bandwidth,” he said. “Some sites are in better shape than others.”
But don’t lose the forest for the trees, Hardy cautioned: “You want to focus on the end deployment, so you don’t fall into a problem where you’ve done a good job on the first hub but the rest still need a lot of work.” Meanwhile, there’s human-related factor for a successful deployment, according to Hardy: making sure everyone’s on the same page. “One of the things we’ve learned is that switched digital video is almost more of a corporate initiative than a site initiative,” he said. “The sites know it’s important — but they’ve got a business to run. So you need to make sure that everybody’s on board and that the lines of communication are clear.”
SIZING THE ROLLOUT
In terms of capacity planning, with video-on-demand rollouts, the rule of thumb has been that 10% of subscribers in a given group will be accessing video-on-demand at any given moment.
For now, there are no similar rules for SDV. In general, according to operators and vendors, a logical starting point is to design the network with a 2-to-1 oversubscription ratio. That means, for example, about 80 of a service group’s least-watched digital channels are allocated to the space that used to be taken up by 40 channels — four quadrature amplitude modulation (QAM) channels.
In fact, some major operators are designing in even more headroom, said Tandberg’s Adams. If the initial analysis suggested they would need six QAMs to serve switched digital video at a 2-to-1 oversubscription ratio, they put in eight.
“In every conversation I’ve ever had with someone in an operational role at a cable operator, when I ask, 'What’s your acceptable probability of blocking?’ the answer is always, 'Zero,’ ” Adams said.
Service groups for switched digital video are typically manageable at between 500 and 1,000 tuners, according to C-COR’s Matarese. His ballpark range is delivering 100 to 150 switched channels over six to nine QAMs — with the caveat that every individual area will have somewhat different characteristics. “You really have to go through and do the modeling to see what it will look like in practice,” he said.
The reason for erring on the side of caution is to make sure there’s virtually never a case in which there’s not enough space for a switched channel.
Of course, the only real way to ensure that is to go back to broadcast. Adams pointed out that telephone systems are engineered to assume about 12% utilization. Only with extremely heavy usage does the telephone network run out of room — in the telephone world, it’s the Mother’s Day scenario that yields “all-circuits-busy” congestion.
What’s the Mother’s Day scenario for switched digital video? Sood said BigBand described it as “the snowstorm”: Everyone is stranded indoors, and they’re all randomly flipping channels.
But in the last few years, even during major weather events, BigBand hasn’t seen the statistical probabilities of original SDV analyses get busted. “The likelihood of a service group tuning into channels that weren’t previously popular — it just doesn’t happen,” Sood said. “We don’t see a large diversity of viewership.”
SHARING AT THE EDGE
Just as switched digital video is a more efficient use of a QAM channel, the next step is to share multiple services over a group of QAMs, gaining even more efficiencies.
Using a system that allows sharing of edge-QAMs, an operator could allocate a set of QAMs for on-demand, SDV and cable-modem services. That way, if subscribers in an area aren’t accessing on-demand video as much at a given time, dedicated QAM channels aren’t just sitting by idly — they can be used for high-speed Internet traffic or switched video.
“QAM sharing is definitely very important, because you’ve got three volatile services — VOD, SDV and high-speed data — for which the traffic patterns could change,” Tandberg’s Adams said. “Operators say, 'We don’t want to spend money on edge-QAMs and have them just dedicated to switched digital video.’ ”
Motorola, Scientific Atlanta and Tandberg Television are among the vendors developing capabilities to manage multiple services for a pool of edge QAMs.
Another player in this nascent area is Camiant, which is introducing the Universal Edge Resource Manager (UERM) for providing dynamic, policy-based QAM resource allocation. The system, to be available by the end of the year, is an extension of Camiant’s core products that provide policy management for Internet Protocol networks.
Camiant chief technology officer Susie Kim Riley said UERM has been tested with major on-demand video platforms — including those from SeaChange International, Concurrent Computer and Tandberg — and the two primary protocols in use with SDV today, developed by Time Warner Cable and Comcast.
“You can’t expect operators to roll out a dedicated set of QAMs for every new application or service,” Riley said. “That’s absolutely not efficient.”