President Bush and his Democratic challenger, John Kerry, are both telling voters that they want to make high-speed Internet connections available to all Americans.

	Tony Tye, Post-Gazette Raj Reddy envisions a device called the PCtvt, which would include a PC, TV, video recorder and telephone to connect residents of poor villages. Among those working on the project with him are systems manager Edward Walter, left and graduate assistant Vamshi Ambati. Click photo for larger image. Graphic: Features of the PCtvt On the Net: PCtvt Web site

Raj Reddy dreams bigger. He wants to make broadband available to virtually everyone else.

That includes the residents of the small farming village in southern India where he grew up and those of thousands of villages like it. It includes the 4 billion people who live on annual incomes of less than $2,000 a year. And it includes people who can't read or write.

For more than a decade, Reddy, former dean of computer science at Carnegie Mellon University, has been pondering how the developing world could make greater use of today's information and telecommunications technology.

"Does technology have a role to play at the bottom end?" Reddy wondered. Internet connections might allow families to reach a distant doctor for medical advice, or to buy goods and services at competitive prices not available in their village. E-mail might help keep them in contact with distant family members, or enable them to reach government officials.

But are these potential benefits enough to justify the cost to poor families and villages? Would a family be willing to forgo, say, a third daily meal to pay for it?

The answers aren't easy or obvious, but Reddy hopes to begin finding them soon.

Sometime next year, Reddy plans to begin field tests in India, China and Africa of a device he calls the PCtvt ---- a combination personal computer, television, video recorder and telephone that wirelessly connects with the Internet. It all comes at a projected cost of $250 apiece.

These computers, in turn, would link up with an ultra-low-cost broadband technology being developed at the University of California, Berkeley. The goal, said A. Richard Newton, dean of Berkeley's College of Engineering, is an antenna, power supply and other equipment necessary to provide wireless Internet access for a village for about $500.

"Raj and I have been talking about this whole area, just the two of us, for the last six to eight years," said Newton, who serves with Reddy on a Microsoft technical advisory committee. But increasingly it seems that the time may finally be ripe to put their ideas into action.

"The technology," he added, "is now reaching the price point where we can reach these goals."

Reddy said he began working on the project in earnest a year ago, when he finally found a company ---- South Korea's TriGem ---- willing to build the PCtvt prototypes for $450 apiece, with a reasonable chance of building them for $250 a copy once in mass production.

"It's a full-fledged computer with full-fledged functionality," not a stripped down model, Reddy said.

That's in contrast to another computer designed for underdeveloped nations, the Simputer, which went on sale earlier this year in India. The Simputer is a $220 handheld device with a monochrome screen and no keyboard; it can be connected to the Internet.

Reddy took a different approach, giving the PCtvt multiple functions so that it might be more valuable ---- and saleable ---- to poor families who lack many electronic devices. It can be used as a color television, a telephone that uses Voice over Internet Protocol, or VoIP, technology, as well as a standard, Internet-linked computer.

But Reddy also is convinced that it needs to be a powerful device if it is to be useful to people who are illiterate.

"A person who is illiterate needs more computing power and more bandwidth than a PhD," he explained. "They need 100 times more bandwidth, 100 times more memory. This is counterintuitive to most people."

For instance, the PCtvt is equipped with a Web cam so that it can handle video mail, which might be its most important function for a person who can't read or write. E-mail address books use photos of respondents, not just names.

The PCtvt has a keyboard and a mouse, but also can be controlled using what appears to be a TV remote control.

Rather than paying a flat monthly fee for Internet access, as is common in the United States, users of the PCtvt would pay a small fee, perhaps a penny, for each use ---- each e-mail sent, each TV program downloaded, etc.

Assuming that one computer will have multiple users, the primary memory isn't a hard drive, but a flash memory key issued to each user that plugs into the computer's universal serial bus, or USB, port.

Reddy has enlisted the help of students at both CMU's Oakland campus and its California campus, as well as colleagues at the University of Washington, the India Institute of Science and the International Institute of Information Technology in Hyderabad, India.

"Our students, both at Berkeley and CMU ... are really motivated," Newton said. "I cannot tell you how excited these students are. It's the most excited I've seen students in 25 years. They're lining up to work on this."

ICT4B

At Berkeley, the initiative is called Information and Communication Technology for Billions, or ICT4B. The National Science Foundation has provided Berkeley with $3 million to help develop low-cost broadband networks.

This summer, students deployed a prototype system in southern India. But they found that the system, which employs antennas of equal height, had problems in that flat region. "It turns out the trees get in the way," Newton said, so the researchers will have to go back to the designing board, developing a system that uses a few high antennas and many shorter ones.

Building tall antennas at low cost is forcing a new research effort, he noted. One scheme involves using tethered helium balloons equipped with antennas that are electronically steerable, so they can automatically adjust themselves as the balloon moves. "It works ---- on paper," Newton said.

Each village would be able to join the broadband network by erecting its own antenna, as long as it was within range of another village's antenna. "We see this deploying virally," he said, with the network expanding from a few cores.

Like the PCtvt machines, the network is designed with the assumption that connections will frequently be disrupted, such as by power outages. Some of the changes in hardware and software design necessary for this intermittent service might eventually be used to make U.S. network systems more robust, Newton said.

The PCtvt itself also is evolving. Reddy said he's now debating whether he can afford to build the machines with flat-screen displays. Flat screens are less power hungry than cathode ray tube monitors, which is an advantage in underdeveloped nations. But the higher cost of flat screens would likely result in use of a small, 8-inch diagonal screen, which would be unsuitable for most TV viewing.

Even the modest costs that Reddy and Newton envision for their computers and networks might seem beyond the reach of most of the world's population; per capita income in India is $550 a year. But Newton said new business models suggest that the technology may be viable even in poor nations.

One key is the sheer size of the impoverished market. For instance, the Grameen Bank, which operates in Bangladesh and 20 other countries, has shown that it can make profits by making "microloans" of $100 or less. The bank's repayment rates are actually higher than those of many Western banks, Newton said.

Moreover, if farmers can use the computer to get better prices or to arrange timely transportation of crops, or to negotiate lower prices for the goods and services they need to purchase, the computer may pay for itself.

Socially valuable?

The field test that Reddy plans next year will help determine whether this "sustainable computing" is possible, or even desirable. In addition to installing computers and broadband networks, he plans to send students trained in both computer science and social science to baby-sit each computer.

Reddy compares it to HomeNet, the experiment CMU researchers conducted in Pittsburgh in the mid-90s. Fifty families were given computers and Internet links, while researchers studied how family members used the machines and how they affected their lives and moods.

Likewise, the students in the field test will not only evaluate how well the technology works, but whether the people use the equipment and find it valuable.

"Technology we know how to do," Reddy said. "It's the social aspects that we have no clue about."