March/April 2006

Pervasive Wireless

Can't all our wireless gadgets just get along? It's a question that Dipankar Raychaudhuri is trying to answer.

By Neil Savage

This article is the eighth in a series of 10 stories we're running over two weeks, covering today's most significant (and just plain cool) emerging technologies. It's part of our annual "10 Emerging Technologies" report, which appears in the March/April print issue of Technology Review.

In New Brunswick, NJ, is a large, white room with an army of yellow boxes hanging from the ceiling. Eight hundred in all, the boxes are actually a unique grid of radios that lets researchers design and test ways to link mobile, radio-equipped computers in configurations that can change on the fly.

The ability to form such ad hoc networks, says Dipankar Raychaudhuri, director of the Rutgers University lab that houses the radios, will be critical to the advent of pervasive computing—in which everything from your car to your coffee cup "talks" to other devices in an attempt to make your life run more smoothly.

Wireless transactions already take place; anybody who speeds through tolls with an E-ZPass transmitter participates in them daily. But Raychaudhuri foresees a not-too-distant day when radio frequency identification (RFID) tags embedded in merchandise call your cell phone to alert you to sales, cars talk to each other to avoid collisions, and elderly people carry heart and blood-pressure monitors that can call a doctor during a medical emergency. Even mesh networks, collections of wireless devices that pass data one to another until it reaches a central computer, may need to be connected to pagers, cell phones, or other gadgets that employ diverse wireless protocols.

Hundreds of researchers at universities, large companies such as Microsoft, Intel, and Nortel, and small startups are developing embedded radio devices and sensors. But making computing truly pervasive entails tying these disparate pieces together, says Raychaudhuri, a professor of electrical and computer engineering at Rutgers. Finding ways to do that is what the radio test grid, which Raychaudhuri built with computer scientists Ivan Seskar and Max Ott, is for.

One problem the researchers are addressing is that different devices communicate using different radio standards: RFID tags use one set of standards, cell phones still others, and various Wi-Fi devices several versions of a third. Linking such devices into a pervasive network means providing them with a common protocol.

Take, for example, the issue of automotive safety. Enabling cars to communicate with each other could prevent crashes; in Raychaudhuri's vision, each car would have a Global Positioning System unit and send its exact location to nearby vehicles. But realizing that vision requires a protocol that allows the cars not only to communicate but also to decide how many other cars they should include in their networks and how close another car should be to be included. As programmers develop candidates for such a protocol, they try them out on the radio test bed. Each yellow box contains a computer and three different radios, two for handling the various Wi-Fi standards and one that uses either Bluetooth or ZigBee, short-range wireless protocols for personal electronics and for monitoring or control devices, respectively. The researchers configure the radios to mimic the situation they want to test and load their protocols to see, for instance, how long it takes each radio to detect neighbors and send data. "If I want cars not to collide, it cannot take 10 seconds to determine that a car is nearby," says Raychaudhuri. "It has to take a few microseconds."