Cable Tech

High Noon Arrives in VDSL Battle

There's no network TV coverage, no drug testing, and no French judges. But for a few chip makers, the VDSL Olympics are a heart-stopping finale to a year-old standards fight, with the results possibly deciding the winner in three standards efforts.

The Olympics are an eight-week series of tests and benchmarks conducted by impartial labs at Telcordia Technologies Inc. and British Telecommunications plc (BT) (NYSE: BTY; London: BTA). Contestants won't be going head-to-head, as in speed skating, but rather will be judged one at a time, as in figure skating. And, as with figure skating, it's a little hard to tell what the results will mean.

The battle is over which line-code scheme should be standardized for very high-speed digital subscriber line (VDSL) technology: quadrature amplitude modulation (QAM) or discrete multitone (DMT). The QAM side is being represented by Infineon Technologies AG (NYSE/Frankfurt: IFX) and Metalink Ltd. (Nasdaq: MTLK); DMT's supporters are Ikanos Communications Inc. and STMicroelectronics NV (NYSE: STM).

Broadcom Corp. (Nasdaq: BRCM), originally a QAM supporter, has switched camps to DMT, using technology acquired from Element 14 (see Broadcom to Acquire Element 14). But Broadcom doesn't have a DMT chip available and therefore isn't allowed to play.

Strangely, none of the companies interviewed said anything about going for the copper.

The three standards bodies watching the scoring are:
  • T1E1.4, determining VDSL standards for North America
  • ITU-T Study Group 15 Question 4, standardizing VDSL internationally
  • IEEE 802.3ah, a.k.a. Ethernet in the First Mile, which has chosen VDSL as a short-reach backhaul scheme but needs to smooth out the particulars

The T1E1 group is sponsoring the Olympics in order to see the ramifications of selecting QAM or DMT, with a decision likely in June.

Theoretically, the other two groups will follow T1E1's decision in the interest of setting a single common standard. But that's not guaranteed -- in fact, the ITU reportedly has heard a proposal to simply keep both VDSL standards.

Chip makers expect North American equipment makers to press for a single standard. But some carriers, particularly in Europe, say they'd like both line-code schemes to fight it out in the market, and to heck with the Olympics.

"I think the operators see this [QAM-DMT competition] and don't want it to stop," says John Egan, Infineon product marketing manager. The technologies are pressing one another to advance and driving one another's price down -- Egan expects a VDSL physical-layer chip to cost $10 by 2004.

So what are the differences between the two? For starters, folks on the DMT side are emphasizing the programmability of their technology. "Because of the way it's implemented, on DSP [digital signal processor] based engines, it's a very flexible platform," says Aidan O'Rourke, director of product marketing for Broadcom.

O'Rourke claims that will be important as DSL variants multiply. ADSL, for example, has seen Annexes A, B, and C crop up for U.S., European, and Japanese requirements; and new standards for high-speed ADSL are on the way.

But the DSP implementation is also what makes DMT more expensive in terms of both chips and engineering, says Infineon's Egan. "QAM is a much simpler technology. Its approach to VDSL is different from the DMT approach."

QAM also does have some programmability -- a feature that's been added as the technology faced increasing competition, according to Egan.

QAM seems to be favored in many pre-standard VDSL implementations -- in fact, Broadcom's still selling its QAM chip, the BCM6020. "We see a lot of pre-standard VDSL being sold into Korea, for example. But we see carriers moving to a standards-based VDSL in the back half of the year," O'Rourke says.

DMT is also making its pitch into Asia. In particular, Ikanos claims its 50-Mbit/s VDSL technology is enticing carriers such as Hanaro Telecom Inc. and KT Corp. (formerly Korea Telecom), both of which have been running ads for the service. What they've got now is 15-Mbit/s technology based on QAM from Broadcom and Infineon, Sekar says.

Here's the Olympics schedule: Each company will be the subject of a two-week test by BT and Telcordia concurrently, alternating between QAM and DMT. Infineon (QAM) volunteered to go first and completed its testing, Egan says. One of the DMT companies (we're told it's Ikanos) began its testing run on April 28. That means Metalink is next, followed by STMicro, with competition ending the first week of June. As far as we can tell, there are no plans for a pairs competition.

The results are being kept confidential until T1E1 has a look-see and determines its winner. That's expected to happen at the group's June meeting, with the IEEE and ITU decisions possibly following in July.

— Craig Matsumoto, Senior Editor, Light Reading

link 12/5/2012 | 12:05:30 AM
re: High Noon Arrives in VDSL Battle The QAM vs DMT thing happened in ADSL and DMT won out. Is there anything different about VDSL from ADSL that would point to a different outcome as the arguements all sound the same as when they occurred in ADSL?

How relevant is this? With ADSL2 coming from the low end and FTTH coming from the high end, is there any room for a VDSL solution? Any idea of how big this market is projected to be? What do we see as the driving the VDSL growth anyway... just faster internet, video, other? Who makes VDSL now?

It's an very interesting article but I feel like half the story is missing?

MrLight 12/5/2012 | 12:05:02 AM
re: High Noon Arrives in VDSL Battle link in your statement "The QAM vs DMT thing happened in ADSL and DMT won out." It was more like "The CAP (carrierless QAM) vs QAM vs DMT thing happened in ADSL and DMT won out."

In reality it was mostly about intellectual property rights - for ADSL they all are good enough. VDSL is another story, none of them will do a good enough job without some enhancements to ensure they can cover at least 90% of the loops.

I will be interesting to see if the neo-VDSL movement will solve the biggest issues with VDSL - expensive to deploy, conflict with T1 being in the same binder groups, reach and loop quality restrictions and lack of bandwidth symmetry for business.

On top of that VDSL needs to support multiple HDTV connections (at least 19Mbps, unless it is further compressed to say 12Mbps with assocaited lost of information) and support Ethernet and not ATM, support TDM over EThernet for Business, Rate adapation (RVDSL) to suit the loop, etc. (hey maybe even an emergency line service for VoIP on power failure, in case the subscriber doesn't have a cell-phone.)

MrLight :-) who has worked on this stuff, including the pre-BCM6020 technology, and who would like to see something better than a QAM or DMT approach based on what has been learned about ADSL. But really the end game is fiber, so it doesn't realy matter as long a whatever is picked is good enough until then.

P.S.A point about QAM is it has a steep BER falloff so when bit error burst take place you are on a slippery slope to getting zero throughput. Where DMT on the other hand tolerates bit error rates bursts better, reslulting in a more graceful degradation.
Sign In