Skip to main content

How to Leapfrog Telcos’ Fiber Dreams

Last time, we explored why 6 Megahertz channels are 6 MHz, and whether they still matter in an all- digital world.

This time, we’ll look into the clamor of activity around “channel bonding,” which also goes by “wideband” and “DOCSIS 3.0.”

A word of caution about the latter term: To be accurate, channel bonding is likely a component of DOCSIS (Data Over Cable Service Interface Specification) 3.0, but it isn’t the only thing under consideration for inclusion in the specification. In fact, it hasn’t yet been decided what all goes into DOCSIS 3.0, involved technologists say. So we’ll leave that for now.

The idea of channel bonding is to paste a number of 6 MHz channels together, sum the aggregate throughput, and proceed with one larger tunnel inside the big pipe that is a digital cable system.

Say you adjoined three 6 MHz channels at the headend. Each one, using 256 quadrature amplitude modulation (QAM), has a carrying capacity of 38 Mbps. Multiply that by three, and voila, you get a pipe capable of 114 Mbps, aimed at a single modem. Bond 10 channels, get 380 Mbps — again, pouring into a single modem.

Of course, one could pour that much bandwidth into a home (as opposed to a single device) right now. In theory, a home that has a cable modem, an MTA for voice-over-Internet protocol and a set-top containing a cable modem, already gets some portion of three 38 Mbps downstream geysers — one to each device.

That’s why an important distinction about channel bonding and wideband is the delivery of that much bulk bandwidth to a single device.

It’s beyond my craziest daydreaming powers to envision a service that needs 100-plus Mbps to a single modem.

But saying “we’ll never need it” is forever an invitation to be proven wrong. People used to say they’d never need more than 330 MHz. Today’s carrying capacity for a cable system is 750 MHz, sometimes 860 MHz. I’ve yet to meet a cable operator complaining of excess capacity.


Competitively, channel bonding is a handy “back-atchya” to telco promises of 100-plus Mbps to homes. As bandwidth leapfrogs go, there was 28 Kilobits per second dial-up, then 56 Kbps dial-up. Then cable modems leapfrogged dial-up in a big way. Enter the digital subscriber line modem. The speed wars began (and continue).

As speed wars go, you want to be the guy with channel bonding and wideband in your back pocket. If deployable as promised, suppliers say, they stand to leapfrog the telephone industry’s sexiest technique: fiber to the home.

Channel bonding also helps to assuage the developing trends in overall broadband-Internet bandwidth.

Technologists are already starting to suspect, for instance, a sort of Moore’s Law brewing for broadband Internet bandwidth, where downstream speeds need to double every 18 months while pricing remains static.

If that logic holds, the cable provider who bumped up to 5 Mbps downstream in mid-2004 needs to get to 10 Mbps by the end of this year, and 20 Mbps by mid’07, and so on. (If this “law” holds, it should perhaps be dubbed “Bowick’s Law.” It was Chris Bowick, chief technical officer of Cox Communications Inc., who first uttered it — at least to me.)

Channel bonding also provides a capacity plan for the delivery of multichannel video, both standard and HDTV, over the IP (Internet protocol) side of the plant — again, to a single receiving device in the home.

From a pure speed perspective — and putting aside rights-management concerns — wideband means cable operators could send multichannel video to broadband-connected PCs; set-tops with embedded cable modems; two-way digital TVs, or any of the handheld video gizmos that glitzed with such intensity at this year’s Consumer Electronics Show.


From an equipment perspective, rest assured that the Big Names in IP technology are already years deep into wideband and channel bonding. At the recent Emerging Technologies conference, put on every January by the Society of Cable Television Engineers, both Cisco Systems Inc. and Broadcom Corp. dispatched their best brains to describe the concept. Many others are also working on it.

Cisco also used ET to host a demo of its wideband techniques, calling it the biggest thing since HFC (hybrid fiber coax) and the original DOCSIS modems. That’s a pretty tall claim, but enthusiasm that big seems to beg to be noted.

(Brass from silicon startup Broadlogic was silently on-hand for the Cisco demo, which makes it plausible that they’re in the wideband works too.)

The demo showed downstream speeds of 200 Mbps, with the capability to surpass 600 Mbps. The inclusion of statistical multiplexing techniques, similar to squishing more digital-TV channels into a single 6 MHz channel, also apply to manufacture more bandwidth, they said.


When it’s time to start planning the actual implementation of channel bonding as a means to wideband, it’s important to know this: It isn’t “found bandwidth.” You still need to locate and clear off however many 6 MHz channels as you want to bond.

Once you’ve done that, according to the supplier community, you’ll need to tweak existing cable-modem termination systems (CMTS), or add more CMTSs, plus QAMs. But again, we’re at demo stage, from the vendor side, and specification stage, at Cable Television Laboratories Inc.. Things will change between now and launch.

In the meantime, start thinking about what you could do with 100-plus Mbps to a single device in someone’s home — because soon enough, the means will be there to provide it.