Hui Zhang today has a blank sheet of paper in one hand and $7.5 million in the other.

So equipped, the Carnegie Mellon University computer scientist has been charged by the National Science Foundation to design a new communications infrastructure, one that likely would entail ripping out all of the copper telephone wires that link the nation's homes and businesses and replacing them with fiber-optic cable.

The goal is to provide high-speed, broadband Internet access to virtually every U.S. home. To do so will require not just replacing telephone wires, but also redesigning the Internet itself, eliminating bottlenecks that slow network traffic, while adding safeguards that would enhance the security of data -- hence the clean slate.

"This is the opportunity of the century," said Zhang. He added that such an undertaking could easily cost $100 billion or more, but "it's probably going to be there for a hundred years."

Zhang's project, which also includes collaborators from the University of California at Berkeley, Stanford University, Rice University, AT&T Research Labs and the Pittsburgh Supercomputing Center, is one of eight large information technology studies to win funding from the National Science Foundation. The researchers have been asked to develop a technical blueprint for a nationwide high-speed network and to identify the technology necessary to make it a reality.

The agency announced a total of $169 million in awards for its Information Technology Research program, including 175 mid-sized and 180 smaller projects.

Among the eight large projects was a $9.4 million award to biologists, including Carnegie Mellon's Robert Murphy, at four institutions, to develop automated methods for analyzing microscopic images of cells.

Building a new communications infrastructure ultimately will be an undertaking of private industry, Zhang said, but government policies and regulations can guide the construction. Pressure to upgrade what was originally designed as a telephone system is increasing, he maintained, noting that other countries, such as South Korea and parts of China, are already doing so.

"It is critical if this country is to be globally competitive to have an up-to-date infrastructure," he said. Now, only about 10 million American homes have broadband Internet connections, either DSL or cable modems, with top speeds of about 500 kilobits per second. Zhang proposes wiring 100 million households with connections that are 200 times faster -- 100 megabits per second.

But speeds on the existing Internet would probably top out at just 2 megabits per second, Zhang said, so a new design and new components, such as fiber-optic cable, will be necessary.

What can you do with 100 megabits per second? Wendy Huntoon, the Pittsburgh Supercomputing Center's assistant director for networking, can speak from experience because she and many of her colleagues already have 100 megabit connections to their office desks.

"You don't worry about the size of files you send," she said, noting that she thinks nothing of assembling a bunch of digital photos into an e-mail and sending them to a colleague. At home, she would think twice because it would tie up the computer too long.

At her office, video conferencing, in which the computer is transformed into a sort of videophone, is routine. Video recordings can be easily shared between colleagues. Sending movies over such a system is a piece of cake.

"Quite frankly, in the university setting, people do this kind of stuff all the time," Huntoon said. What might be possible once such connections are available on the home front can only be imagined, she said.

Because the supercomputing center staff already is acquainted with such high-speed connections and because it has access to fiber-optic cables that cross the country, the center will play a key role as a test bed for some of the new technologies that Zhang and his colleagues devise.

How much testing -- and how much technology development -- occurs during the project will depend on whether additional funding can be located, Zhang said.

The existing Internet is a series of computer networks interlinked by routers -- devices that forward packets of information from one network to another. It's a design that has allowed the Internet to expand and provided great flexibility, but it's also a design that limits data transmission speeds. The next system, Zhang said, will need to find an alternative to routers.

More attention also will be paid to data security, he added. The Internet was designed for a "friendly" environment, but it now is a public utility that needs to be concerned about cyber-attacks and misuse. Michael Reiter, associate director of Carnegie Mellon's Center for Computer and Communications Security, is one of the investigators on the project, as is former computer science chairman Raj Reddy.

In the other large project, "Data Mining for Biomolecular Imagery," Murphy and colleagues at the University of California at Santa Barbara and at Berkeley and at the Massachusetts Institute of Technology are designing computer-controlled systems that can analyze and compare microscopic images.

In Murphy's lab, for instance, he and his team are working on "location proteomics," an effort to describe where in a cell different types of proteins can be found and how those locations change in response to development, or illness or drugs. Because each cell might contain thousands of different proteins, the undertaking requires studying many thousands of microscopic images.