Smart and Tiny Forest Fire Detectors

Sept. 13, 2002 -- If a single tree in the forest catches on fire and no one is there to see it, will it still burn?

The answer is pretty clear-cut. Or, at least it becomes evident once the lone burning tree ignites the surrounding forest into a blazing inferno.

Still, spotting and stopping an isolated fire — caused by a strike of lightning or a careless human camper — isn't an easy task for human forestry workers who must watch over millions of acres of wilderness.

But researchers at Accenture Technology Labs in San Jose, Calif., are working to develop a high-tech aid: a network of tiny electronic sensors that automatically watches for and warns against signs of burning wild foliage.

Mighty Mites

The system would rely chiefly on experimental machines called "Smart Dust" being developed by Kristofer Pister and a team of researchers at the University of California in Berkeley.

As the name implies, the devices are extremely small — ideally measuring no larger than a few grains of sand — and contain instruments to measure temperature, humidity, light and other environmental conditions. More importantly, the "smarts" of the devices would come from embedded microprocessors, software code and wireless communication systems.

Chad Burkey, the senior researcher for the project at Accenture, says that basic principles of the system is fairly simple.

The Smart Dust sensors would be sprayed across a forest from an airplane or another distribution method. Once embed within the trees, each speck of dust locates and establishes a wireless link with each nearby mote.

Collaborative Dust Clouds

Each sensor is programmed with proprietary algorithms developed by Burkey's team at Accenture to look for specific environmental signals that would indicate potential problems. "A rapid increase in temperature and light levels with a rapid decrease in humidity is a fairly good indicator that something is burning, for instance," says Burkey.

When a speck detects a possible anomaly, it contacts the other nearby dust-sized devices to determine what their readings are. And by collaborating multiple readings from multiple sources, the sensors themselves can determine if there is danger to the tree they're planted on.

"In essence what we are trying to do is rather than rely on a central data collection point, we're building intelligence into the system itself," says Burkey.

Once a danger is determined, the group of triggered sensors uses its wireless connections to send a single message back to forestry workers monitoring the network of sensors.

Promising Technology…

Michael Palma, principal analyst with the Gartner research firm also in San Jose has seen prototypes of the system in Accenture's labs and is impressed.

"It looks like an interesting application combining microprocessor and communications," says Palma. "It gets to the heart of how do you monitor a huge forest without deploying a large number of expensive human or mechanical assets."

What's more, both Palma and Burkey are optimistic that such a dust-based system could be extended to other environmental monitoring applications as well. A slight tweak of the embedded software could allow the system to be used in pipeline to monitor for leaks, for instance.

…That Needs Some Time

Still, Burkey does note that the system still has some tough development work ahead.

For one, smart dust developers still have to figure out how to make the devices truly mote-sized. Researchers at UC Berkeley have designed dust sensors that are about 100 cubic millimeters in size. But they aren't working prototypes yet.

Burkey says that Accenture has garnered some interest in the system from perspective clients, such as paper and lumber companies. But a commercial system is probably still three to five years away.