IPTV Strategy

As befits an emerging technology, Internet protocol TV (IPTV) is hard to nail down. The total estimated number of international subscribers is a matter of educated guesses. Standard definitions of terms are missing.

Some analysts, for instance, exclude Verizon’s million-plus video subs on sound technical grounds. But others include the deep-pocketed telco, whose aggressive cable overbuild may yet shift into an all-IP delivery in a few years.

Not surprisingly, given the confusion surrounding basic terms and data, projections of global growth vary, too. In February, FierceIPTV‘s Tom Burton underscored the differences between the following analyst shops’ predictions of how many IPTV subscribers would populate the world by 2011:

• RNCOS, 103 million
• Research and Markets, 80 million
• MRG Group, 72.6 million
• Informa, 38.4 million

Whether IPTV grows by a factor of four or 10 by 2011, however, the clear upward trend spells opportunity for some and threat for others. Where does cable fit into this mix? Definitions A lot depends on definitions, especially where consumer demand and MSO marketing are concerned. As a slogan, "I want my IPTV" lacks a certain amount of precision, for instance.

"It’s like saying, ‘I want to drive a motor vehicle,’" said Rich Prodan, vice president and CTO of Broadcom’s Broadband Communications Group. "Well, OK. Is it yours? Do you want to buy it or lease it? Is it a truck? Are you looking for a race car, a street car?"

More directly, Prodan said MSOs have three IP video options: stream and deliver bulk, controlled end-to-end MPEG encapsulated video; deliver bulk transfers; or simply stream. The first option aligns with IPTV as understood in terms of a TV experience; the other two correspond respectively with current over-the-top, transfer-and-save services from AppleTV or NetFlix; and YouTube, or Comcast’s own Fan (with an added measure of quality of service, QoS).

MSO executives are quick to note early convergence of DOCSIS and video technologies. "We are purchasing set-tops with modems built in," said Mike Hayashi in an interview with CT late last year. He pointed to "caller ID on TV" as a "great example" of bridging the PC and TV worlds. Coming at convergence from the other direction, Time Warner also has trialed TV programming delivery to the PC. (See "Device Shifting: Lessons from TWC’s Broadband Television Trial," CT February 2006.)

Time Warner is not alone. The residential network gateway (RNG) set-top family specified and outlined by Comcast executives a year ago houses a cable modem. Charter Communications moved even earlier in this direction, having deployed for several years a Digeo set-top box that uses DOCSIS for most of its communications.

None of this, however, translates into IPTV as a scaled and widely deployed TV service. "Most of the operators that I speak with (in the United States, Europe and Asia) are still trying to determine what role IPTV has in their future," said Mike Cookish, director of product management at Motorola’s connected home solutions division.

"These operators have a huge installation and legacy of MPEG video that’s working for them today," he said. "My gut sense is that it’s not today or tomorrow." Content and speed What is happening now is on the high-speed data side of the aisle, namely the gradual shifting into a DOCSIS 3.0 framework for delivering higher throughputs. Linked to that migration, in some early instances, has been a need for video to demonstrate the capabilities of such speeds.

The challenge that faced UPC Vice President for Design and Architecture Mourad Veeneman during a pre-DOCSIS 3.0 field trial in 2005 was not so much reclamation of spectrum for the bonded channels, but rather "finding applications and content that participants could utilize to take advantage of the dramatic data rates."

"We ended up with some high-definition (HD) video content obtained from our sister company Chellomedia and used it to stream to the trial users," Veeneman explained to CT late last year.

A similar situation faced Mike Reynolds, president of StarHub in Singapore, which launched a 100 Mpbs channel-bonded data service in late 2006. "Clearly, with higher bandwidth speeds available, there is also a need to ensure that customers have content in which to use the available bandwidth," he said.

One of the answers to StarHub’s situation was to partner with Limelight Networks, an international content distribution network (CDN) that offers service providers a chance to steam video with greater control.

"We have an optical network that’s globally spread, and we peer directly to the telcos and MSOs," said Adam Wray, Limelight Networks vice president, strategic alliances. "This is different from the traditional approach of CDNs … where they’ve deployed disparate boxes across many different networks and just bought bandwidth as needed."

In a similar vein, Level 3 announced in April an expanded CDN to meet demand from customers with large libraries of digital assets; and on the content side, Starz Entertainment launched Starz Play, a variant on its direct-to-consumer service that offers a greater role for distribution partners, such as MSOs. Bonded and bypassed Channel bonding opens up a range of options. In a keynote address at the opening session of the IPCable track at the IPTV World Forum in February, Cisco Systems Director of Cable Video Architectures S.V. Vasudevan focused on the potential efficiencies of a variable bit rate (VBR)-friendly, channel-bonded future.

In a 24-channel bonded framework, which he said was doable by 2015, MSOs could leverage VBR to deliver a Gigabit to the home, realizing 30 to 40 percent efficiencies over alternative constant bit rate (CBR) encoding formats. Separate research supported by the European Commission, conducted by Polish academics and shared by ARRIS Group VP Advanced Technology Mark Bugajski later in the conference, corroborated Vasudevan’s projections.

Apart from VBR and CBR, most discussions of cable’s delivery of video over DOCSIS presume a 3.0 future. Where these talks diverge is in positioning the cable modem termination system (CMTS).

"For the initial rollout of IPTV, sending (video) downstream through the CMTS is a perfect solution," said Motorola’s Cookish. "But after a certain point, we see the CMTS becoming … both an operational and a cost burden."

In other words, Cookish et al still stand behind the so-called DOCSIS IPTV Bypass Architecture (DIBA) arguments presented more than a year ago by Motorola’s Michael Patrick, who inveighed against the inefficiencies of sending video on a "hair-pin turn" around the CMTS’s processing core and back out again.

"The logic is that that the MAC (media access control) processing capability that resides in the current CMTS is very expensive," Cookish said. "The other concept is that the cost of the edge QAM (modulator) is coming down. It’s on a Moore’s Law kind of curve."

Motorola is not alone in focusing on the edge QAM modulator. Harmonic’s "direct-to-edge" cable IPTV solution positions its narrowcast services gateway (NSG) 9000 universal edge QAM device as a bridge between an MSO’s DOCSIS 3.0 and digital video infrastructure for IP video delivery. It picked up an award for "IP Cable" at the IPTV World Forum.

Harmonic VP Business Development and Marketing Communications David Price characterized it as a second take on an IPTV proposal offered at last year’s CableLabs Summer Conference. That particular demo swept that event’s Innovation Showcase polling and was subsequently explained in the November issue of CT. What about QoS? While not formally linked, the Harmonic and Motorola approaches face opposition at several levels. As indicated earlier, Cisco executives tend to emphasize the value of aggressive channel bonding. They also portray existing CMTS architecture as fundamentally sound.

That latter point is underscored by Broadcom’s Prodan. "If you look at DIBA, what they’re thinking of doing is not connecting these edge QAMs to the M (modular)-CMTS core MAC going over DEPI (downstream external physical interface). They’re thinking of going over an IP network, with some layer 2 switches, just to connect to video servers."

"DIBA ignores the QoS requirements of video streams," he said, adding that there is no way to do rate shaping or load balancing with DIBA. Although the DOCSIS tunneling protocol could be used to sidetrack the CMTS, Prodan’s question is who or what is managing that protocol.

"There’s a missing piece," he said.

Conceding that this approach could work well for a bulk, bit-transfer business, Prodan doubts that cable operators see that as their end game. Instead, the CMTS’s support of IP with video encapsulation could be a very big advantage in maintaining a host of qualities.

Apart from the doubts levied by Prodan, the Harmonic and Motorola approaches face somewhat more friendly criticism from another camp. The point that GoBack TV Vice President of Marketing Rei Brockett makes is not that bypassing is a bad idea, but rather that it doesn’t to wait for downstream channel bonding and wideband cable modems. (See Figure 1.) "GoBackTV’s particular implementation allows a single DOCSIS core to support thousands of DOCSIS 1.1/2.0 – and soon 3.0 – cable modems tuned to multicast and unicast video across tens and even hundreds of QAM channels," Brockett said.

Moreover, GoBack technology is live in the field. (For schematic, see Figure 1.) The deployment discussed at the IPTV World Forum event was that of Marco Island Cable, a niche operator serving both year-round and seasonal subscribers on the southwestern edge of Florida.

The GoBack approach looked "like an easy solution to implement … and a good start," said Motorola’s Cookish, who adds that from his perspective, the bypass architecture is "not focused on the small operator as much as the large-scale operator who has to buy lots of CMTS capacity."

Motorola’s goal in this area, he said, is to reintroduce the concept of standardizing the bypass architecture in the second half of 2008. It’s hard to predict how the MSO and CableLabs leaders of DOCSIS will view DIBA this time around, but much probably depends on how the status – and prospects – of 3.0.

"I can’t tell you whether they’ll give us a thumbs up or thumbs down," Cookish said.

Jonathan Tombes is editor of Communications Technology. Reach him at [email protected].