No Sign of Consensus In Cable IP Migration

John Burke, SVP & GM, converged experiences, Motorola Mobility

John Burke, SVP & GM, converged experiences, Motorola Mobility

March 21, 2011 – Notwithstanding some aggressive baby steps by Australia’s Telstra, Time Warner Cable and other MSOs in the direction of delivering live as well as on-demand cable content in IP mode, the industry’s progress toward IP TV promises to be a longer, harder slog than would be the case if there were greater consensus on key technology issues.

Most of the early efforts to formulate migration strategies are under wraps, but from discussions with vendors and other sources, it’s clear the question is no longer whether but how and when the transition from reliance on traditional MPEG transport to IP-based distribution of TV content will be made. Some, like Telstra and Time Warner Cable, are starting out with offering some IP TV services in parallel with legacy MPEG services, typically with the help of bonded DOCSIS 3.0 channels to accommodate the video streams.

In Telstra’s case, subscribers’ ability to receive the service depends on whether they purchase a “T-Box” hybrid set-top or connected TV or Blu-ray player licensed for use with the MSO’s “BigPond TV” service. TWC at this stage is relying on a native IP device, the iPad, to receive the service and has announced plans that include service delivery to TVs supplied by Samsung and Sony and a trial of IP TV service employing the Microsoft Mediaroom middleware for access over IP set-tops.

Comcast, now preparing for a trial slated to get underway by midyear, is taking a different approach tied to use of IP gateways for accessing pure IP TV linear and on-demand content. Liberty Global, on a somewhat faster track, plans to introduce all-IP services starting in The Netherlands later this year, then moving to Switzerland in early 2012 and other markets after that. Other MSOs will rely on new premises media gateways that will encapsulate legacy MPEG-2 TV content into IP streams for distribution to all devices, including low-cost IP set-tops, in the home.

Complicating things even farther, variations within each of these approaches from one MSO to the next make it impossible to deliver a one-size-fits-all solution even within, say, the pure IP-over-DOCSIS mode. Not only is there a debate over whether to use or bypass the CMTS (cable modem termination system) in such applications; there are important differences pertaining to whether key functions such as multicasting should be facilitated through use of PacketCable Multimedia servers or via the usual Internet HTTP (Hypertext Transfer Protocol) approach

“It’s not that we, as a vendor, can’t accommodate all these different strategies,” says one senior supplier executive, asking not to be named. “But the way things are unfolding is a significant departure from the cooperative spirit we’ve seen in the past, where new standards served to expedite deployments and keep costs down.”

Confusion over the best way to migrate to a converged IP service infrastructure, which most agree is the end game, is causing many MSOs to hang back in hopes of learning more from those who are plunging ahead. “We’re seeing RFPs from some big companies in North America and Europe, but far from the majority, and there’s even less activity in Asia and elsewhere,” the executive says.

But that’s not to say business isn’t picking up on the IP migration front. “Telstra is one of the first of many pay TV operators we’re working with around the world to bring a next-generation TV experience to customers,” says Marty Roberts, vice president of sales and marketing at thePlatform.

The Comcast-owned provider of online video publishing services has been a key player in accommodating cable operators’ transition to TV Everywhere with an initial focus on delivering Web-hosted on-demand premium content to PCs and other devices. Now thePlatform is putting its content management system to use in supporting distribution of linear as well as on-demand video to the TV.

“All the things we like about on demand [in the Web space] – near-infinite video libraries, advertising capabilities, dynamic guides for discovering content with integrated search and recommendations – are migrating back to the TV,” Roberts notes. A key challenge is that the IP linear and on-demand content has to be delivered at quality levels commensurate with traditional pay TV viewing.

“The goal is to provide a dynamic user experience with recommendations, content discovery and with a management layer that sits behind it that operates at a carrier-class level,” Roberts says. While Telstra is only delivering seven channels of linear news and sports programming in IP at this point, complementing a large portfolio of on-demand content, the operator is well positioned to add additional content to the linear IP stream over time.

That’s because “the management layer for all video is all in IP,” Roberts says. “The way the guide is presented, the way users are authenticated for playback, etc. – it’s all managed as a middleware layer that thePlatform operates for Telstra.”

The IP streams are directed at the transport level over the router-based Content Delivery System supplied by Cisco Systems, with interfaces to SeaChange International and Alcatel-Lucent servers as part of the CDN (content delivery network) infrastructure. Content protection on the IP side is provided by the digital rights management system supplied by Google’s newly acquired Widevine unit.

For households that want to augment their regular cable TV service with the IP-based service, Telstra is deploying the T-Box from Netgem, and it has deals with Samsung and LG Electronics to enable subscribers who buy those suppliers’ connected TVs to access the new service as well. A number of operators are opting for this type of strategy where the gateway set-top can accommodate IP and MPEG streams as two separate services while leveraging the efficiencies of an IP-based unified service management plane, Roberts says.

Indeed, this is what thePlatform had in mind when it designed its new mpx management system, he adds (see February 2010, p. 26). “We manage all the video independently,” he says. “The customer may have different windows for distribution with different restrictions on which devices can be targeted. They can model all the functionalities with enforcement of those rules on a per-subscriber basis at the time of playback. And they can personalize the experience for each user.”

Roberts adds: “Telstra is one of the operators we’ve been working with that is taking a very progressive approach by looking at best-in-class solutions up and down the entire technology stack.” The deal is “one of many we’ll be announcing over the next several months.”

Ironically, while this best-in-class modular approach frees MSOs from reliance on end-to-end turnkey solutions such as telcos used in earlier IPTV rollouts, it also frees the cable industry from reliance on a consensus-based solution. For traditional suppliers of cable network and customer premises solutions that puts a premium on software expertise as the key to driving new business.

Motorola Mobility is a case in point. “Our Medios suite of software solutions is very extensible, because not every service provider has the same strategy for how they want to deliver content in a multi-screen world, and not every service provider has the same business model driving their deployment,” says John Burke, senior vice president and general manager for converged experiences at Motorola Mobility.

“Some large service providers want to deliver everything in an IP format straight from the network to the device,” Burke says. “Other operators are looking to leverage devices like set-tops that are in the home and want to be able to move content from those set-tops, from those DVRs, to mobile devices as well. Medios provides a broad array of software tools to allow operators to address multi-screen opportunities as they see fit.”

One way around many of the uncertainties attending streaming of services in pure IP mode, not to mention the bandwidth issues imposed by simulcasting of the same content in IP that’s delivered over MPEG, is to deploy home gateways that convert the MPEG channels to IP in cases where a household is interested in an all-IP service. “We’re seeing several of our customers who have actually started deploying that approach,” Burke says.

“It’s an advantageous approach for the operator because it doesn’t require a wholesale replacement of their video infrastructure,” he adds. “When you bring your MPEG content as well as your IP content through this multi-tuner gateway, you can serve that content up as IP streams to very thin set-top devices or directly to these new Internet-enabled TVs and Blu-ray players.”

In some cases operators are looking on the use of IP distribution as a way to drive demand for new service tiers, such as multi-room service. “They’re using that to justify going into the house and putting in new set-tops that are MoCA enabled,” Burke notes. “But other operators are taking a more ubiquitous approach saying, ‘It’s more valuable to me to get a home network infrastructure deployed as quickly as possible,’ because they see the value that could bring to a whole host of converged experiences and applications they see coming down the pike.”

Liberty Global, which is relying on NDS as middleware supplier to gateways built by Samsung, is taking a more aggressive approach by delivering all TV services in IP mode to premium subscribers who opt for installation of media gateways (see November issue, p. 1). In an appearance at the European Cable Congress in Lucerne, Switzerland last month, Liberty CEO Michael Fries framed the MSO’s strategy as one that would be so feature rich it would dissuade subscribers from any interest in Google TV and other OTT options.

“We create a comprehensive multimedia centre for your home,” he said, as quoted by the European press. Critically, he added, Liberty is moving away from reliance on the walled garden as the exclusive source of content for its subscribers. “The new Horizon box is intended to change all this and open up Web content to the TV set as well as act as a hub in the home,” he said, noting that navigation across all service categories will be facilitated by the NDS Snowflake user interface platform.

In this migration strategy, the people who get the home gateway will be on a DOCSIS 3.0 bonded-channel feed that delivers all services over IP using MPEG-4 compression to conserve bandwidth. At the outset these will be premium subscribers, Fries said, but, eventually, the MSO intends to sign up new subscribers on the gateway-based IP service and at some point will be able to convert everyone to the all-IP service, eliminating the MPEG-2 flow altogether.

As quoted by Digital TV Europe, Fries made clear the economics of the advanced gateway approach are working in favor of this approach to IP migration. “In the end, you have to have a hard drive, advanced processing power in the device and an amazing user experience,” he told the publication. “There is only so much of that you are going to achieve in the cloud.”

Nonetheless, he added, “This isn’t costing meaningfully more than digital boxes did, say, two years ago. We’re not putting such an expensive device in the home that it’s going to blow our economics – that’s not the case at all. It’s an advanced device, but it will still be heavily reliant on our network – it will receive VOD content, it will source Web content; it will be upgradeable remotely.”

While officials haven’t discussed the way they’re going about allocating bandwidth to support IP simulcast, any MSO moving at this early stage to the full-service simulcast approach must deal with the limitations on channel bonding imposed by the current generation of cable modem microprocessors. While eight-channel bonding is available in the latest modems, four-channel bonding is the norm at this point.

But the total throughput available over four bonded channels, aggregating to about 160 mbps, may be sufficient to meet the bandwidth requirements for delivering all broadcast and on-demand narrowcast content to a limited number of subscribers. That’s because the IP mode of delivery allows operators to bring several techniques to bear in reducing bandwidth required for delivering all broadcast and on-demand content in comparison to the huge amount of spectrum required to support broadcast and narrowcast over MPEG-2 transport.

For starters, H.264 MPEG-4 compression reduces the per-stream bandwidth requirement by half. Through use of statistical multiplexing across all bonded channels and all DOCSIS-delivered content, including high-speed data and voice, operators can achieve another 40 percent or better reduction in comparison to how bandwidth is consumed when services are siloed into separate IP, MPEG-2 broadcast and MPEG-2 unicast streams (see September 2009, p. 32).

Beyond these steps the MSO can use the IP equivalent of switched digital video, which involves a complex approach to IP multicast that only sends a live linear program to a service group when at least one person in the group asks for it. This serves to minimize the number of channels being delivered to any given service area at any given time by a significant amount compared to what the channel count would be if the entire broadcast lineup were delivered at one time.

“The bandwidth savings depends on how many channels are available over the multicast network in proportion to the number of users, with the savings increasing as that ratio increases,” notes the unnamed vendor executive quoted earlier. The bandwidth savings could be anywhere from 20 to 60 percent, depending on that ratio.

Thus, in the case where a very small proportion of users on a given node in the HFC network are equipped with gateways to receive all-IP service it’s likely an operator, using all the techniques described here, could offer a full-channel load along with unicast streams to serve on-demand requirements over just four bonded channels. “A lot depends on how many unicast streams you have to deliver, which depends on how many devices are accessing on-demand content,” the executive says. “That unicast count can go up pretty fast, so you’d want to get beyond four bonded channels as soon as possible.”

Operators can also apply various techniques on the unicast side so as to limit bandwidth consumption, possibly in conjunction with use of adaptive streaming, where the data rate and therefore the quality of resolution on certain content could be throttled down when there’s a threat of congestion. But how far operators would want to go in this direction is questionable.

The larger question respecting the way IP content is delivered using these various techniques is whether cable operators want to manage their IP flows through the capabilities of the PacketCable Multimedia protocol, which is an extension of the PacketCable protocol developed to support voice-over-IP service in cable, or to rely on the mechanisms used in conjunction with HTTP over IP. This is a huge question with respect to the types of gear and architecture to be deployed in support of all-IP services. On the one hand PCMM provides an extensive set of tools for managing services that are well suited to the DOCSIS HFC environment; on the other hand an HTTP-based approach to streaming architecture keeps the MSO on track to exploit ongoing developments in the Internet technology domain.

One of the advantages to pursuing the gateway approach to IP where the lion’s share of service is delivered over MPEG transport and then encapsulated for IP distribution to IP devices in the home is that operators can take advantage of the whole-home benefits of operating in IP mode while postponing the transition to all-IP delivery until chips supporting 16- or 24-channel bonding are in the market, which will happen over the course of the next couple of years. At that point it will be possible to convert a large number of customers to the IP flow, including “flash-cut” scenarios where, on a node-by-node basis, once all users are equipped with the hybrid gateways, the MPEG transport stream could be turned off and the IP stream turned on, obviating the need for simulcasting.

It happens that availability of such channel-bonding chips in mid to late 2012 will coincide with the projected time frame for commercial availability of the chips designed to support the new CMAP (Converged Multimedia Access Platform) architecture developed by Comcast engineers and now embraced by a number of major MSOs, though not all. As previously reported (June, p. 8), this technology will provide operators a flexible approach to assigning bandwidth to MPEG transport streams versus IP streams while greatly reducing the number of QAMs (quadrature amplitude modulators) required to support multiple types of service.

With the availability of CMAP devices, which would supplant today’s CMTSs, operators would be positioned to transition to IP TV much more gracefully and at lower costs than is possible today. But here again the question of consensus on key issues arises, as various MSOs move to adapt the CMAP concept to different strategies. Sources report that these variations are adding still another layer of uncertainty in efforts to develop solutions that can benefit from the economies of scale that standards-based solutions enjoy.

“There’s been a lot of talk about the transformation to IP – is it a revolution or evolution?” Burke notes. “What you’re seeing is most operators saying the evolution approaches make a lot of sense right now. So I think that’s the path you’ll see over the next couple of years.”