As the co-founder of Inktomi, a Web search and software company that flourished in the late 1990s, Berkeley EECS Professor Eric Brewer had the opportunity to travel the world and meet, as he says, "many interesting and talented people from developing regions." Over and over, he heard about the struggle of developing countries to absorb technologies that were clearly not designed with their needs in mind.
In the Berkeley Intel Research Lab, EECS Professor Eric Brewer (left) and Sonesh Surana, a graduate student, make adjustments to a prototype of low-cost wireless technology. The technology was deployed as part of a collaborative project between UC Berkeley, Intel, and the Aravind Eye Hospital in India. (Photo by Peg Skorpinski) The problem, he realized, is that the technological advances of the Information Age have been driven primarily by the demand of developed countries for bigger memories, faster computers, smaller devices, and new ways to connect to the Internet. But in places where electric power is unreliable, and basic parts are hard to come by, powerful is less important than functional, robust, and cheap. "Technology researchers need to create new, more appropriate options for developing regions," Brewer says. "Current solutions are too complex, too expensive, and too power hungry."
In 2003, after Inktomi was sold to Yahoo, Brewer launched a groundbreaking new project at Berkeley called Technology and Infrastructure for Emerging Regions (TIER). TIER brings computer scientists together with social scientists, city planners, and development experts, with the goal of identifying the unique needs of the developing world and using this knowledge to create new computer and networking technologies that can be applied in a wide variety of areas, including health, education, agriculture, and finance. Brewer’s vision is "to make a few products that can be used in hundreds of places without our assistance," says Sergiu Nedevschi, an EECS graduate student working with TIER.
In order to meet this challenge effectively, it's essential to have a firsthand knowledge of the conditions in which these systems and services will be deployed. Thus, Brewer and his students are involved in ongoing projects in India, Rwanda, and Ghana, as well as exploring the potential for future projects in such far-flung places as Uganda, Malawi, and Mozambique. The goal is to understand what the common challenges are and figure out the best ways to address them with technology that is low in cost yet not inferior. "The systems we have developed stand on their own in terms of technical quality, complexity, and novelty," says Michael Demmer, another student working with TIER.
The Doctor is In (via Wireless)
The Aravind Eye Hospital in Theni, India serves a population of two million, but many of these people do not receive timely treatment because they live in rural areas and cannot easily get to the city. The hospital used to send a van to outlying villages, but this was hardly an optimal solution. "Without a permanent presence in those towns, you can’t do prevention and education," Brewer says. "Eighty percent of the problems they see are preventable. The biggest one is cataracts, which are caused, in part, by indoor cooking with wood fires."
A TIER-sponsored project in India equipped shared computers with multiple mouses that operate independently. (Photo courtesy of Microsoft Research India) In 2004, one of Brewer’s students, Sonesh Surana, met with the management of Aravind to discuss how TIER could help. Over the following months, Surana helped the hospital to set up a new type of wireless network connecting the doctors in Theni with clinics in the villages. The doctors can now see their patients via a videoconference link, talk with them and an on-site nurse, and download photographs taken with an optical camera. The virtual office visit is good enough to provide diagnosis and treatment for most patients and to identify the ones who need to come to the hospital for eye surgery.
The wireless technology used in the Aravind network is, by design, a combination of widely-available hardware and custom-designed software. For projects in developing countries, all of the hardware must be inexpensive, durable, and easy to repair and replace. No such limitations constrain the software, however. The cost of software is mostly the intellectual capital of the developers—and that is precisely what TIER can provide for free.
TIER's wireless network in India uses off-the-shelf WiFi transmitters and receivers, with one significant alteration: The 802.11 WiFi protocol restricts the practical range of transmissions to only a few hundred meters, which is sufficient for "hot spots." By rewriting the driver and using different antennas, Brewer and his graduate students extended the range of links to 50 kilometers or more. The cost of each link is low, about $800, and the power requirement is also low, around eight watts. Yet the bandwidth is high enough to support high-quality video transmissions. The system can run on solar power, which is an important consideration in a country whose power system is subject to frequent outages and brownouts.
The new network came online on a trial basis in January 2006. Five clinics are now up and running, for the most part, autonomously. “We don’t want a project that ends when we leave,” Brewer says. Because of its emphasis on empowering local technology experts, the TIER model is having a multiplicative effect: The Aravind Hospital hopes soon to set up 50 similar vision centers, each treating up to 10,000 patients a year.
Observe, Then Innovate
Brewer says that collaborating with faculty from other disciplines has taught him a new philosophy for applying engineering solutions to the developing world. "In general, engineers are interventionists," Brewer says. "But the rest of the campus is more observational. We’ve changed our strategies as a result. We do more before-and-after studies. You don’t want to delay something that is helpful, but you want to make sure that it will be helpful. Social science is much better than computer science for replicating results and better for understanding the cultural context."
The policy of observing first, to see what the host country really needs, led to one particularly clever innovation: In 2005, Joyojeet Pal, a graduate student in the Department of City and Regional Planning, conducted a study in which he observed how children in public schools in India made use of the limited number of computers available to them. At the beginning of a session, typically ten or more students would cluster around one computer, excited at the chance to use it. Then one student would monopolize the mouse, and gradually, the other students would lose interest.
Why not, Pal thought, attach more than one mouse to the computer? Pal searched for studies on equipping computers with multiple mouses and found a scant few of them; even those were motivated by theories of collaborative learning, rather than by economic necessity. So Pal and two researchers from Microsoft Research India decided to address the issue themselves. They reprogrammed Windows to display five cursors controlled by five different mouses. It was not an easy fix—the assumption of one cursor per computer is embedded deep in the operating system—but the results were gratifying. Even though there were still more children than mouses, they shared control, no one was left out, and no one got bored. Instead of spending a lot of money on new computers, the school could spend a little on mouses and get almost the same results.
It is "vastly fulfilling," Pal says, to make a tangible contribution to the world while at the same time doing research toward a dissertation. As Brewer says, "There aren’t many research projects anywhere in the world which have so much potential to affect the lives of real people."
Graduate student Melissa Ho, of the School of Information, climbs a wireless transmission tower as part of her work for a TIER project in Legon, Ghana. (Photo by R.J. Honicky) The chance to do good, while at the same time doing good engineering, has fostered a real esprit de corps among the students involved in TIER. On their blog and in meetings, they enjoy sharing stories of the obstacles they encounter abroad, from sockets that don’t match the plugs on the equipment to trucks that break down in the middle of nowhere. "There’s little you can learn in the classroom that teaches the kind of common sense you learn with TIER," says Demmer.
The students involved in TIER feel they are building a new discipline: information technology for development. "We’re creating our own job market," says Melissa Ho, a graduate student in the School of Information. So far, TIER work is a specialized niche. But as the world becomes more aware of the technology divide, the TIER students expect the demand for their expertise to increase greatly. Companies such as Intel, which is sponsoring some of their research, are beginning to recognize that developing countries are an enormous market for their products. To tap that market, they will need to figure out how to make technology work both in and for the developing world.