DSL Prime News: The Inside Source

Brand new DSL technology could—really—provide subscribers 100 Mbps in each direction. Verizon tries to please the FCC by planning a buildout, while SBC refuses to build as it lobbies the states for total domination.

by Dave Burstein DSL Prime [April 2, 2003]

"100 Mbps everywhere! Both directions."
—Professor John Cioffi of Stanford

Nobody believed Joe Lechleider when he calculated a copper pair could carry megabytes, but today millions have 10 and 12 Mbps service. Few today believe John Cioffi's results, but his previous achievement, DMT ADSL, is now serving over 30 million users around the world. Early field tests confirm his lab results, although it's still a long way to proven in the real world.

Our April 1 issue, "Fool's Day" issue will come out later this month; suggestions welcome.

Cioffi: DSM doubles VDSL, speed at distance
Great early results
The Japanese and Koreans are proving customers want 5 Mbps and more already, including a major shift to VDSL. DSM aims to double speeds, and allows a long loop particular performance improvements that may prove crucial.

Even I can understand the principle. Just as a 50,000 watt radio station reaches further than a 100 watt, increasing power to a DSL line gives greater reach/rate, along with increased interference. The industry, led by T1E1.4, has worked for a decade to set sensible limits designed to keep that interference manageable. But Cioffi's Dynamic Spectrum Management works with the actual bundle of wires, not a theoretical model. By actively protecting lines from interference by testing and adjusting, the DSLAM can selectively upgrade performance.

The results Cioffi reports are dramatic, including increasing the rate of VDSL from about 50 Mbps to about 100 Mbps in many situations. More directly valuable is the improvement by waterfilling of remote performance—3 to 10 times better at distance, 100 Kbps increased to 300 Kbps or even 2 Mbps. (In a two line example, one short and one long, the short line at 2 Mbps moves to higher frequencies without loss of performance. That allows the long line to operate efficiently on lower frequencies, with better reach, and run much faster.)

Cioffi's presentation, with credit to grauduate students Taek Chung, Jungwon Lee, Dimitris Toumbakaris, M. Mohseni, Bin Lee, & V. Pouramod, is here [.ppt]. This is exciting, partially proven work, exceedingly controversial. In Cioffi's DSM models, non-ADSL symmetric service, whether G.shdsl or CAP VDSL, is not included. That's contrary to the plans of many telcos and regulators. These are hard decisions, with still only limited information.

How Verizon's building to 80 percent
Alcatel DSLAMs in COs and remotes
BellSouth proved that adding a small DSLAM next to existing remotes works and is an inexpensive way to add coverage. Verizon plans to cover over 3,000 this year, primarily with Alcatel equipment. Some will be the new small DSLAMs, others may be upgraded with Litespan linecards. Alcatel will also be lead vendor on 1,000 central offices, for a total of over $100 million in equipment. Catena offers a lowcost upgrade to older remotes that Verizon has used in some cases (Lucent SLC-5's). AFC's newer units simply require a line-card upgrade (one analyst estimated that will be over $7 million in business for AFC.) A group of vendors (Siemens, ECI, Adtran) have been pitching Verizon on using their equipment as well, but none have been announced as breaking through.

An internal estimate from Verizon is a cost of $500 million for the DSL build, which is about three times the direct installed equipment cost. Some of the $500 million will apply to overhead, some to adding more facilities in the already served markets. Many details are still to come in what was a rushed announcement. Chris Stern in the Washington Post did a great article about the millions being spent on advertising and lobbying to influence the 30 or so people at the FCC involved in writing the final triennial review. Verizon's early announcement presumably is part of that effort, although I have confirmed with several sources that Verizon is actually doing that build. SBC has angered potential friends in D.C. by refusing to build despite being given a virtual DSL monopoly, and Verizon is trying to do damage control. SBC, like Verizon, will build out to at least 80-85 percent because the business case is overwhelming. They are refusing to say so because it's a convenient political lever.

The follow-up to this item will be Next Step: from 80 percent to 95 percent. Solvable problem at an affordable price.

Copyright 2003 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.

Go to page 2: DSL Numbers >