Rensselaer Magazine * Rensselaer Links
* * *
* * *
* * *
President's View Mail Class Notes Features

Connecting It All
The potentially rapid growth of pervasive computing means networks of connected devices will themselves grow larger. In order to ensure pervasive devices are performing well, a strong network infrastructure also will be vital.

The study of networks is one of the principal preoccupations of researchers within Rensselaer’s Center for Pervasive Computing. Several members of the center — including Szymanski and Shivkumar Kalyanaraman, associate professor of electrical, computer, and systems engineering — have joined forces to develop a variety of large-scale network simulations to investigate the behavior of both the Internet and wireless sensor networks.

* *

Some of the most-publicized pervasive computing devices are tricked-out appliances, like the camera-equipped refrigerator. Before you go home, you could look inside the fridge via Web cam and see if you need to buy milk. Photo by IBM

A related research focus involves getting devices to sense the network itself. Tools in networks of pervasive devices communicate directly with each other — as opposed to today’s cell phones, which bounce messages off towers — and the relatively ad hoc nature of wireless networks, as devices come on and off line, complicates this issue considerably.

“The ability to understand where things are, and how they are, in a distributed fashion, is one of the real challenges of large-scale pervasive computing,” says Ken Vastola, professor of electrical, computer, and systems engineering. “You have many nodes that are not planned in advance and they have to figure out who they are and what else they can relate to in a network.”

For instance, Vastola and Lisa Shay, Ph.D. ’02, have developed a system that can predict when a wireless connection may soon fail. This system can be used, for example, in military wireless networks that link individual soldiers or units in areas of uncertain terrain. “You would like to figure out the quality of the connection, and how long it will last,” says Vastola. “Is it in danger?” Data being transmitted produces a pattern upon reception, so by analyzing that pattern, it may be possible to let soldiers know when they are losing touch with others in their unit.

This and other Rensselaer projects have garnered funding from the U.S. Department of Defense, another key player driving pervasive computing research. But research at Rensselaer has a broader scope. Kalyanaraman, an expert in Internet traffic management and networking systems, has several NSF projects on ultra-high-speed last-mile wireless networks using cheap radio-frequency and optical-wireless technologies. One of these projects involves building an organically evolving community wireless network on the Rensselaer campus that will include radio frequency antennas fashioned from Pringles cans by undergraduate researchers.

Bulent Yener, associate professor of computer science, is working on security in wireless networks. Yener has extensively investigated the optimal power allocation, coverage, and routing problems in ad hoc wireless networks. “Securing wireless networks of power-constrained devices is a very difficult task and vital in the fight against ‘power attacks’ by mobile adversaries,” says Yener.

Security and Privacy
Before their products gain traction, firms developing pervasive computing devices still face obstacles, especially in the areas of security and privacy. Connecting more devices together, after all, means more possible access points for people trying to obtain data illicitly.

“Security is a huge issue in pervasive computing, and most pervasive computing projects focus primarily on the hardware feasibility of the connectivity,” says SRIC-BI consultant Powell. “Security comes after that.” A problem, Powell notes, is that “the reduction in size of the sensors itself leaves very little room for anything but the ability to communicate. There’s just a tiny sliver of processing power and memory left over for the security software.”

Linux WatchPad

IBM Research’s Linux WatchPad. Photo by IBM Research

Narayanaswami’s IBM group is addressing concerns regarding battery power by having the watch’s operating system be mostly event driven and at other times “wake up” only periodically — say, once a minute — thus conserving energy for times when it is needed most. He believes processing power will increase over time, and, he says, “you can run standard security protocols on top of these small devices as well.” Still, Narayanaswami acknowledges, pervasive computing security will take a familiar shape: “If there is commercial benefit in hacking into these things, people will attempt it, and people will obviously try to develop better technologies to prevent that.”

Privacy is also a concern, since the flip side of having sensors on your car for insurance purposes, for example, is that the same information could work against you. Consumers could be wary of pervasive devices that monitor their cars, appliances — or themselves. In this case, says Prial, “it’s how people manage the information that matters. Industries need to do a good job policing themselves.”

Then there is the perennial challenge of creating compatible operating system standards. The value of something like telematics — the ability of cars to send messages to each other — would drop considerably if, say, Hondas could only communicate with other Hondas.

“If a company is talking about proprietary standards, or locking you into a single platform, that’s a problem,” says Prial. Of course, IBM prefers open standards because they could allow the company to sell its software and services more broadly. Still, Prial argues, it can be in the financial self-interest of companies to use shared standards. “With great network ubiquity, great device availability, and open standards, the price of entry for [pervasive] solutions has significantly decreased.”

The hurdles facing pervasive computing are familiar to the rest of the technology and business world. Those who have been attracted to the field, however, say they expect innovations to continue, often in ways not yet imagined. “Ten years from now people will be doing things with pervasive computing that we haven’t thought of,” says Vastola. “I believe that’s very exciting.”

And many agree that as pervasive computing becomes more pervasive, the field may no longer be seen as an esoteric segment of technology, but simply as mainstream computing. Prial suggests it eventually will be so familiar to us, its name may no longer be necessary: “The term ‘pervasive’ will not be around forever, but the technology will be.”

<<Previous | 1 | 2

Peter Dizikes is a freelance writer based in Arlington, Mass.

Rensselaer Magazine: Summer 2004
Rensselaer Home Page | RPInfo | AlumServ Home Page

Rensselaer (ISSN 0898-1442) is published in March, June, September, and December by the Office of Communications.

Opinions expressed in these pages do not necessarily reflect the views of the editors or the policies of the Institute.
© 2004 Rensselaer Polytechnic Institute. All rights reserved worldwide.
Rensselaer Polytechnic Institute (RPI), 110 8th St., Troy, NY 12180. (518) 276-6000
Web site design by the Rensselaer Office of Communications.
Jane Van Ryan, Assistant Vice President, Office of Communications.
Questions? Comments?
Please contact us.