This “Golem Dust” mote is a sensor that detects ambient light and acceleration, and incorporates a tiny radio antenna (the cross) for communication. Credit: Brett Warneke, Kris S.J. Pister, Berkeley Sensor & Actuator Center, University of California, Berkeley

The coming generation of sensors will have to be made of stern stuff. After all, many of the potential applications call for lots of sensors, scattered so widely through the target area that there's no hope of tending to each one individually. Instead, the devices will have to operate on their own for days, weeks, or months at a time—in places where there are no power sockets, no broadband cable connections, no tech support, and no protection from being soaked, baked, frozen, buried, stepped on, or even eaten.

This makes for some daunting engineering challenges. Among the toughest is power: a sensor that's made to be tiny, inexpensive, and mass-produced is a sensor with very little room for a battery. That's why designers often arrange to have the devices spend most of their existence in sleep mode, where they can survive on just the barest trickle of power. From there, they have to wake up only for a tiny fraction of a second every now and then, so that they can take a quick instrument reading and, if need be, beam back a few bits of data.

Another challenge is getting those bits back to headquarters. Out in the field, where there is no Internet, the latest sensors can do that by passing the data from one to the next via wireless networking technology—in effect, making their own network. But this is a lot trickier than it sounds. For one thing, the connections are typically restricted to very low power, very short distances, and very low data rates. Worse, the transmissions can be very noisy and erratic; furthermore, if the sensors are connected to a vehicle—or an animal—they will frequently be moving around.

Click on the trees to see how ad hoc networking works and view other sensor animations. Credit: Nicolle Rager, Zina Deretsky, National Science Foundation

That's why many engineers are emphasizing ad hoc networks, in which the sensors are programmed to reach out, find their nearest neighbors, and form network links on their own—without anyone to show them how. If any of those links are blocked or broken; moreover, the sensors will automatically reach out and find new links to replace them.

And then there is the whole realm of societal challenges. What is the best way to build in privacy safeguards, for example, so that the new-generation sensors don't become a tool of Big Brother? And how do you build in equally strong security safeguards, so that hackers can't just eavesdrop on the wireless data stream?

NSF-funded researchers are pursuing solutions to all these challenges and more—as are researchers supported by other agencies, and by industry. Still, enough sensor technology is already in hand to support a host of applications. Read on for more examples from Environment & Civil Infrastructure, Industry & Commerce, Health and Safety & Security.