Wednesday, October 1, 2003

What would it cost to bring superfast Internet access to almost every home and business in the United States? And how would it affect the flow of information to offer such access -- at 100 times the speed of most DSL connections?

A group of universities and research laboratories led by Carnegie Mellon University hopes to find out. The institutions are starting an ambitious five-year project that will examine what it would take to build and sustain a fiber-optic network that would reach 100 million locations throughout the nation. "We want to take a fresh look at the architecture of the Internet," said Hui Zhang, an associate professor of computer science at Carnegie Mellon who is the project's principal investigator.

The National Science Foundation pledged $7.5-million over five years to the project, which is being called "100 Megabits for 100 Million Homes." It seeks to develop and test the feasibility of an all-new, high-speed glass-fiber telecommunications network that would improve the reliability and speed of Internet access.

"There has been a misconception that, with the success of the Internet, all the research into networking architecture has been done," says Mr. Zhang. "But we really need to make the network much more robust and dependable, and higher-speed."

To design a more efficient and manageable communications network, project leaders must decide how best to use fiber-optic technology and how best to approach broader concerns, such as the economic and social impacts of a high-speed network, according to Mr. Zhang. The research will be conducted by an interdisciplinary team of computer scientists, engineers, and economists at Carnegie Mellon, Rice and Stanford Universities, the University of California at Berkeley, Internet2, and a collection of supercomputing laboratories and for-profit research centers.

The first two years of work on the project will be devoted to designing and modeling potential glass-fiber networks, according to Mr. Zhang. After that, researchers will test small-scale prototypes to determine whether telecommunications providers could use the same blueprints to form a nationwide network. The workload will be divided among participating institutions, and the results will be analyzed at annual conferences.

The $7.5-million grant is "not a lot of money," Mr. Zhang said. "We're very grateful for the support, but we have to view it as seed money because we need much more to sustain this effort, even with modest prototyping and deployment."

"Since we've used copper-based telecommunications networking for so long, this is the first time in 100 years of science that we've seriously re-evaluated network architecture," Mr. Zhang said. "The Internet has really showcased the potential for better communications, but it has a lot of limitations."