DSL Prime: Debunking the Hype

Even we have fallen for some of these, but not for most.

by Dave Burstein of DSL Prime and Future of TV and the Web Video Summit [January 8, 2008]

John Papandriopoulos Will Not Make DSL 100 Times faster
Solid work on improving DSM
Headlines around the world wildly overstated his accomplishments: "ADSL broadband connections up to 100 times faster" and "Aussie Claims Copper Broadband now 200x Faster" at the usually reliable Slashdot. Although Papandriopoulos, still in school, didn't revolutionize the field, his SCALE algorithm is respected by experts. One professor writes "SCALE is an efficient algorithm for a computer workstation to compute the relative spectra of the users, along with a message-passing algorithm from the workstation to users over a management plane. Since DSM can improve performance from 30 percent to 100 percent +, any contribution to the field is a step forward.

John Cioffi and others have been predicting DSL speeds will rise to perhaps 250 megabits over very short loops (hundreds of feet). That's a substantial improvement over the 100 megabits in each direction achieved by millions in Japan, Korea, and soon New York. The improvement will mostly come by reducing noise (crosstalk) from other lines. (See below, Alcatel Confirms DSM Improvements.)

Unfortunately, the calculations necessary for that optimization stretch the state of the art. Cioffi in Berlin told me he believes the coming 65 nanometer chips should probably have enough processing power to meet most of the need, but improved efficiency is very welcome. Here's how Papandriopoulos describes his work in an engineering paper [.pdf]:

A novel technique involving a series of convex relaxations to derive an algorithm called SCALE (Successive Convex Approximation for Low-complExity). We show through numerical simulation that SCALE performs significantly better than IWF, and with comparable complexity. An important feature of SCALE is that it may be distributed with the help of a Spectrum Management Center (SMC). The resulting method may be viewed as a distributed computation across the DSL network, in contrast to the centralized OSB and ISB schemes. Importantly, we outline how the overhead associated with this approach can be managed, and show that it degrades gracefully to the same performance as that attained by IWF when no inter-user communication is available. Our final contribution involves a fresh look at IWF. We derive a new algorithm called SCAWF (Successive Convex Approximation for Water-Filling) that simplifies existing IWF approaches and enjoys low complexity implementation.

Mistakes I've discovered
I was wrong
Eli Noam calls me out for being enthusiastic about European fiber builds that haven't happened yet. He points out that Verizon alone has installed more fiber to the home than all of Europe (possibly excluding Fastweb), so I should be careful talking about how well they are doing in Europe. I'm confident that Free, Neuf and FT will quickly expand fiber in France, KPN in Holland, and the cities of Geneva and Cologne.

The KPN Holland fiber network will be Active Ethernet, not GPON, Ron van Os of Genexis writes. Bravo to the Dutch regulator for inspiring KPN to choose point to point fiber. Incumbent carriers usually go GPON, which is generally impractical to share. I should have checked further.

David Goldie tells me dark fiber or similar is available to 90 percent of the U.K. I had told Ed Richards only 70 percent could be covered, based on a comment from an ISP CTO. Goldie's Carphone Warehouse announced they were buying dark fiber across the nation.

Mathias Kurth of Germany was apparently right that unbundling from the street cabinet will find some takers. He predicted at least one competitor would jump in, and some have applied. It's never worked anywhere else.

Quoted

• "provide adequate capacity." DSL Reports on the right way to solve the congestion problem.
  
  
• "CTIA says wireless carriers added 14,747 cell sites in 1H07 which still doesn't explain why call quality continues to suck on pretty much every network." Om Malik, pointing out what everyone in the U.S. seems to know except the FCC.
  
  
• "Any company that can make a mobile phone with no buttons, no picture messaging, slow web access and no video capture into the most desirable phone on the planet can easily make tablets popular." Crave at CNET, who believes the rumors are true that Apple will release a "tablet PC", probably an iPhone clone with a big screen.
  
  
• "My closed network is more open than your closed network" David Bujnowski, on the real message of the Verizon and AT&T announcements

Copyright 2008 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.