CONTACT: David F. Salisbury, News Service (650) 725-1944;
Stanford technology to deliver television, high-speed computer access over existing phone lines
Telephone lines. They're not just for talking anymore.
That slender copper wire that was originally designed to carry voice conversations has been the biggest bottleneck in the information superhighway. Despite the doubling and tripling of the speed of the voice modems that currently provide computer communications, phone lines have seriously restricted the usefulness to home users of the burgeoning Internet.
But the telephone companies, which are engaged in a multi-billion dollar race with the cable television industry to provide high-speed Internet access to the home, are ready to begin piping megabit streams of computer data, as well as high-quality television programming, down that same slender copper wire.
And much of what they are doing is based on the work of John M. Cioffi, an associate professor of electrical engineering at Stanford, who pioneered a technology that wrings enough added capacity from existing phone lines to deliver computer access that is 30 to 50 times faster than current rates, with room to spare for several video channels and a number of normal voice lines.
The technology has proven so successful that Cioffi was able to sell off a start-up company he founded to Texas Instruments for $450 million, and Stanford's Office of Technology Licensing says his invention is likely to become a major new source of income for the university's resarech and educational activities. After coming up with the method as part of a federal award, Cioffi found ways to develop it into a commercial product despite initial industry indifference. He even beat AT&T in a national standards competition.
In 1993, the technology, called discrete multi-tone (DMT), was written into a new national standard for high-speed computer communication over the phone system, called asymmetric digital subscriber lines (ADSL). But only in the last few months have the major telecommunications and data communications companies lined up behind it. Only last month a new standard for a lower-cost approach called G.ADSL Lite backed by a powerful consortium consisting of Microsoft, Intel, Compaq, GTE, and four of the five Bell telephone companies was approved, based on the Stanford technology.
"It took a while, but I think everyone is now on the same page," Cioffi says.
In 1986, when Cioffi decided to work on the problem of sending digital television signals over a phone line, there was little indication that it would turn out to be so successful. He had won a Presidential Young Investigator award from the National Science Foundation, which required that he pick a research project with matching industry support.
"Joe Lechleider at Bellcore [the research laboratory for the regional Bell telephone companies] talked me into looking into this idea. He saw it as a way of providing video-on-demand over the phone lines," Cioffi says. "I liked the idea. We knew that copper had a tremendous amount of dormant capacity, but no one knew how to tap it. With a billion loops of copper wire worldwide, that is a lot of capacity going to waste."
It took Cioffi and his students three years to come up with a reasonable solution to the problem. It was a "smart" technology that tests the quality of the telephone line 10 to 100 times a second and finds the optimal frequencies for sending information at rates 30 to 50 times faster than current modem technology.
"No two phone lines are exactly alike," he says. "They're different lengths, some have radio-signal noise, impulse noise, crosstalk noise, et cetera. DMT uses intelligent modems that learn about the line, tell each other about it and change their signals accordingly."
Then Lechleider retired and the Bellcore funding dried up. When Cioffi approached other telecommunications companies, he didn't find much interest. The phone companies were investing billions of dollars to replace their copper lines with optical fiber, which has a much greater capacity. So they weren't interested in a method to increase the capacity of copper lines, Cioffi says.
About that time, Cioffi saw a brochure that he says "changed my life." It described a program sponsored by University Technology Transfer International (UTTI) for transferring university technology into start-up companies. "On a lark I submitted a proposal. It was one of two that received funding," he says. His lab was slated to receive $250,000. But UTTI went bankrupt in 1990 so he only received about half that amount. Nonetheless, it gave the group "enough momentum and enough contacts to continue on our own with careful spending," he says.
By then, Cioffi had disclosed the details of his research to Stanford's Office of Technology Licensing (OTL), and the university applied for four patents on the DMT method.
In 1991, another bit of serendipity came into play. A small Israeli company named ECI had contracted with Deutsche Bundespost (the German postal and telephone service) to install phone lines in the former East Germany. But the company had overestimated the capabilities of conventional technology and so had underquoted their bid. ECI had heard of the Stanford technology and decided to take a chance on it.
With the ECI contract in hand, Cioffi decided to start a company: "Starting a company was a last resort. I only did it when it became clear that there was no industry interest in commercializing the technology." He took a two-year leave of absence from Stanford and named his company Amati, after the 18th-century Italian violin maker who taught Stradivarius his craft. Jackie Chow, one of his students who had worked on the original technology, joined him at the new company. A second student, Peter Chow (no relation to Jackie), joined a year later when he completed his degree.
Stanford granted Amati an exclusive license for the DMT technology, and OTL set its royalties at a lower-than-normal rate in return for an equity share in the company.
The products Amati developed for ECI were highly successful. "They actually exceeded the requirements. The Bundespost was very happy," Cioffi says.
Despite this success, the telephone companies were still investing in optical fiber technology, not copper. Not until 1993 did Amati have a chance to demonstrate the capabilities of its technology. The American National Standards Institute (ANSI) had scheduled a competition to establish a new standard for computer communications over the telephone system. The odds-on favorite was AT&T's CAP technology, which operated at 1.5 megabits per second. CAP treated all telephone lines as equal, so its performance varied substantially with line quality. DMT, on the other hand, was capable of 6 megabits per second, and its performance was less affected by variations in line quality.
"Everyone told me that we couldn't win, because we were up against AT&T. We decided to enter anyway," Cioffi says. When the tests were finished, DMT had performed as predicted and the verdict was unanimous: It was the new standard. Despite this confirmation, it has taken the highly competitive industry five years to line up behind the technology.
Last November, Texas Instruments Inc. purchased Amati for $450 million. Stanford realized nearly $8 million from cashing out the equity that it held in the company. In addition, OTL has received about $500,000 in royalties.
"We hope that this [technology] will bring in between $20 million and $30 million more in royalty income over time, which would make it the second or third largest revenue generator that we have had to date," says Katharine Ku, director of OTL.
On April 8, OTL and Texas Instruments renegotiated the terms of the Amati's exclusive license to enhance TI's ability to sublicense DMT patents. For any patent to be part of a standard, standard bodies like ANSI require that licenses to the patent be made available on a fair, reasonable and non-discriminatory basis, explains OTL licensing associate Linda Chao.
The new agreement also gives TI the option to take out non-exclusive licenses on improvements to the technology that Cioffi and his students continue to develop. Other companies interested in licensing these improvements will do so directly from Stanford.
One recent sign of movement on the digital-subscriber-line front was an announcement made on April 20 by US West. The phone company revealed its plans to offer television programming and high-speed Internet access to 400,000 customers in Phoenix by the end of the year at a cost comparable to the monthly fees charged by the local cable company. US West says it will use an alternative technology to ADSL called "variable digital subscriber lines" that can deliver more information over shorter stretches of copper wire.
By David F. Salisbury