Converging on IP: Starting from Parallel Networks
Over-the-top content is forcing the issue, but how much is it forcing the schedule?
Cable started with an RF-based network optimized for video and added a separate IP-based network to carry data services. Running two parallel networks, even when necessary, is inefficient, and almost since those first DOCSIS networks were built, there’s been theorizing about a migration to some sort of converged network.
Network convergence could potentially have remained theoretical for a decade or more, but circumstances have conspired to move the start date up. The end date, however – the day when networks are fully converged – may not have shifted so much.
The growing popularity of the Internet is such old news that it’s easy to lose sight of the fact that ever-expanding broadband usage has ongoing ramifications. One immediate consequence of the spectacular non-stop growth of the Internet is that it is pushing the subject of network convergence.
As the Internet grows, building and operating Internet protocol (IP) networks just keeps getting progressively cheaper. Consolidation on IP networks has long seemed likely, but now it’s a near certainty, but because of the economics, convergence is almost certain to mean ultimate consolidation on an all-IP network.
According to Dan Dodson, a fellow at IBB Consulting, “The three big reasons to migrate are: One, you get a richer app environment; two, capex savings on CPE; and three – after you’ve got IP video rolled out – you get bandwidth savings in the 60, 70 to 80 percent range. You convert to IP, you’re using a fraction of the bandwidth required today you can use for new services.”
While the potential benefits of IP are attractive and are giving MVPDs some economic impetus to begin moving toward network convergence, the factor that’s really forcing activity is consumer behavior.
Soaring broadband usage has been encouraging consumer electronics companies to keep pumping out ever-more-capable IP-based devices that can dangle off the ends of service provider networks. Consumers’ enthusiasm for those IP devices only keeps growing, increasing demand for IP services, especially those provided by multichannel video programming distributors (MVPDs), including – or, to be more accurate, especially – video.
MVPD customers consuming vast amounts of video on the IP pipe is what has forced the issue. It’s customer-driven convergence.
Many MVPDs have already taken some first steps toward the ultimate goal of converged networks, and the rest have such measures incorporated in their technological roadmaps.
(It bears noting that of all the major wireline MVPDs, AT&T arrived pre-converged, if you will, having opted for an IPTV service that at the get-go relied entirely on its DSL network.)
The first order of business for most MVPDs, then, has been to learn about IP (few, if any, have institution-wide expertise with the technology), and preferably to learn by doing.
Luckily for MVPDs, getting IP content, including video, on IP devices doesn’t necessitate network convergence at all. It merely requires taking advantage of the broadband pipe, ideally with terminal devices that provide at least some rudimentary opportunity for being managed. The industry has initiated what can justifiably be considered as convergence at the customer premises equipment (CPE).
One of the first, most obvious places to start has been with an IP distribution point many consumers already had in their homes: Wi-Fi routers.
MVPDs originally looked at Wi-Fi as unreliable because it was not secure, especially in comparison with a wire. But Wi-Fi is increasingly more robust, and consumers were using it anyway. With the introduction of management facilities, MVPDs have been distributing modems paired with Wi-Fi routers.
Charter Communications and Time Warner Cable are among the cable operators distributing this form of gateway. CenturyLink is another that distributes combination modem/routers.
That concept is getting stepped forward with specialized Wi-Fi routers designed to provide a greater level of management, and therefore content security. One example is Comcast’s AnyPlay, built by Motorola, which plugs into a set-top and can distribute not only over-the-top video, but also cable linear channels to iPads and other connected devices.
Another example is the box AT&T is using, built by Cisco, that plugs into an existing wireless router and forwards signals to Cisco Wi-Fi-receiving modules plugged into a TV that can be placed anywhere in the home.
IP out of set-top boxes is less common, but a notable example is the Slingbox, first adopted by Dish Network (a sister company of manufacturer EchoStar), and subsequently picked up by Time Warner Cable.
AT&T, Verizon, Comcast and others are also delivering services through Xbox Live game consoles.
Of course, an Ethernet port is now standard issue in everything from modems and routers to set-tops and gateways. For the moment, that port is a nicety, but cable is clearly planning to use it in the future.
Working through CableLabs, the industry is developing a means to deliver Ethernet services over DOCSIS networks. The DOCSIS Provisioning of EPON project, or DPoE, creates a rare gem among acronyms in that not one, but two of its constituent letters derive from other acronyms.
The CableLabs DPoE version 1.0 specifications describe a common approach for using DOCSIS back office provisioning processes to deliver business data services over Ethernet passive optical network (EPON) access networks.
So while DPoE is aimed at business services first, that’s to be expected because that’s where cable EPON is at the moment. It’s a common expectation that a future evolution of hybrid fiber/coax (HFC) networks will be to PON – more on that later.
Cable is getting close to DPoE, having just wrapped its fourth interop in less than a year.
“We’ve gone from basic registering and provisioning of devices to complex interoperability issues such as secure software download, classification and encapsulation,” said Curtis Knittle, director of digital video services at CableLabs and project lead for the DPoE specifications.
Participants in the fourth DPoE interop were Adva Optical Networking, Alcatel-Lucent, Arris, Aurora Networks, Broadcom, Ciena, Cisco, CTDI, Finisar, Huawei, MRV Communications, PMC-Sierra, Qualcomm Atheros, Shenzhen Gongjin Electronics, Sumitomo Electric Industries, ZTE and three companies new to the program: Oliver-Solutions, NTT Communications and Ixia.
That CPE convergence is perfectly adequate in the near term for handling consumer demand for IP video in the home buys some time for MVPDs to work on network convergence.
There are numerous architectural options for evolving toward all-IP, but one possibility is immediately ruled out – ripping out existing plant and installing all-IP equipment is too expensive to be a viable solution.
One of the most prominent of the recent proposals is the Converged Cable Access Platform (CCAP) project. CCAP lays out an architecture that enables MSOs to couple IP video processors to data processing CMTSs.
Until recently, a cable operator would have to deploy some QAMs dedicated to MPEG video and others dedicated to IP. The advent of hybrid models makes QAMs one locus of convergence.
To transition to all-IP, cable operators will need to make more efficient use of bandwidth and will have to make incremental investments that generate a return, according to a paper presented at Cable-Tec Expo written by Patrick Wright-Riley and Chris Poli, both from Motorola. They also recommended that any change that is made should support mobile devices.
There are several means of making more efficient bandwidth. They can be used in combination, and knowing which ones to use is likely to depend on individual MSO goals and circumstances. These means include employing switched digital video, probably for on-demand and/or long-tail content; deploying DTAs so that analog bandwidth can be reclaimed; and transitioning from MPEG-2 to the more bandwidth-efficient MPEG-4.
Upgrading to DOCSIS 3.0 is a necessity, Wright-Riley and Poli said. Establishing network DVR and VOD in the cloud are also elements of an evolution to IP.
Dodson said that if he could recommend one thing, it would be to “move on-demand to IP sooner rather than later. A hybrid settop box will help you right away.”
"IP video will take less bandwidth than QAM video," Dodson continued. "You use variable bit rate transmission and MPEG-4 for video-on-demand; that saves you bandwidth, too. Now you've given yourself the flexibility to do other things, including – possibly – network DVR."
That same infrastructure can be used to host applications. An MVPD can still differentiate itself from its competitors simply by offering a rich set of apps.
Back on an architectural note, the industry is also talking about putting edge QAMs in the node, said Aurora Networks vice president of marketing John Dahlquist.
The optics are all-digital; transmission could be all-IP to the node, where the QAM would turn it to RF. An incidental but not unimportant benefit is that the use of more optics automatically drops power consumption. And everything would remain compatible with all of the other equipment deployed today, Dahlquist noted.
In a new report, Jim Barthold of NPD Group examined the evolution of cable to all-IP and concluded that it is both inevitable and unlikely to happen anytime soon. (The NPD report is called “The benefits of managed IP video delivery: Why service providers are migrating – slowly but surely – toward IPTV.”)
In his report, Barthold says: "Every MVPD will migrate to a managed all-IP network, and some are pursuing this migration more aggressively than others. Comcast and Verizon – not surprisingly the two biggest video providers in their respective genres – have already moved some services to an all-IP format and are trialing others, while taking care not to impact existing RF services. Other top MSOs, all of which have taken slightly different approaches to IP, are less aggressive. Second-tier cablecos are hard pressed to move away from traditional RF video delivery to what would be for them an expensive migration to all-IP services."
First, there is too much investment in legacy MPEG equipment, both in the access network and in subscribers’ homes in the form of CPE.
Furthermore, cable simply isn’t seeing enough competitive pressure to make the evolution to all-IP a higher priority. Competitive services coming over the top are unmanaged, and therefore are less viable threats.
AT&T is all-IP, and while that gives AT&T benefits such as lower operational complexity, lower costs and the potential for faster service velocity, the company must work within the constraint of the limited bandwidth afforded by its DSL network. So AT&T isn’t going to generate much competitive pressure on MSOs.
“OTT isn’t pushing them. The telcos aren’t pushing them,” Barthold said. When it comes to the timing on a transition to IPTV, “some guys say this won’t happen for 10 years.”