Full Duplex Is a Go; Cable Aims for 10 Gig
CableLabs is moving from exploration of Full Duplex DOCSIS technology to an official specification development phase, according to multiple Light Reading sources.
UPDATE: CableLabs has now confirmed the move as well. The timing fits with earlier estimates forecast by the organization for progression to the next stage of Full Duplex development. (See Full Steam Ahead for Full Duplex DOCSIS.)
The news also comes on the heels of an announcement from Cisco Systems Inc. (Nasdaq: CSCO) earlier this week stating that it has released a silicon reference design for Full Duplex networking. (See Cisco Launches Full Duplex Silicon Design.)
Full Duplex is an important innovation in the cable world because it has the potential to enable symmetrical 10-gigabit broadband over legacy hybrid fiber-coaxial (HFC) networks. Even as DOCSIS 3.1 is designed to deliver speeds up to 10 gigabits per second in the downstream, the upstream is still constrained to top potential speeds of around 2 Gbits/s. Compared to the symmetrical capabilities of fiber-to-the-home deployments, that's a serious disadvantage. But Full Duplex could help close the gap. (See What If Cable Does the Full Duplex?)
As CableLabs moves into developing a specification for Full Duplex, Cisco is the first out of the gate with a public proposal for how part of that spec should be written. And the company claims it has the only proposal in to CableLabs that effectively offers a digital echo canceler as part of its reference design.
Echo cancellation is fundamental to the Full Duplex concept because the technology works by allowing signals to be transmitted and received on the same frequency at the same time. The sending device has to filter out its own transmission so that it can "hear" the signal coming in over the same wavelengths.
According to John Chapman, Cisco fellow and CTO of the Cable Access Business, Cisco's proposal is the only one that doesn't rely on the physical allocation of cable modems to different output ports on a fiber node. Instead, Cisco logically groups modems into different DOCSIS 3.1 interference groups with the option to swap frequencies as needed while still canceling the resulting echo. It’s a virtualized approach rather than a physical one.
Cisco's technology is also built to work with existing cable modem termination system (CMTS) architectures (a benefit for Cisco) and will sit inside a Remote PHY device deployed by an operator.
"This echo canceller will go inside of an RPD, a Remote PHY Device, and that RPD will be located down in an optical node, a box that's typically in a deep-fiber environment," says Chapman.
That design complements Cisco's earlier efforts in the Remote PHY space. In February, Cisco released open source software for RPDs in conjunction with CableLabs in a project called OpenRPD. Cisco says that software can be used by any legacy node vendor to create new Remote PHY nodes. Cisco is also developing its own RPD hardware. Chapman says the product is expected to go into early field trials later this year, with general availability following in the first half of 2017. (See Cisco Open Sources Remote PHY Device.)
Of course, there is still some disagreement about how widespread the use of Remote PHY technology will be throughout the cable industry, and controversy around Cisco's specific vision for a distributed access architecture (DAA). Several operators have started deploying integrated Converged Cable Access Platform (CCAP) devices, which, as the name suggests, converge more cable technology into a single chassis rather than distributing functions throughout the network. (See Don't Give Up on Converged Cable Access Yet.)
Where DAA deployments are concerned, there are also multiple approaches possible. Cisco's Remote PHY strategy is one. Another is the approach heralded by Gainspeed, the company which was recently acquired by Nokia Corp. (NYSE: NOK), and which developed a virtual CCAP device designed to move more cable functions deeper into the network than even a Remote PHY device would. (See Nokia Swings Deal for Gainspeed.)
In all likelihood, different cable companies will use different strategies in different locations; sometimes deploying an integrated CCAP, sometimes pursuing Remote PHY and sometimes moving to a virtual CCAP solution.
Meanwhile, Cisco isn't the only company trying to gain the upper hand in Full Duplex development either. Kumu Networks has also been working with CableLabs on Full Duplex technology, and Nokia demonstrated its XG-CABLE proof-of-concept in May to prove the feasibility of delivering 10-gigabit speeds over an HFC network. The company says XG-CABLE can be easily integrated into CableLabs' Full Duplex DOCSIS 3.1 efforts. (See Nokia Bell Labs Boasts 10Gbit/s Symmetrical Using DOCSIS 3.1.)
Regarding timing for the complete development of cable's Full Duplex technology, estimates are still hazy. But Chapman believes the cable industry could get to lab trials in 2018 and take Full Duplex out into the field in 2019. Previously, CableLabs CEO Phil McKinney has said he believes the process could happen even faster than the development of DOCSIS 3.1, which took about three years.
However long it takes, starting the official spec development is an important and noteworthy first step.
— Mari Silbey, Senior Editor, Cable/Video, Light Reading