Cohere Networks Inc. still isn't talking about what it's up to, but a patent recently granted to company founder Shlomo Rakib sheds some additional light on what the startup has in mind.

Rakib, a founder of Docsis pioneer Terayon Communication Systems (now part of Motorola Mobility), was awarded U.S. Patent No. 8,311,412, which describes a "distributed" cable modem termination system (CMTS), on Nov. 13.

That seems to chime with what's leaked out so far about Cohere's purported plan -- that it's developing a "micro CMTS" that could be squeezed into fiber nodes and essentially keep the Docsis transport signal in digital format longer. That shift in architecture could help cable pursue higher-order modulations that pump out more bits per hertz and wring more life out of the hybrid fiber/coax (HFC) network. (See Startup Cohere Lands Big Names & Big Bucks and Terayon Founder Plans Access Revolution.)

The patent goes into some technical detail on how Rakib's idea would work, while also giving a name to this new class of distributed device: the Cable Modem Remote Termination System (CMRTS).

Rakib's patent doesn't outline any specific capacity targets, noting only that a key goal of the CMRTS and its more neighborhood-level, granular approach is to provide "a method for enhancing the data carrying capacity" of an HFC network.

The filing also claims that this distributed architecture would be backwards compatible with the existing, more centralized architecture, able to be "configured to mimic a conventional cable plant CMTS." It goes on to say that the CMRTS could be deployed on a targeted basis in neighborhoods that require a capacity boost.

And there seems to be a software-defined networking (SDN) wrinkle to what Rakib has in mind. The filing also describes a "virtual shelf manager" that would essentially control the complex CMRTS configuration functions in software. That component would "shield this complexity from the rest of the system when needed," the filing notes.

This over-arching computer-controlled, network configuration and data distribution scheme would seem to address a potential weakness identified by some cable engineers who are familiar with Cohere's design -- that a distributed CMTS architecture could present significant device management challenges.

Docsis 3.1: Cohere's entry point?
The patent offers no sense on when this new class of product would be ready to roll. Multiple industry sources say Cohere, now run by former Infinera Corp. (Nasdaq: INFN) CTO Drew Perkins, was giving private showings of a prototype at last month's Society of Cable Telecommunications Engineers (SCTE) Cable-Tec Expo in Orlando. Cohere declined to comment for this story.

"They have a breadboard version of it," a source says, referring to an approach that lets companies create temporary prototypes and experiment with design. Cohere hasn't commented on its product plans, but it's understood that the company hopes to have a test model ready by next year.

Cohere is also coming on the scene as operators continue to deploy traditional CMTSs and make preparations to introduce new, dense Converged Cable Access Platform (CCAP) products that combine the CMTS and edge QAM functions. Because of Cohere's disruptive approach and the fact that cable operators typically don't do anything radically new until they are forced to, Cohere could try to make its move as cable operators start to make plans for Docsis 3.1, a CableLabs spec-in-the-making that is targeting higher modulations and downstream speeds as high as 10Gbit/s. (See Docsis 3.1 Targets 10-Gig Downstream and Docsis 3.1 Set for a Spring Fling .)

"Docsis 3.1 might be a natural break point for them [Cohere]," says a source who is familiar with Cohere's plans and how the cable access market could shape up in the next three to five years.

— Jeff Baumgartner, Site Editor, Light Reading Cable