NSF seeks broad Internet research agenda

One of the Internet's pioneers will explore the feasibility of designing a new Internet. David Clark, a senior research scientist at the Massachusetts Institute of Technology and former chairman of the National Research Council's Computer Science and Telecommunications Board, has begun a yearlong preliminary study to determine how computer scientists could create a new Internet architecture that would gain global acceptance.

The study is a first step in re-engineering the Internet, said Clark, who has led efforts to develop the Internet since the mid-1970s. Overhauling it could have a profound impact on society by eliminating many of the security problems that plague the Internet today. Agencies, such as the National Archives and Records Administration, are already benefiting from an Internet-focused research agenda at the National Science Foundation, which is funding Clark's study.

NSF's agenda includes a proposal for creating a new office devoted to cyber infrastructure. But the agency's tight budget could thwart such projects, some observers say.

The agency that created the earliest version of the Internet, the Defense Advanced Research Projects Agency, used to focus on computer science research. But lately it has shifted many of its resources to advanced weapons systems development. NSF has taken the slack by shouldering more responsibility for basic network research, said Guru Parulkar, a program director in NSF's Division of Computer and Network Systems.

In the next year, Clark will try to answer this question: If computer scientists could design the Internet from scratch, would they do anything differently? Clark's answer today is this: Make it secure. "I don't want to have a sudden meltdown [or] blackout or have the Internet used as a vector for a widespread terrorist attack," he said.

This fall, NSF plans to solicit proposals for a next-generation Internet architecture, Clark said. Agency officials declined to comment further on Clark's grant or on the agency's upcoming solicitation, saying only that they would likely make an announcement in August.

Peter Harsha, director of government affairs at the Computing Research Association, said the proposed cyber infrastructure office might free officials in NSF's Computer and Information Science and Engineering (CISE) Directorate to devote more attention to Internet experimentation.

"As far as we know, under the new plan, the operational aspects of cyber infrastructure will be handled by the new office, leaving CISE to focus on its core research mission, which includes the research that will enable next-generation cyber infrastructure," Harsha said.

"What we don't know is whether research on [next-generation] cyber infrastructure will also be a focus of the new office," he said. "The new office doesn't really do much to alleviate the burden on NSF created by DARPA's withdrawal and the expansion of the field and the increased number of faculty, unless the new office also comes with new money."

Academic scientists are devoting more energy to the field of network design and are hoping the federal government will do the same, said Larry Peterson, a computer science professor at Princeton University and a network architecture expert who recently worked on an NSF project in which Clark also participated. In January, NSF provided a small grant for a workshop at which Peterson, Clark and other members of the research community released a report calling for the federal government to help fund long-term experiments on network architecture.

NSF's research plans could have already had a significant impact on one high- profile effort, NARA's Electronic Records Archives project. Last week, a NARA official said he has been working closely with NSF.

The $500 million project is an ambitious effort to save the government's records and make them available on whatever type of hardware and software people might be using in the future. Robert Chadduck, research director for the ERA project at NARA, said that "all work presumes Internet access."

In thinking about how to preserve records for posterity, ERA managers say experimentation weighs heavily in their current decision-making. Through NSF's Internet research program, Chadduck said, ERA managers can see into the future.

The National Science Foundation is gathering information for a possible effort to design a new and improved Internet.

NSF officials say a fresh look at the basic design principles of the Internet is justified because:

• The Internet's architecture is more than 30 years old.
• The current architecture is fundamentally insecure.
• More than 30 years of modifications have made the Internet increasingly difficult to secure or change in fundamental ways.

— Aliya Sternstein