The company has now expanded the service into its New England franchises on its way to meeting a goal of reaching two thirds of its subscriber base with bonded-channel connectivity by the end of next year. The immediate impetus behind the ultra high-speed service, which operates at 5 mbps on the return path, is rising demand for IP transport capacity resulting from multiple users in the home, greater volumes of video consumption, online gaming, telecommuting and other usage patterns, says Cox spokes person David Deliman. “Our research continues to show that one of the reasons people are attracted to Cox is that we offer the fastest broadband speeds,” he says.
But at Cox, as has become the case at many cable companies, the underlying assumption is that eventually everything will move to IP, which requires sensitivity to long-range needs as short-term requirements are met, notes Randy Kinsey, executive director of engineering technology for Cox. “Everything is evolving into the IP space,” Kinsey says. “When that will happen is the only debate.”
For now, Cox seeks to get the most out of its legacy infrastructure and set-top boxes, understanding that IP-based services can be layered in, as has been the case in Cox’s transition from circuit-switched to IP voice. But a smart migration path will require that choices be made at the underlying transport level sooner than later to make sure the migration moves as smoothly as circumstances dictate, Kinsey adds.
“We always try to keep an eye out on things that are coming our way in the transition to IP or, eventually, we’ll find our options are restricted,” he says. “We’re making sure the work we’re doing now is foundational to meeting those future requirements.”
A case in point is the work that must be done to prepare for Cox’s move to a next-generation mobile service, which, as previously reported, will likely be based on LTE (Long Term Evolution) technology, in contrast to the 4G WiMAX platform pursued by MSO partners in the Clearwire venture. 4G mobile will be an all-IP service that will require preparations on the fixed network side to accommodate the multi-platform synergies between fixed and mobile. “We’re looking at that pretty hard,” Kinsey says.
So far, all of Cox’s DOCSIS 3.0 deployments, now covering its Northern Virginia, Lafayette, La., and Palos Verdes, Calif., territories as well as Rhode Island and other New England segments, are utilizing Cisco Systems’ uBR10012 Cable Modem Termination System (CMTS) in conjunction with its DPC3000 cable modems. This is a modular CMTS platform, which means the RF modulation process, known as quadrature amplitude modulation (QAM), is performed on QAM units positioned at the network distribution hubs away from the CMTS rather than within the CMTS chassis.
There has been considerable debate within cable technology circles over whether the initial launches of DOCSIS 3.0 are best carried out on integrated CMTS (I-CMTS) or M-CMTS infrastructures. Most engineers agree that, eventually, the industry will want to exploit the M-CMTS architecture in order to take advantage of the multi-service efficiencies of universal edge QAMs. U-QAMs can combine different types of service streams into 6 MHz RF cable channels, thereby maximizing bandwidth availability for any given service category and eliminating the need to add service-specific QAM resources over time. But adaptation to U-QAMs has been slow, leaving many engineers inclined toward use of I-CMTS architectures.
Cisco, which currently only offers an M-CMTS DOCSIS 3.0 option, has “no particular religion” on the issue, asserts Dave Brown, cable solutions manager at Cisco. “It so happens that the product we’re shipping initially uses the modular approach,” he says. “But we’re also delivering new hardware that’s capable of supporting the integrated approach.” This is a development that has yet to be formally announced.
But there are many advantages to the modular design stemming from the efficiencies associated with universal edge QAMs, Brown adds. “With universal QAMs, if you need more data capacity to handle peak rates or higher tiers you can take advantage of available capacity on the existing QAMs before adding new ones,” he says. “If you have 2 megabits of capacity left on an RF channel, it may not be enough for a TV channel but it can be used for DOCSIS.”
There are also important implications for migrating services to IP, he notes. “Say you’re offering traditional video on demand and in the future you want to migrate those QAMs and start offering IP video services over DOCSIS channels,” he explains. “It’s very straight forward to migrate over to IP on those QAMs,” as opposed to duplicating QAM capacity by having to use the I-CMTS QAMs for the new VOD service.
But not everyone is persuaded. “One of the things we’re witnessing is that there are two camps forming around the issue of integrated versus modular,” acknowledges Murali Nemani, director for market management in Cisco’s Service Provider Marketing group. “Customers who are going to an aggressive IP deployment strategy are looking at how we’ve expressed our vision for Medianet [a highly integrated, distributed intelligence-based approach to enabling IP services] and how you make the transition to an all-IP network,” Nemani says.
“It’s a vision that assumes a shift to a unicast model that supports a high level of personally interactive services,” he continues. “Certain major cable operators in America and Europe are making these hard calls now. Other customers are saying, we’re still largely in a broadcast mode and don’t foresee the need for a major change anytime soon.”
While Cisco is moving to accommodate the I-CMTS segment of DOCSIS 3.0 demand, its vision is that the evolution to an all-IP network “from the core to the devices through the access network” is inevitable, Nemani says, which necessarily brings M-CMTS architecture into play. “That’s where we see the advantage of being able to scale QAMs to as many as 400 per chassis, which is what we do in our Gateway 10 product,” he adds. “When you map to this capacity, you have a very efficient way to expand the volume of video over DOCSIS and to migrate to a unicast model. We think the most innovative service providers are driving toward the model we’re talking about.”
The Cisco executives dispute the notion that the I-CMTS approach is more cost effective than M-CMTS in terms of immediate needs to merely create a higher-speed, more efficient pipe for broadband service. “We have some cost curves that show how we’ve dramatically reduced the cost per bit in the UBR 10012 chassis,” Brown says.
Adds Nemani: “The industry has been trying to understand the scalability and density rationale behind our modular approach in comparison to the cost calculations on the I-CMTS model. We’ve done a tremendous amount of modeling based on industry roadmaps, and our customers are agreeing with our conclusions.”
Indeed, Nemani notes, Cisco’s overall CMTS port count, including all DOCSIS generations, didn’t hit one million until 2007. Now it’s over two million. Cisco’s DOCSIS 3.0 platform is now deployed in 130 cable systems worldwide, he says.
Interestingly, despite Cisco’s rationale for choosing a modular approach, Kinsey says Cox’s choice was driven by the fact that it already was using Cisco’s CMTSs for DOCSIS services and preferred to stay with that vendor rather than switching to another. Noting that in some of its cable systems the company uses CMTSs supplied by Motorola, which follows an I-CMTS approach with DOCSIS 3.0, Kinsey says, “Our launches of DOCSIS 3.0 don’t depend on the use of modular features. Where we’re deploying with Cisco this is the natural evolution of the platform we have in place to get to DOCSIS 3.0.”
But that’s not to say the MSO won’t decide a modular approach makes the most sense at some point. “We are looking at those kinds of things,” Kinsey says of the migration advantages that might attend modular. “There are situations where it does make sense from a density and bandwidth perspective, but those considerations aren’t driving our decision in this case.”
Channel bonding, which in the present case entails combining four channels to support raw throughput of close to 160 mbps downstream, allows operators to offer user rates exceeding the 38 mbps capacity of a single 6 MHz RF channel. But it also helps free up capacity for streams delivered over DOCSIS 2.0 and 1.1 platforms, thanks to the statistical multiplexing efficiencies that come with turning four RF channels into a single pipe. Brown says the efficiency gains of using DOCSIS 3.0 can add up to as much as 20 to 30 percent of the allocated bandwidth.
While this is indeed a benefit, Kinsey says, the Cox strategy is driven primarily by the need for a higher tier of broadband. “We need to get to higher speeds and broaden the service capacity,” he says. In fact, to support even higher speeds, Cox offers a 10 percent power boost on the 50 mbps service at no extra cost if bandwidth is available and a customer wants to burst to 55 mbps.
In Rhode Island Cox is charging $144.99 per month for the residential version of the 50 mbps service. This includes three unique IP addresses, 10 e-mail addresses and 50 hours per month of remote dial access. Prices vary elsewhere. For example, the rate is $139.99 in Northern Virginia, with an introductory price set at $109.