Computing's power problem

Credit: Carnegie Mellon Electrical and Computer Engineering

When it comes to computing, much of the focus is on getting more powerful computations out of ever-smaller chips (think of your cell phone) but it isn't always about intense calculations. There is a different issue: power. If we want to deploy sensors in the corners of the world or the crannies of our lives, the chips need a continuous source of stable power or the software basically breaks. Brandon Lucia at Carnegie Mellon University is trying to solve this in hopes that low-cost, low-power computer chips can be used with less risk of failure.

"This isn't about high intensity computing. It is about not having a battery," Lucia says. He is developing a programming language and computer systems to be included in the world's smallest satellite, expected to launch next year.

The battery problem: Computing can be a tradeoff between the computation to be made and the energy that it requires. Batteries weigh too much and tethering the chips to power sources limits where they can go. Instead, tiny solar panels can harvest energy from the environment but the power they provide isn't consistent.

The software problem: Software typically runs in sequence — from step 1 to 2 to 3 and so on. When the power cuts, a program will start at step 1 again or another intermediate checkpoint step. Too many restarts like that can cause inconsistencies in the program, mess up data in the computer's memory, and stop the program. Currently, expert programmers custom build - over months - software that operates under these conditions but even then there is no safety net when a program fails.

Lucia is addressing this in two ways:

  1. He created a new programming language and software system that retains the data in the memory, allows a program to pick up where it left off, and keeps trying to execute the task until completion.
  2. By combining that system with a hardware platform that buffers energy, the chip's computations can work with intermittent power. "We're creating useful systems out of those flaky components by changing the way hardware and software communicate," he says.

Testing: The system has been field tested on the university's campus with an application that collects compass and gyroscope data. But a big test will come next year when a chip satellite (chipsat) employing their design is scheduled to launch into space. Zachary Manchester, who created the KickSat mission to launch chipsats into low Earth orbit in the name of democratizing space exploration, says:

"What I like about Brandon's work is the combination of low-cost off-the-shelf hardware and clever software to make robust systems that can work in harsh environments. Right now, a lot of the computing hardware used onboard spacecraft is essentially custom built to survive in space. That makes it way too expensive to use in small low-cost satellites."

More applications: Lucia says the terrestrial applications of these low-power, 1-in. square, 4mm thick chips are just as exciting:

  • Maintenance: Lucia's team is working on a system designed to be used on a pipe to monitor its temperature by placing sensors along its length. "If you have 40,000 devices sensing and providing data, you can't go switch the batteries all the time," he points out.
  • Energy conservation: Imagine an indoor occupancy detection system consisting of sensors on a board that are placed throughout a building so the HVAC system can be aware of how many people are in the room and adjust the temperature accordingly. Running power to them wouldn't be ideal nor would switching batteries on 42 stories of sensors.
  • Medicine: People have proposed ingestible robots but a key advance would be to get batteries away from the body, Lucia says.
  • Reactive science missions: Lucia and Brett Streetmen from Draper both envision using chipsats to monitor wildfires, volcanic eruptions, or wildlife movement in remote areas. Streetmen says they could be launched into space and deployed to study these events.

What about space: Chipsats can conceivably be blasted in droves into deep space or through Europa's icy atmosphere to study whether there is life there. Lucia hopes to create a robust system for these harsh conditions. "It enables a radically different spacecraft design," Streetman says.