1541-4922/04/$25.00 © 2004 IEEE
Published by the IEEE Computer Society
Universities Focus on Pervasive Computing Research
Terry Costlow
As people use computers more and more, it's increasingly challenging to ensure they can all interact and share data without making proprietary or personal information too widely available to others. The Cambridge-MIT Institute (http://www.cambridge-mit.org) is focusing its research efforts on pervasive computing challenges, including security, networking, and user interfaces.
CMI will invest more than US$5 million to explore ways to make computers more responsive to human needs, reducing the amount of effort people must spend interacting with the growing number of systems used in everyday life. Dubbed the Pervasive Computing Community (http://www.pervasive-cmi.csail.mit.edu), the research initiative is the fourth Knowledge Integration Community created by the two universities, which formed an alliance in 2000.
Sharing Research to Improve Technology
A key benefit of the initiative is to give researchers the freedom to share ideas and results more easily and more often than in the normal course of academic openness. "It lets us do research on both sides of the Atlantic and increases the occurrence of chance interactions. Quite a lot of research is pure serendipity, getting the right people together at the right time to get a noble outcome," says Simon Moore, lead researcher at Cambridge University's Computer Lab.
The two teams are working on projects aimed at creating atmospheres based on personal preferences. "You might be able to walk into a room and the drapes would open or close, or the radio station might change," says MIT's Umar Saif, the initiative's project manager. MIT is working on ways that disparate devices can work together, while Cambridge researchers are looking at how to use this technology in broader networks.
As they go forward, the they will enhance the research by adding partners who can increase funding and add knowledge. "We are working on a cooperative partnership with industry partners in the US and UK, talking mainly to small- and medium-scale companies," Saif says. They'll look at the various technologies and ways to exploit those that have the highest commercial possibilities, he says.
A major aspect of improving these interactions is to upgrade peer-to-peer networks, Saif says. "We want to make peer-to-peer models more effective."
As sharing data gets easier, researchers must also prevent others from accessing personal data such as banking or diary information. "With mobile devices, we want anyone to pick up hardware at any point and log in," Saif says. "That's got to be handled by a secure system with secure processes."
One security method is using data encryption. The teams have already spotted flaws in popular technologies and have created fixes. "Crypto APIs are a very important area. Teams on both sides of the Atlantic have identified weaknesses in current technology," Moore says, adding that this information might go into the public domain before long.
On the hardware side, researchers are developing power management and battery technologies so people can use mobile computing systems for longer periods. Because mobile products are important components in the pervasive computing environment, they need to be on most of the time. Minimizing the amount of time spent worrying about recharging batteries meshes with the goal of making computing more invisible.
Another critical area is in improving user interfaces. Today, humans must interact with computers by learning the computer's language. Using speech instead of a mouse or stylus is one way to help the computer learn human commands. Another is to let the computer watch its operator and interpret the person's actions. "The computer vision project is based on understanding human gestures," Saif says.
Research in speech recognition is already advancing to improve foreign language coursework. "We're doing speech work, improving voice recognition and improving language learning," Moore says.
That language learning uses computing power to provide far better learning experiences than today's machine-oriented learning aids. Instead of the user mimicking an audio tape with minimal feedback, future systems could be more interactive. "Someone could repeat the phrase, then the system can help them correct their pronunciation or accent," Moore says.
The teams will also explore ways to open communications with someone and continue even as the participant moves about. Connections are continuously upgraded or downgraded as the person changes locations, for example shifting from video conferencing, to a cell phone, to walkie talkies or iPaqs. The initial work was extremely effective, making researchers realize that an important part of pervasive computing is the ability to get away from the computing environment, but it wasn't flawless, Saif says. "The problem was that there was no way to gracefully shut it down—it was too good at keeping the conversation open."
DOWNLOAD PDF 
Peer-to-Peer Networks in Action (http://dsonline.computer.org/0201/ic/w102gei.htm) FREE IEEE Internet Computing, (Jan./Feb. 2002) 
June 2003)