Power Play

A new bare-bones wireless experiment gets ready to hit the airwaves.

By David Cameron

March 27, 2002

To tell Berkeley professor Jan Rabaey that he can't sink any lower is, for him, the highest praise. After all, his group at Berkeley's Wireless Research Center would feel vindication for their lofty goal: to create a network of wireless nodes, what they call picoradio, that run advanced information systems, use an absolute bare minimum of energy, and cost next to nothing.

"Pico" is a prefix that means "one trillionth," and while it doesn't describe the dimensions of the system, it conjures Rabaey's target of creating a network that can go as far as possible on as little as possible. "The questions we've been asking are, how low can we go? What is the bottom limit in terms of power dissipation for the wireless world?" says Rabaey, who teaches electrical engineering and computer science at Berkeley and heads the picoradio project at the center, an industry and government funded research group that focuses on design issues for next generation wireless devices.

That focus on "going low"-ultra low power and ultra low bit rate-is at the forefront of today's wireless research, according to Bhusan Gupta, a research manager for Geneva-based STMicrolectronics, an early backer of the Berkeley Center. Although the wireless world is awash in competing technologies like Bluetooth and the various gradations of the 802.11 standard, "the real unexplored space is where Rabaey is today," says Gupta.

According to Rabaey, each node in the picoradio network will draw no more than 100 microwatts of power, three orders of magnitude less than 802.11b's 300 milliwatts. The benefits of such a low power network are obvious: no batteries are needed because each node can harvest all the energy it needs from its environment. Even a Bluetooth-based radio, which can operate at 70 milliwatts, still needs a battery to operate and thus will always need to be maintained.

But this benefit raises an equally obvious question: is it possible to create a network that can to do so much on so little? The Berkeley group will have an answer soon enough. In the next two weeks they will complete their first crude prototype of a wireless network that will demonstrate the feasibility of picoradio's low-energy protocol. They intend to build fully operational, built-to-scale network prototype by December.

Each node, in its final form, will be no bigger than a shirt button and will contain several elements: a sensor, or set of sensors, for measuring elements like temperature, pressure, motion, light or sound; a microprocessor; an interface to communicate with other nodes; and a power component, either solar cells or a piezoelectric polymer that can convert vibrations from the environment into electricity.

A sample application, Rabaey says, is climate control for large buildings. Rather than raising and lowering a single thermostat, the nodes would monitor variables like temperature variation, light conditions, humidity factors and building occupancy. The network could then configure itself-without any human intervention-to activate mechanical devices to open or close vents, valves, or blinds in different locations at particular times of day. To operate on such low power, each node only communicates with the node closest to it, which in turn signals the next closest node, creating what Rabaey describes as a "multi-hop network."

"It's almost like querying a database," says Rabaey. "If I send a request into the network saying, 'Give me the temperature in the kitchen,' it propagates through the network until it meets a node that says, 'I'm in the kitchen, and it's 70 degrees.'"

Rabaey and his group envision nodes so inexpensive that a person can purchase them at RadioShack, then-without having to do any sort of network configuration-program them at home to control climate, lighting or even security.

Rabaey's work has generated a fair amount of interest-and money-from both government and industry. Potential applications go far beyond checking temperatures. Since each node is essentially a blank slate that can do whatever it's programmed to do, the network can be used for other jobs that employ radio frequency identification technologies, like tracking items or people in a contained space. Bob Graybill, program manager of DARPA's Power Aware Computing/Communication, says that Rabaey's research "is one of the key technologies that we'll be evaluating over the next few years for our distributed-sensor technology."

Not everyone is convinced that Rabaey's intention to run a network with such an infinitesimal power supply will work-and even if it does work, that it is at all necessary. Says Ken Dulaney, vice president for mobile computing at Gartner, "Because you need lots of power for so many other factors, in the end it might just make more sense to use a standard technology like 802.11 that uses more power but can get more distance." Additionally, he says, reducing one cost can inflate another. "A technology can cost a buck, but 20 bucks to install."

Story continues below

And what if Bluetooth gets its price down before Rabaey and his team work out all the kinks in their system? "Something at the same price with longer range will certainly beat the picoradio system," says Dulaney.

But the success of picoradio may not be the success that the Berkeley group currently envisions. Says Gupta, "I'm not sure if Rabaey can deliver on everything in the exact way he intends, but I believe he'll deliver quite a bit." Dulaney adds, "And that's what research is all about. In the end, you may only hit on one dimension, but that dimension may be the one that matters most."