Compaq Computer, Intel and Microsoft are leading an industry consortium in the development of a consumer standard for high-speed Internet access through telephone lines. And although it may not seem that way from what I have to say below, I'm barracking for them.
Trouble is, the consortium is doomed, or at least it's showing all the signs.
The consortium is mired in standards bodies and their acronyms, none of which I've used yet in this column. It has decided to rely on technology decisions made by telephone monopolies, back when said monopolies thought video on demand, not Internet access, was the next big thing. Worse, the consortium is now trying to sweep its various problems under various rugs.
Compaq, Intel and Microsoft, otherwise known as CIM, have attracted more than 70 members to their Universal ADSL Working Group (UAWG). UAWG is rushing to specify a plug-and-play version of Asymmetric Digital Subscriber Line (ADSL), which it hopes will receive an initial blessing as G.Lite from the International Telecommunications Union (ITU) at its October meeting in Geneva.
UAWG is determined that G.Lite will be based on American National Standards Institute (ANSI) T1.413 ADSL using Discrete Multitone Technology (DMT) modulation. And as if ANSI and ITU weren't enough, UAWG is supported by yet another consortium, the more than 300 members of the ADSL Forum (http://www.adsl.com).
UAWG is right that today's analog dial-up telephone modems are not adequate for Internet access from homes. UAWG is right that packetising phone lines is the leading alternative to dial-up modems. UAWG is right that an international standard would greatly accelerate the proliferation of Digital Subscriber Line (DSL) technology.
UAWG is wrong, however, in assuming that ADSL can simply be slowed down to reach far into many distant homes without rewiring them. I warned about this right after UAWG was formed.
Back then I argued that because Nortel and Paradyne have both recently introduced non-ADSL DSLs, there must be something wrong with DMT. And now sources within UAWG are telling me that I'm starting to look right.
But UAWG is not admitting the problems it's having with ADSL. Go to its Web pages (http://www.uawg.org) and look for any bad news at all. Nothing but glad tidings and the claim that G.Lite will be the preferred modem standard, not by Christmas 1998 or 1999, but by 2000. Slip.
Efforts to hurry G.Lite through ITU are running into reality. Conscientious engineers are now trying to insist that the modelling and testing of actual home wiring be made part of the ITU specs. The leadership, however, is working hard to hold to schedule and maintain appearances.
For example, you will not read on UAWG's Web pages that DMT's developer, Stanford Professor John Cioffi, is admitting DMT cannot smoothly handle telephones going on and off hook on shared G.Lite lines.
Others complain that it is not realistic to model home wiring as a simple star of Category 3 copper wiring.
It's just that because of DMT's high frequencies and power, G.Lite will require in-line phone filters; or new telephones; or the full-rate ADSL splitters that UAWG was aiming to eliminate with G.Lite; or ugly deployment restrictions to keep G.Lite from interfering with plain old telephone service, T1, ISDN and itself.
I may yet be proven wrong about ADSL being broken beyond UAWG's repair, but I'm positive that UAWG's current telephone company politics and propaganda will, if not fixed, prove fatal. UAWG proceedings should be opened up so that we all can track its problems and progress. Now, out in front of all this, the few remaining Baby Bells [US telcos] are talking quite a bit about their upcoming launches of DSL service -- they are feeling the competition of cable modem companies. And it sounds like ISDN all over again.
If I were cynical, I'd suggest that telopolies want UAWG to fail. The longer they can delay DSL deployment, the more time they have to prepare for their retirements by gouging us for T1 services.
Before we see a universal DSL standard, I think we're going to see chips that automatically sense and support various combinations of V.90 56K-bits-per-second modems, DMT and Carrierless Amplitute and Phase Modulation ADSL, G.Lite, Paradyne's Multiple Virtual Line, Nortel's EtherLoop, and others. These chips will be valuable in absorbing differences among emerging standards and in adapting to wiring differences.
The hard part will be getting the various DSLs deployed in, under, around, and through the US telephone monopolies and their public utilities commissions.
(Internet pundit Bob Metcalfe invented Ethernet and founded 3Com in 1979. Send e-mail to [email protected])
