End-to-End Ethernet Gains Traction as Step Past DOCSIS

Jay Chambers, VP, edge processing business development, ARRIS

Jay Chambers, VP, edge processing business development, ARRIS

December 9, 2011 – It will take some time to get there, but the outlines of an all-new mode of signal distribution over cable plant are coming into focus with major implications for long-term savings in capital and operations costs.
While the details of how these new concepts will shake out are a long way from solidifying there are strong cases to made for several ideas which, when combined, add up to a compelling picture of a network architecture that leverages the traditional hybrid fiber coax physical plant to create an Ethernet-based end-to-end digital stream of IP-based services. Should these concepts come to fruition cable operators will be able to leave behind the complications and costs associated with amplitude-modulated lasers, capacity limitations imposed on coaxial plant by severe signal attenuation and the mounting burdens of maintaining separate infrastructures for TV and data services.

As reported nearly two years ago (January 2010, p. 1 and March 2010, p. 18), an Ethernet-based pipeline to the home offers many advantages as a replacement to the traditional DOCSIS system, but, until recently, the industry showed little interest in changing course as it pushed ahead with deployment of the latest iteration of the standard, DOCSIS 3.0. However, in early November the IEEE began an initiative with participation of key industry players that aims to define specifications for what is known as Ethernet Passive Optical Network over Coax or EPoC.

At an exploratory meeting attended by 73 engineers and other interested parties on November 8 during the SCTE Cable Expo convention in Atlanta, participants voted overwhelmingly to form a study group for purposes of developing criteria for an EPoC infrastructure, with the first meeting scheduled for late January. With representatives of five cable companies in attendance listed as supporters, including officials from the top three MSOs, with Time Warner Cable principal engineer Kevin Nolls among the four presenters at the meeting, the initiative appears to have the backing long lacking for an Ethernet-centered approach to cable infrastructure.

“EPoC is definitely getting some notice,” says Jay Chambers, vice president of business development for edge media processing at ARRIS. “There’s a lot of forward thinking going on now, rather than continuing to stick with the idea that you can solve all problems with broadcast and DOCSIS. There’s definitely a shift underway.”

ARRIS’ just-completed acquisition of BigBand Networks, where Chambers served as chief architect, is one reflection of that shift. As ARRIS veteran Jeff Brooks notes, the intelligent edge system developed by BigBand to support direct IP-based video distribution without utilizing downstream resources on the DOCSIS CMTS (cable modem termination system) expands the ability of ARRIS to address changing architectural requirements.

“Some operators want to be able to ignore the transport layer entirely as they move to IP services,” says Brooks, who serves as vice president of business development at ARRIS. “The bypass technology developed by BigBand allows them to evolve services and apply intelligence for new applications over the network very easily, much as things are done over an IP-based telco infrastructure.

“The bypass technology is also useful where operators wanting to migrate to IP video have a lot of QAM deployments but don’t have DOCSIS 3.0 infrastructure in place,” he adds. “And some operators might want to use the platform because they have some operational method to go past MPEG2-based management systems and go straight to IP. So we now have multiple ways to meet operators’ requirements.”

This last category of potential users of the DOCSIS bypass mode includes many major operators worldwide who are beginning to converge their backend TV operations management systems onto an IP-based infrastructure where encoding output for legacy MPEG TV streams becomes just another distribution category for content that is ingested, stored and processed through the IP-based core infrastructure.

As previously reported (July, p. 1, September, p. 1), operators such as Comcast in the U.S., Australia’s Telstra and Europe’s Liberty Global have publicly embraced the IP-based backend integration for all TV services, although none have publicly disavowed adherence to the DOCSIS 3.0 approach to implementing TV over IP broadband infrastructure. But the “dumb pipe” idea represented by the EPoC initiative could radically alter the physical layer IP migration path while opening new possibilities to use intelligence in the network at whatever points are most conducive to maximizing operational flexibility and efficiency.

Critically, as Time Warner Cable’s Kevin Nolls noted at the November IEEE meeting, EPoC must be designed to maximize migration flexibility so that the distribution of Ethernet signals can coexist with legacy modes, including analog and digital TV, DOCSIS services and voice, in whatever ways are most appropriate to an MSO’s service strategies and HFC infrastructures. In other words, the transition point or “PHY” where the digital Ethernet optical stream is modulated via quadrature amplitude modulation (QAM) onto RF spectrum for transport over coaxial plant to the home must be designed to accommodate virtually any spectrum allocation for the Ethernet streams in the downstream and upstream paths.

“The EPoC PHY would need to be flexible and permit re-provisioning to make use of more RF spectrum as it is made available by the cable operator,” Noll said. As he and others have noted, however, introducing EPoC wouldn’t just be a matter of trading out bandwidth used for existing distribution modes to make room for the Ethernet streams. Instead, by using low-cost digital lasers rather than amplitude-modulated lasers to deliver the EPoC signals over dedicated wavelengths, operators would increase the signal power delivered optically to the RF PHY unit at the node, which would have the effect of freeing up more bandwidth on the coaxial plant.

In other words, where the attenuation impact of coaxial plant amplifiers on higher spectrum frequencies limits use of those frequencies when AM lasers are feeding the node, those frequencies begin to open up for service delivery if the optical signal-to-noise input can be increased, as would be the case with digital lasers. The frequencies freed up for EPoC transport might range anywhere from a couple of hundred megahertz above the current cutoff point at 750 MHz or 860 MHz to one gigahertz or more in cases where fiber penetration is so deep that there is no need for amplifiers on the coaxial plant.

Making all this a practical alternative without a huge forklift upgrade of headends and other infrastructure is a key goal of the EPoC initiative, which means there will have to be an EPoC MAC (media access control) specification that would serve as the interface between the data pipe and everything else to ensure that the specific quality-of-service parameters attending each type of service are maintained across the distribution network as ever more services are fed into the EPoC transport layer. The new MAC would have to “talk” to the traditional DOCSIS MAC, MPEG MACs and other data protocols and translate them into commands at the Ethernet MAC layer.

Beyond the capacity benefits, the larger argument for moving to EPoC has to do with the lower costs of operating over an IP/Ethernet infrastructure, including the savings that go with convergence of all processes into the IP domain as well as the low costs of CPE and other equipment. On the downside, opponents of this approach argue the placement of the required intelligence in the node not only adds costs to the node but exposes that intelligence to the climatic and other disruptive forces that impact outside plant. Moreover, they argue, whatever specifications are created to support this architecture, they will not have the proven field performance and reliability that make the standard HFC architecture with DOCSIS a dependable way to migrate to IP.

One key promoter of the EPoC idea is chip maker Broadcom, which has proposed refinements in its commercial EPON MAC chips that would incorporate the PHY conversion to RF for distribution of the Ethernet signals over coax right into the fabric of the ASIC. In fact, the IEEE meeting in November was a key step toward the industry support Broadcom says it is looking for if it is to move in this direction.

Another vendor with product capabilities that would exploit EPoC once specifications are better understood is Aurora Networks, which over two years ago introduced a product idea, “BitCoax,” that envisioned use of Ethernet in conjunction with low-cost small form factor digital lasers to open up spectrum on coax for IP-based services. Now, as reported in May (p. 14), Aurora is preparing to introduce an expanded version of the concept as a product once specifications are set.

“People are looking for ways to expand capacity without adding CMTS blades,” says John Dahlquist, vice president of marketing at Aurora. “We’ll be there with the product once the specs are set.”

While just how the execution of the MAC requirements might be achieved remains to be determined, Aurora offers a field-proven model that pushes back against the argument that the complexities and time delays associated with the distributed processing model are too daunting. This has to do with the firm’s use of time division multiplexing over Ethernet as a way to emulate legacy T-1 infrastructure used in cellular backhaul to facilitate the transition from copper-based backhaul networks to the higher capacity capabilities of fiber.

As a solution certified to be compliant with the Metro Ethernet Forum’s MEF 18 specification, the Aurora TDM-over-packet technology had to overcome limitations of traditional “pseudo-wire” solutions by assuring the jitter, wander, frame loss and packet delay metrics on a per-link and aggregate backhaul network basis are well within tolerance levels set for all generations of GSM, CDMA and WiMAX networks. In cellular operations synchronization must extend across hundreds or even thousands of base stations connected to any one controller.

As for the need for distributed intelligence that will be able to support functionalities such as switched video, ad insertions and adaptive rate transrating uniformly, irrespective of the transport modes, the technology developed by BigBand provides ARRIS the means to support these requirements, notes Jeff Brooks. “The edge management capabilities of the [BigBand] MSP platform gives us the tools to help with session management and exploit the IP strengths while continuing to operate over the legacy MPEG system,” he says. “It doesn’t really matter what the format of the session is, it manages all sessions. It’s a very smart platform.”