DSL pioneer wins top telecommunications honor

By David Orenstein

In January 1993, Stanford electrical engineering Professor John Cioffi had a startup with barely enough funding to build a prototype high-speed modem. With more confidence than cash, he and a small team of students and friends managed to fly to New Jersey to enter the gadget into a competition the American National Standards Institute was holding to find a new standard for data communications over home phone lines. Cioffi walked into that competition a dreamer and walked out the primary author of what has become the international standard for digital subscriber line (DSL), a technology for broadband Internet connection. DSL now has more than 200 million customers worldwide.

"Nobody at the big telecommunications companies believed our system would work," he recalls. "But we went head to head in that competition. We got it working pretty close to theory and beat them by a large margin. It was like Palo Alto High School taking on the New England Patriots in the Super Bowl and winning."

Today, of course, Cioffi is no longer the underdog. He holds more than 70 patents (about 30 are assigned to Stanford) and the Hitachi America Professorship in the School of Engineering. He has founded a couple of companies and serves on several boards. But one of Cioffi's most meaningful distinctions is also his newest. He is the 2006 Marconi Fellow. Bestowed by the foundation named for the inventor of the radio, the fellowship is widely considered the most prestigious in telecommunications. Cioffi will receive the award at a gala Oct. 12 at the Menlo Circus Club in Atherton, Calif.

The award gala will follow a daylong symposium at the Computer History Museum in Mountain View, Calif., where Cioffi and other luminaries will discuss the future of the communications industry. There Cioffi will share his vision of how consumers may soon enjoy data transfer rates as fast as a gigabit per second through their phone lines. That's much faster than today's 1.5 megabit-per-second DSL lines. Such speeds would enable services such as delivery of full-length, high-definition movies over the Internet on demand.

Building on Bell

Enabling ever greater communications is the bottom line for Cioffi. He fervently believes that people will use as much bandwidth as engineers can provide and that demand for data transmission will always increase. Ever since he was an engineer at Bell Labs in the 1970s, Cioffi has been inspired by that basic principle. It has driven him repeatedly to find ingenious ways to wring capacity out of the temperamental twisted pair of copper wires known as the home phone line.

For almost 100 years after Alexander Graham Bell invented the telephone, nobody needed or cared to use it for much more than transmitting voltages in the narrow band of frequencies that correspond to the human voice. In theory, these wires can carry many more frequencies and therefore more information. In practice, the vexing physical realities of phone lines made the realization of DSLand a newer round of improvements Cioffi has developedvery tricky.

"Phone lines are big antennas that radiate into one another," Cioffi says. "They are their own worst enemies when they are all bundled together. Any kind of [electromagnetic] noise from AM radios, fluorescent lights or your vacuum cleaner can get into these things and cause problems."

The solution Cioffi began to imagine at Bell Labs and finally produced at Stanford in conjunction with students was to transmit data between modems (one at the phone company and one in the home) over 256 separate 4-kilobit channels on the phone line. Each modem would be equipped with a sophisticated adaptive learning algorithm that could compensate for interference or other glitches affecting any of the channels. If a channel, for example, was experiencing interference, the modems would notice that and stop transferring data on it. The traffic would be switched to other less troubled channels. The technique, called "bit swapping," was the essential technology in the prototype modem Cioffi's first startup, Amati, brought to the standards competition in 1993.

"That is the heart of the DSL modems that are in use today," Cioffi says.

Continuous improvement

Keeping in mind that people would want ever more bandwidth, Cioffi kept working throughout the 1990s on more sophisticated management systems that could deliver far greater data rates through standard phone lines. The result is a technology he calls DSM, or dynamic spectrum management.

DSM adds many more channels than DSL, particularly at higher frequencies, because it manages interference much more aggressively than bit swapping does. Higher frequencies are the most prone to interference, so they had been off-limits in the earlier versions of DSL. But DSM combats interference by anticipating it and encoding signals that minimize it. DSM also can identify troublemaking channels and shut them off, like tow trucks removing disabled cars from highway lanes.

DSM also does away with the "asymmetry" of the most common flavor of DSL, called asymmetric DSL. Asymmetry means that 90 percent of the channels are reserved for downloads and only 10 percent are used for uploads. This design reflected a consensus among engineers in the '80s that people would spend most of their time downloading things, Cioffi says. In an era of file sharing and video conferencing, these assumptions may seem more like an archaic constraint than a necessary design.

Cioffi will be on leave this year from Stanford to help build a new startup, Adaptive Spectrum and Signal Alignment (ASSIA) Inc., a company named after his wife, Assia, and dedicated to commercializing his DSM research. But as Cioffi embarks on that leave he is hardly forgetting Stanford. He is leaving the $100,000 prize that comes with his Marconi fellowship to his research group. The money will help fund research into optimizing the management of communications transmitted wirelessly and by other means.

Cioffi's relentless desire to improve global communications is ultimately what made him a Marconi Fellow. "The annual Marconi Prize recognizes those living scientists who, like Guglielmo Marconi, share the determination that advances in communications and information technology be directed to the social, economic and cultural improvement of all humanity," says John Jay Iselin, president of the Marconi Foundation. "John Cioffi's remarkable career personifies this spirit."

David Orenstein is the communications and public relations manager at the Stanford School of Engineering.