DSL Prime: TI Boosting 100 Mbps

U.S. users are ready to switch to broadband, just as vendors and ISPs begin preparations to bring 100 Mbps service to the U.S. On the other hand, Q2 subscriber adds in the U.S. are likely to disappoint.

by Dave Burstein DSL Prime [June 22, 2004]

"People on the other end of the line didn't know I was using VoIP until I told them." —Mike Langford in the San Jose Merc.

"Will Texas Instruments 100 megabit VDSL displace most of fiber to the home?" asked J., an analyst. The question embarrassed me, because J. is a regular reader and apparently I haven't made clear that 100 Mbps VDSL is coming. Since the demo in March at FNF, it's been obvious that VDSL + fiber to the pedestal/basement is a viable alternative to fiber all the way home. KT has already shifted most planning from fiber to the home.

I'm about to fly to SUPERCOMM, and couldn't clear my backlog of stories to write. Still to come: Mike Capellas's side deal with the Bells (DSL via New Edge), UTStarcom's breakthrough in Europe at Tiscali, Broadcom and other "better than telco" calls, Ireland seeing some advanced competition, South Africa's overpriced but poor service, why Masayoshi Son should come to America (complete with business plan for him and/or AT&T), and the rural telco story begun below in the wall street section. Also an appreciation of some rural telcos that offer DSL to 100 percent of their customers.

Say hello to the round fellow with a beard at SUPERCOMM.

TI's Promises
100 Mbps symmetric, full ADSL2+ included, lower power, DSM, bonding, 4 years away
TI is exciting the world about 100 Mbps for everyone, promising inexpensive chips and systems just a few years away. The high speed requires short loops, probably about 1000 feet. That's great for 20 million Americans in apartments, much of East Asia, London, Paris, and Milan. They believe the same chip can support a VDSL2 mode for 3,000 feet, as well as standard ADSL2+ for longer loops. Not specified yet, but likely to come, is a 5,000 to 6,000 foot symmetric mode, possibly the 10/10 Mbps one Bell plans.

These speeds are not science fiction, with 100/30 Mbps on short loops heading to customers in Japan and 100/100 Mbps promised for next year. DSM performance is part of the design, no surprise since Cioffi has worked with TI since they bought Amati. DSM/MIMO is a major industry trend, and I'm seeing a startup's MIMO/DSM results privately in Chicago.

Bonding is an easy way to double those speeds, as ADTRAN, George Hawley, Actelis, Hatteras, and others have demonstrated, and the line count drop means there's enough fiber in many places. TI points to an extra 20 percent+ performance boost from vector bonding.

TI brings to the table the clout of a respected major company, and a pr staff second to none in telecom. Don't underestimate how that can change minds, especially when the major objection to moving ahead is inertia. The top four DSL stories on Google were all about TI's 100, and Slashdot picked it up as well as all the usual media. TI people and publications were clear production chips are easily two years away, with equipment even further behind.

They announced only a few days after BT, learning from the Asians, told us entirely new networks are the way to go. Suddenly, slow ADSL is looking very tired. That means the most interesting question is how far to bring the fiber: FTTH (theoretically unlimited), FTTC/1,000 feet (100 Mbps), 3,000 cabinets (20 to 40 Mbps) feet or 5,000 foot (10 Mbps symmetric, 20 Mbps ADSL). My guess if FTTH in new builds and where topology makes it cheap, FTTC/VDSL many other places. Others believe the costs are too high, and are sticking to 3,000 and 5,000-foot goals.

The new holy grail—ultra reliable, line-powered pole boxes
100 Megabits+ travels 300 meters/1000 feet or perhaps more, which will require a basement, field cabinet, or pole unit served by fiber. The equipment cost is only a third what fiber costs today, although that will drop. The construction in many topologies is dramatically less, often with no trenching required. Even underground, the fiber can often go in existing conduit. The maintenance and installation expense of so many field units must not be underestimated, which is why few have committed so far.

KT is already installing the 24 port Tellion TelePrism, with data rates of 50/17 Mbps soon to increase. The TelePrism is 300mm (W) x 240mm (D) x 460mm (H) and weighs less than 12 kilos. It's designed for indoor and outdoor use, but does require utility power.

TI believes they can reduce the power required by 50 percent, which will allow field boxes to be powered over spare telephone pairs. Line-powered remotes are driving down ADSL costs, and the techniques to reduce power consumption are well known. Tripath has shown the way, with a new generation of low power line drivers (Class D), which are already in use in some Alcatel DSLAMs; TI and others expect to deliver similar over the next year.

Convincing companies like SBC to use VDSL instead of PON will be much easier if the equipment makers can ensure repair costs will be low. That requires extraordinary engineering for reliability (failsafe wire connections, remote diagnostics and repair, temperature extreme hardening) rather than fancy features. Last DSL Prime reported how much Matt Bross of BT expects of a vendor; proving rock-solid quality will be the best way to sell to Bross and his peers.

Copyright 2004 Dave Burstein.
The DSL Prime Newsletter is reprinted with permission.

"The power of the printing press belongs solely to those who own the presses"
—A.J. Leibling

The Internet is the cheapest printing press ever invented.