Skip to main content

Debating the Next Move for DOCSIS

2.0b or not 2.0b — that appears to be the question for CableLabs and cable-technology vendors.

At issue is a proposed addition to the Data Over Cable Service Interface Specification family dubbed DOCSIS 2.0b. Considered by many as a first step toward bonding together two or more data channels to boost cable’s data pipe, DOCSIS 2.0b isn’t as yet a formal development project for CableLabs, the industry’s technology consortium.

Actually, it might never be a formal development project. Concerns that adding the specification would divert attention from the developing products based on the draft DOCSIS 3.0 scheme — which offers a greater ability to bond data channels — has many arguing that CableLabs should not create a separate DOCSIS 2.0b development track.


But even if CableLabs opts to scrap DOCSIS 2.0b as a specification, it appears it will survive in vendor products. That’s being driven by the fact that DOCSIS 3.0 cable modems or network gear won’t be available until at least 2008. Meanwhile, there is mounting pressure from telco players — including Verizon Communications Inc., with its FiOS fiber-optic service — to boost data throughputs north of 100 Megabits per second to individual users.

Reports that CableLabs would drop the DOCSIS 2.0b idea gained fuel at last month’s Society of Cable and Telecommunications Engineers’ Cable-Tec Expo show.

During a panel session, Comcast chief technology officer Dave Fellows appeared to throw cold water on the idea, saying DOCSIS 3.0 was the priority.

Headlines afterward proclaimed DOCSIS 2.0b was dead. But that wasn’t accurate, according to Marwan Fawaz, departing chief technology officer at Adelphia Communications and soon-to-be CTO at Charter Communications Inc.


Fawaz, who also took part in the panel session and is a member of the DOCSIS business team at CableLabs, said the cable consortium will soon issue a plan for dealing with pre-DOCSIS 3.0 technology. That isn’t necessarily a death sentence for DOCSIS 2.0b.

“I want to make sure that people [aren’t] reading too much into it, that we are killing 2.0b,” he said. “We — the MSOs and CableLabs — are working very hard right now to clarify that. There will be some communications shortly that will come out from CableLabs that will be coordinated with the MSOs about what are the processes from a product development related to any pre 4.1 product.”

The goal for CableLabs and the operators is to encourage interim technologies, as long as they are compatible with older DOCSIS specifications such as DOCSIS 2.0 and can be upgraded to work with future DOCSIS 3.0 schemes, he said.

“The issue is where do you focus your resources as an industry, and we want all of our suppliers to focus on [DOCSIS] 3.0,” Fawaz said. “CableLabs is working very hard to get the specs out, and we are within weeks of that. So as soon as those are out, we want full-speed ahead — get us 3.0 products.”


While there is no formal guideline, DOCSIS 2.0b technology essentially would allow cable systems to two or possibly more downstream data channels in the near term.

That would provide less of a bandwidth boost compared to the minimum four channels DOCSIS 3.0 technology could weld together.

But even two channels bonded at 38 Megabits per second each could enable 76 Mbps in total bandwidth to parcel out to throughput-hungry customers.

Chip-making giant Broadcom Corp. has already come up with a chip product adding channel bonding to a DOCSIS-2.0 silicon — a la DOCSIS 2.0b — that has been distributed in samples to box makers for four months now.

The BCM 3381 chip provides three bonded DOCSIS-2.0 channels downstream, totaling 100 Mbps of throughput to a subscriber, but it maintains the lone upstream channel used in DOCSIS systems now, according to Jay Kirchoff, senior director of marketing for broadband communication business unit.

Broadcom’s chip design borrows from the draft DOCSIS 3.0 packet sorting strategy. It reassembles packets coming in over multiple data channels based on each packet’s identification header, which is a sort of digital shipping label containing information about the type, source and destination of the packet.

Operators will probably start deploying hardware and offering services using the channel bonding in the fourth quarter, according to Kirchoff.

“It was a product that was really born out of demand from the marketplace much sooner than there was ever a spec on it,” he said, adding that 90% of the interest comes from outside the United States by cable operators competing against much more advanced fiber optic and very-high-bitrate digital subscriber line (VDSL) services.

That said, “we’re seeing some interest in the U.S. where FiOS or VDSL technologies are actually being delivered,” Kirchoff said. “And so we think it’s real.”


Cisco Systems Inc. also has developed a pre-DOCSIS 3.0 channel bonding technology it calls Wideband. Unveiled last month, Wideband follows the same general path as DOCSIS 2.0b, bonding three downstream channels to offer 100 Mbps service.

As with Broadcom, Cisco is seeing much of the interest in Wideband coming from oversees cable operators.

“We’re definitely getting enquiries from the North American customers as well,” said Dave Brown, marketing manager for Cisco’s cable products and systems. “They are playing their cards close to the vest in terms of what they actually want to do in terms of deployments, but they have expressed a lot of interest in getting into the labs and testing it out. They certainly want to have it as a card that they want to play, and may be working on how exactly they are going to do that and where.”

A big selling point for the Broadcom channel-bonding chip is that it can be certified and run inside a DOCSIS 2.0 modem, and it might also be upgradeable to power a DOCSIS 3.0 modem.

“It doesn’t have all of the [DOCSIS] 3.0 features in it as such, so we can’t call it a 3.0 modem,” Kirchoff noted. “But what I’m hearing is our customers are expecting that they will be using this same technology when they do get 3.0.”

Cisco, which claims the top-selling CMTS unit on the market as well as modems through its Scientific Atlanta and Linksys subsidiaries, also could support DOCSIS 2.0b if it became a specification. Even if that falls by the wayside, Cisco will continue with the Wideband product line, Brown said.


“It seems clear to us that it is going to happen one way or another. It’s going to have some impact, but to us it’s not going to impact our DOCSIS 3.0 schedules or anything,” Brown said.

It is encouraging that early channel-bonding products such as Broadcom and Cisco’s are following in part on the draft DOCSIS 3.0 specification, Fawaz said. But he also points out that DOCSIS 3.0’s importance is also in adding upstream channel bonding — something neither Broadcom nor Cisco’s product will do as yet.

“I think that is important, and I think that is one reason we are pushing so hard to get 3.0 out there,” Fawaz said. If there are “interim solutions with channel bonding that look pragmatic, each company may decide to deploy it until 3.0 is ready. But that remains to be seen.”