Video's Urge to Converge
IPTV is an issue largely because it's the greatest advance in television since Philo Farnsworth developed his version of the technology. That's the general position staked out by AT&T, anyway.
In its strictest definition, the one we'll use for this article, IPTV is delivering video on an IP network all the way to television receivers (or at least to set-tops). That is essentially what AT&T is doing, as are a number of small, mostly rural providers using technology from companies such as Calix (through its recent acquisition of Optical Solutions).
There are clear advantages to the Internet Protocol and IP equipment. More and more applications are being carried on IP and it is being adopted at an accelerating rate. Its popularity encourages ongoing innovation. All of that contributes to steadily dropping costs.
With IPTV, service providers switch video, which means they can offer a nearly infinite selection of channels and content. They can also transmit to individuals—or at least to individual set-tops in every home, which means they can provide targeted content and targeted advertising.
That all sounds pretty good, so Wall Street and portions of the media—including CED—keep asking cable operators when they are going to finally adopt IPTV technology.
And the answer is, no major U.S. cable operator is going to offer IPTV—as a primary delivery method for video— within the next few years, and perhaps not for decades, for a fairly simple reason: they don't need to. "Many of the features of IPTV—it's scalable, it offers infinite channels, you can do targeted advertising—can be done with switched digital," observes Nimrod Ben-Natan, Harmonic Inc.'s VP of solutions and strategy.
Migrating to IPTV is not necessary for cable operators. In fact, starting with IPTV is not necessary. Verizon's FiOS TV service, despite being delivered via fiber-to-the-premises technology, has more in common with the average cable company's HFC network than with AT&T's IPTV network.
Verizon avoids the IPTV migration questions from Wall Street and the media, however, for two reasons. The concepts of IPTV and video service from telcos have been conflated, even though they are, in Verizon's case, two different things. Verizon has not gone out of its way to dispel the confusion, but will explain to anyone who appreciates the distinction that it has a migration path to IPTV.
FiOS is widely considered to be superior to AT&T's IPTV network, anyway, in large part because FiOS affords Verizon with significantly more bandwidth—hundreds of megabits per second compared to the 20 Mbps to 25 Mbps rates that AT&T will get with its forthcoming "U-verse" service.
And that's the key issue—bandwidth, not the transmission technology. "Being able to match and exceed your competitor's bandwidth is the driver," Yankee Group Director of Media and Entertainment Adi Kishore told listeners of a recent CED Webcast [see story, page 14].
There are several technologies that will buy cable operators more bandwidth for advanced services, and that's where transitioning to switched video comes in. Nor is that the only option; others include migrating to all-digital to reclaim the spectrum now used for analog broadcasting, using DOCSIS channels for video, migrating to DOCSIS 3.0, and moving from MPEG-2 encoding to the more efficient MPEG-4.
Migrating to IPTV remains a viable option; it's just that it's farther out. It makes no sense to abandon HFC networks when they can provide the same advantages IPTV can.
Figure 1: Cable operators not only have the option to switch video, they have two different options for
switching. In the multicast approach, the operator transmits a network channel video stream only
when requested by a viewer. Each subsequent viewer on the node who requests the same
channel shares the stream; the operator thereby conserves bandwidth.
Figure 2: Unicast transmission is the original approach to switching video, the approach most
IPTV systems conform to. Each viewer gets a dedicated stream, even when two or
more viewers request the same content.
That said, the migration toward IP is well in progress, and many expect the transition to IPTV is inevitable, though perhaps far off. Most major cable operators have upgraded to Gigabit Ethernet backbones— by nature an IP environment. They have also widely rolled out IP-based telephony. The industry has gained valuable experience with IP through both efforts.
Individual operators are experimenting with other services that take advantage of IP content or IP transmission paths. In fact, if you relax the definition of IPTV, plenty of operators are already doing it.
Time Warner Cable has an ongoing trial in San Diego in which it is using technology from RealNetworks to deliver video via its DOCSIS channel to subscribers' PCs, rather than to their TV receivers. TWC's Broadband TV is available at no extra charge to customers who take both the video and Road Runner data service.
"To date we have no conclusive findings, but yeah, it's a good way to deliver video," says a TWC spokesman. "With all the bluster from the telcos, people forget we are competing. When the telcos finally get around to selling a video product, people are going to realize they're going after a moving target."
Time Warner Cable has been using switching technology for its video-on-demand service, and recently trialed switched digital video in its systems in Columbia, S.C., and Austin, Texas. The company intends to install the technology throughout its entire footprint.
A key benefit of switching MPEG video instead of IP is that TWC will not have to upgrade currently deployed set-tops. "We can deliver advanced services today without bulk investments," the spokesman says.
Cox Communications also conducted a trial with switched digital, a year and a half ago, and is convinced of the benefits.
Comcast data customers can access a wide range of video clips through Comcast's customer portal and a feature called "The Fan." Is that IPTV? Comcast CTO Dave Fellows says video delivered on an IP (DOCSIS) transmission channel most certainly can be.
The issue, again, is not whether or not it's technically IPTV, but whether the operator is making efficient use of its bandwidth.
Comcast's bag of bandwidth management tools includes using its DOCSIS channel for signaling associated with its VOD service. Comcast has been using that approach for three years, Fellows says. Other MSOs do likewise.
The issue does not have to be whether to switch video or not, but how much video to switch. Comcast interprets VOD as switched video. The key is figuring out how much bandwidth to dedicate to VOD.
VOD could be made to look exactly like any other video. All Comcast (or any other operator) would have to do is update its program guide to present an on-demand stream as a channel. The content could be a network channel or it could be a playlist that looks just like a channel. Unless the operator were to charge differently for the on-demand stream, the user would see no difference. Again, it's an issue of bandwidth management
"Three or four years ago, we decided to go 10 percent switched for VOD, and 90 percent broadcast," Fellows says. "We may, going forward, decide to go 20/80, or 40/60. We can always reset that."
Not everyone is enamored of using the DOCSIS pipe for shipping video (and other IP-based applications), but there are several developments that will make it more attractive.
One is a move to a newer version of IP addressing, IPv6, which will be supported by DOCSIS 3.0. Comcast, like many other companies, is now using IPv4, which allows for 16 million IP addresses. "We have more than that in set-tops alone," Fellows notes.
IPv6 will increase the number of addresses available to a virtually infinite number, which will allow Comcast to identify and provide personalized services for as many individual addressable devices as its subscribers might have in total.
Next, with the introduction of modular CMTSs, QAMs can be used to handle both data and video, lowering the cost of shipping video.
Finally, channel bonding will increase the bandwidth available via DOCSIS, and allow operators to ship more video through those channels.
Will all that bonded DOCSIS bandwidth be used for IPTV? It could, depending on how operators elect to use it. "I'm just the broadband plumber. Management decides what to do with the pipes," Fellows quips.
"It's all about gradual evolution," says Seth Kenvin, BigBand Networks' vice president of corporate development. BigBand recently introduced version 4.5 of its switched broadcast product, now being used in several switched broadcast trials. "Cable is already evolving in the IPTV direction. It'll get there, but it might take years—decades even, but it will get there."
Cable operators "have got broadcast. That's a simple, extremely effective mechanism. Maybe there's no reason to switch NBC," says Kip Compton, Cisco Systems Inc.'s senior director of video/IPTV development.
Unless there's a specific new application that shifting to IPTV enables, though, the cable industry will approach the issue without urgency, which is appropriate, according to Kenvin.
"There's not a single service operators deliver that would require a migration to IPTV," adds Harmonic's Ben-Natan.
Not only is there little advantage for MSOs in migrating directly to strict-definition IPTV, but there would also be a penalty, as operators would have to update millions of legacy set-tops.
On the set-top front, chip companies are helping to clear a migration path to IPTV with a new generation of ICs that can handle both MPEG and IP, and can be used to support a new breed of hybrid MPEG/IP set-tops.
There are two possibilities for such boxes, which will by necessity occupy the mid-range and high-end. The first is to encode video in the IP format and handle it as IP all the way to the TV receiver. The other is for the operator to send MPEG video with the MPEG packets encapsulated in IP packets; the processor in the set-top will strip out the MPEG packets, and then process the MPEG.
Broadcom has a new chip, the BMC3255, that operates in channel bonding mode, combining up to three DOCSIS channels and providing over 100 Mbps of throughput.
The cost of IP networking equipment is continuously dropping, it's true, but because cable operators already have fully functional headends, migrating to IP might not save much, notes Peter Schenitzki, product marketing manager for Broadcom Corp.'s cable products division. It's more a technology for greenfields.
"You don't do IP for the sake of IP," says Aiden O'Rourke, Broadcom's director of marketing, IPTV set-top boxes. "It has to be something it gives you in terms of applications that you didn't have before. Okay, maybe somewhere down the road, it'll save operators some cost, but what's the customers' incentive? We've seen some miserable failures in Europe because customers never saw any value-add."
"Think of how the user can use media that he can put in a box and then put it in a pocket. IP is a key piece of that. I can see IPTV being useful for where the rights follow you regardless of the platform. Right now, DRM is unsettled," O'Rourke says, referring to digital rights management.
Intelsat recently introduced a new service called Ampiage. Ampiage is similar to a superheadend that acquires, and aggregates, licensed TV programming. Video can be encoded as MPEG-2, MPEG-4 or MPEG encapsulated in IP—a capability useful for IPTV.
The service is available to MSOs, but as another indicator of MSO diffidence when it comes to IPTV, Intelsat expects most Ampiage customers at first to be telcos using MPEG encapsulated in IP. Later, as more cable operators migrate to MPEG-4, Ampiage expects more MSOs to sign on.
Ken Tagaki, Intelsat's director of strategy and marketing, cites various analysts' projections that the telcos will end up with only single-digit millions of subscribers in four or five years. "But a few million—that's still a good chunk. It will cause the cable guys to respond." Still, he says, "The telcos are starting from zero. They won't be equal competitors for some time."
"It's not unheard of for cable companies to go off in different directions, but when they do that, they usually converge later. Now is a good time for experimentation," says Cisco's Compton. "That's good. The industry's net experience base will grow."