TCS Daily

Revolutionizing a Revolution

By Edward J. Renehan - May 3, 2004 12:00 AM

WICKFORD, RI - Researchers have recently started the process of revolutionizing a revolution. Engineers from Lucent's Bell Labs -- allied with collaborators at the University of California/Santa Cruz, Lehigh University, and Agility Communications -- stand poised to develop a major gear-shift for the Internet. In the process they will dramatically reshape the engineering that lies behind cyberspace, thereby providing enormous (and, in the long run, indispensable) processing economies.

The fiber optic cables that form the backbone of the Internet -- state-of-the-art corridors carrying data at the speed of light -- have long been hindered by their reliance on the relatively prehistoric equipment that direct Internet traffic. The archaic switches and routers of today's Internet (Model-T transmissions clogging up the guts of late model BMWs) involve the slow, inefficient conversion of optical signals to Edisonian electrical signals, and then back again to an optical pulse. This process engages enormous capacity, and wastes literally billions in time and resources across the Internet annually.

With the volume of Internet traffic doubling every year, the routing and scheduling of packets in an efficient manner using "old" switch and router technologies becomes increasingly difficult. Virtually all technologists agree that a new approach is absolutely necessary to handle the huge amounts of traffic forecasted for the future.

The recently-announced Lucent project aims to invent this new approach, thus alleviating the Internet's ponderous conversion bottleneck. Backed by a 4-year, $12.5 million grant awarded by the Defense Advanced Research Projects Agency (DARPA) and the US Air Force Research Laboratory at Rome, NY, Lucent intends to develop a networking technology that will allow densely integrated data to remain in the optical domain throughout every stage of processing.

Lucent's formal charter is to develop the architecture, components and prototype for a wavelength-based optical packet router that can send and receive up to 100 terabits of data -- approximately the information content of the entire Library of Congress -- in just one second.

The project has a cumbersome working name: Integrated Router Interconnected Spectrally (IRIS). Lucent promises that not only will IRIS greatly cut down processing time, but it will also reduce the power consumption and overall footprint of the typical optical system.

The IRIS program aims to have a small working prototype in four years. But it could be 10 to 20 years before a fully optical high-capacity router -- a paradigm shift changing the whole nature of data communications -- is truly ready for the market.

There are obstacles to be overcome. Compatibility between the current infrastructure and IRIS, and the ability to integrate, will be a major hurdle. Even greater issues lay in wait on the technology development side. Data storage, for example, represents a particular problem. A router needs to have internal memory where data are temporarily stored before being forwarded to a new destination. This is very hard to do optically, but researchers at Bell Laboratories have already developed a small photonic integrated circuit that can keep limited amounts of optical data circulating around within a single chip. Although capacity is limited, the current circuit prototype provides a good starting point. Lucent's ultimate goal is to develop the circuit into a dense, highly integrated chip capable of handling more than 100 active optical communications components at any given moment.

"Because of this contract we will develop the world's best optical systems for the United States Air Force," says Rod Alferness, senior vice president of optical networking research at Bell Labs. "Our goal is to develop and demonstrate new optical switching technologies and architectures that will dramatically increase the capacity of packet switching systems, which is critical to future defense, government and industry communications needs."

Lucent rhetoric aside, the new DARPA initiative -- though hardly commented upon by the press -- represents the single most important recent development in the advance of the Internet. 21st century computing has now been officially launched.


TCS Daily Archives