Squeezing the Costs From IP Video; Unbundling data QAMS from CMTS

One of the livelier efforts in cable’s geekosphere is a quest to move digital video bits as cost effectively to cable modems as they move to set-top boxes.

And now this, from the Department of Why Bother: This isn’t about sending video to one or the other, set-top or cable modem. It’s about being able to do both, without cost penalties.

The logic goes like this: Why shouldn’t video streams swoosh through cable modems to whatever connected PC, home network, or handheld device? They all speak in IP (Internet protocol).

Cable modems speak in IP. Why not ride that wave?


The effort goes by many names. Some call it by the product it seeks to become: “M-CMTS,” for “modular cable-modem termination system.”

Similarly, “distributed CMTS.”

Others call it by what it does: “Separation of MAC and PHY,” where MAC is “media-access control” (software that handles the shared aspect of cable’s passageways) and “PHY” is shorthand for “physical layer” (plant and, mostly, modulation.)

Still others call it by where it might wind up: DOCSIS 3.0. A word of caution here (again): There is no official Data Over Cable Service Interface Specification 3.0 yet.

Plus, something like M-CMTS could theoretically work on DOCSIS 2.0.

As technology efforts go, this one is long on brainy.

Here are a few signposts: It’s a video transport thing, to start. It answers this question: How does one send video bits over this vat of gear, and not that vat of gear? Or to both?

M-CMTS is also a cost thing. Say you were to calculate how much it costs to move a digital TV show to set-tops over the cable plant. Say you were also to calculate how much it costs to send that same show over the plant to cable modems.

You’d find that the path leading to set-tops is way less expensive. (That’s “way” with head back and eyes rolling.)

The rest of this translation will examine why that is, and what’s being done about it.


In order to put anything, on any channel, in the digital domain (between 550-750 MHz, spectrally), you need a QAM (quadrature amplitude modulator).

This is true no matter what you want to put on that channel — broadband Internet, video on demand, voice-over-Internet protocol or digital video.

Doesn’t matter. If it’s digital, you need QAMs.

Right now, operators buy QAMs at a rate of one per channel. Example: If you’re using 20 channels (these are 6-MHz channels) for digital video, you need 20 QAMs.

And there’s a multiplier. You need those 20 QAMs at as many distribution hubs as are connected to a headend. Distribution hubs are the place where groups of 500-home nodes are ganged into “service groups.”

Let’s say there are five hubs. That’s 100 QAMs, just for digital video, and not including video-on-demand or HDTV. Digital video isn’t new. Video QAMs aren’t new. When an industry needs lots of something, over time, pricing responds favorably. This rings true for video QAMs, which run in the $500 range.


It’s also true that if you put two 6-MHz channels into action for broadband Internet service, you need two QAMs — and those QAMs run in the $1,000 range.

Why double? Because the QAMs used for broadband, at this moment in time, are built into the headend gear. Specifically, they’re built into the CMTS, which is the companion unit — the master controller — for all cable-modem activities.

When you buy a CMTS, you’re buying a thing that routes Internet traffic downstream, yes. You’re also buying a handful of upstream traffic receivers, and those built-in QAMs.

Until the CMTS can be made less monolithic and more modular, moving video through it is doomed to be considerably more expensive than sending it over paths designed and built for one purpose — to haul broadcast video.

Remember when video-server suppliers were encouraged by their customers (MSOs) to separate streaming mechanisms from storage mechanisms, because both grow at different rates?

The modular CMTS is anchored in similar logic: Break the big thing into its constituent parts, so that more suppliers can compete for the parts they’re good at making, and prices can respond favorably.


Which brings us to the “decoupled CMTS,” or the “modular CMTS.” The concept itself is an outgrowth of the Next Generation Network Architecture (NGNA) mind meld, a Comcast Corp./Cox Communications Inc./Time Warner Cable effort of 2003-04 that’s now being shepherded by Cable Television Laboratories Inc.

It’s still early, as CableLabs projects go. Step one is to identify the interface points within a CMTS that can be “opened up” and made interoperable among suppliers.

One of the constituent parts is the QAM. By unbundling data QAMs from the CMTS, operators move closer to cost parity with video QAMs. That means they can start thinking about IP video without constantly having to stop to pick at cost barriers.

Ultimately, the goal is to allow multiple services to share QAMs — whether voice, video or data. Stop buying “silo QAMs,” in other words, and start buying QAMs that work with any digital service.

All of this pushes the modulation function closer to the “edge” of the network. You’ve probably heard it said, in fact: “Edge QAM.” Where is that? More on that next time.

Stumped by gibberish? Visit Leslie Ellis at www.translation-please.com.