A Black Hawk flies over a foreign city, a fine dust cloud spreading out from the door. The particles of that cloud are each tiny computers. Wherever they land they will harvest power from their surroundings and use it to sense movement, temperature, light, and transmit the data wirelessly. They’ll monitor the populace, tremors in the land, changes in the landscape.
Such tiny, self-sufficient, spying, collecting, and reporting computers are not the thing of some far off future, be it dis- or u- topian. That future is here, at least in prototype form. “What we have today is millimeter scale computing systems, that integrate all of the essential features: sensors that can measure light, pressure, temperature,” says Prabal Dutta, a professor at the University of Michigan. “They communicate wirelessly and harvest energy autonomously.”
The genius—and the slog—behind the Michigan Micro Motes, as the tiny computers are called, has less to do with shrinking the size of any one component, and more to do with the painstaking 20-year long task of getting all tiny parts to work together. “There’s no one single thing that had to happen,” says Dutta. “We had to solve a spectrum of challenges.”
Chief among the difficulties of getting a minuscule computer up and running is how it will harvest, store, and use, power. The amount of energy a device can use shrinks as a function of a cube. Shrink a two-inch wide cubic device down to a one-inch wide cube and the amount of energy it can store will drop eight times.
Sensing, computing, storage, energy harvesting, and wireless communication in 1 cubic mm volume. Image: Umich.edu
“The millimeter volume is important because it’s a concrete goal, it kicks the crutches out from under you.” explains Dutta. “The size is nice, but the size is only part of the issue—we’ve been able to make small chips for a long time. It’s not that we couldn’t bundle small systems before, it’s that we couldn’t scale power systems down, couldn’t match the energy draw.”
The Micro Motes can harvest energy from an onboard photovoltaic cell some thousand times smaller than that found on a solar-powered calculator. The computers can also pull power electrochemically from soil, or from temperature gradients. But the amount of power the Motes have to function with is in the picowatt range. “The magic is how do you use that power? How do convert the energy efficiently.”
The answer is to rework everything. The Micro Motes were not assembled with off the shelf components.
Take, for instance, the Micro Motes’ clock. To conserve energy the computer spends a good deal of time asleep. It needs to wake up, collect data quickly, or transmit data quickly, and then go back to sleep—and it needs a timekeeper to do so. But clocks, though, are notoriously power-hungry devices. Those on the market requiring the lowest power, use in the neighborhood of 100s of nanowatts. That’s more than 100 times the power provided by the solar cell.
So Dutta and his colleagues turned to a phenomenon already sure to occur periodically (though irregularly): power leakage. “You can build a clock of it,” says Dutta. “Now it turns out the clock is terrible. It’s a stochastic process, has a temperature coefficient to it.” But, assuming each Mote is one of hundreds or thousands of comrades switching on and off as their leakage so dictates, sporadic will do.
Though leakage proved essential as a ticker, it was still best to keep it to a minimum. So the team reworked the RAM to make it less leaky, using 11 transistors compared to the more typical six. “That was a huge important thing,” says Dutta. “You can turn off most of the other things—the radio, the processors, sensors—but you can’t just forget the state of the computer. You can’t run a computer if it loses all of it contents.”
Power was also conserved with the radio. Unlike others it doesn’t need to warm up. By using ultra wideband radio transmitting very small pulses, the system can wake up and go back to sleep in nano-seconds.
Similarly, on most chips in a computer there are diodes that wick away static electricity, shunting it back to the power supply so the transistors doesn’t fry. Those diodes, though, leak more power than the entire Micro Mote power budget. So Dutta’s colleagues redesigned the chip to provide electrostatic protection on the cheap. “That’s mundane, but no one’s done it before,” says Dutta. “The key is, it’s not business as usual.”
Aside from spying, what are the potential applications? The snowflake-sized computers could measure tremors and warn of potential earthquakes. They could be put in susceptible eyes to warn of an elevated risk of retina detachment. They could be put in concussion victims to report on cranial pressure—obviating the need for the holes that are drilled in the skull for that purpose now. They could be put on honeybees to help us understand colony collapse disorder. They could be put on key chains to help prevent where-the-hell-did-I-put-my-keys disorder.
“I’m not imaginative enough to come up with a fraction of the applications that others will come up with,” says Dutta. Will those as of yet unimagined applications outweigh those of the Brave New World sort? “We all worry about the Orwellian implications with this kind of work. The question is wait a minute—we have the ability to make millimeter cubes that never die, build them by the millions and put them everywhere. What could go wrong with that?” says Dutta. “I don’t know. We should definitely be cognizant of that.”
Michael Abrams is an independent writer.
The millimeter volume is important because it’s a concrete goal, it kicks the crutches out from under you.
Prof. Prabal Dutta,
University of Michigan