Computing at the edges of the Earth—and beyond

Smaller than a playing card, a ChipSat is a tiny new type of satellite, equipped to sense its environment, compute, and communicate. A 1.5x1.5x0.4 centimeter ChipSat costs about $50 to make—compared to thousands for an already small, but more widespread 10x10x10 cm CubeSat. Their small size and low cost make it possible to fly swarms of hundreds or thousands of ChipSats in Earth’s orbit to sense signals and report them back to Earth to support important missions like climate science, space weather, national security, and disaster response.

With his graduate students at Carnegie Mellon University, Brandon Lucia, assistant professor of electrical and computer engineering, has developed new hardware and software that enables reliable sensing and processing onboard these necessarily simple, tiny satellites. Working together with ChipSat creator and mission lead Zac Manchester at Stanford University, Lucia’s lab launched several ChipSats on March 18 of this year as part of NASA’s KickSat-2 mission. Those ChipSats were part of a swarm of 100 that deployed to low Earth orbit, representing the first-ever successful mass deployment of ChipSats.

In addition to the innovation of their tiny size, low cost, and novel deployment mechanism, ChipSats are an instance of a new area of systems computing research Lucia terms “orbital edge computing.” Edge computing is a technique that allows sensor data to be processed by computers in the same place where it was collected. Edge computing avoids sending data to the cloud, saving time and energy. Orbital edge computing is the concept of edge computing applied to space-based systems. Sensor-equipped nanosatellites—including ChipSats, CubeSats, or both—are outfitted with computing hardware to process the sensor data they collect.

Lucia’s lab is developing hardware and software computer systems to add to ChipSats and CubeSats that will enable orbital edge computing. These new systems apply sophisticated machine inference to the sensor data that they collect and can identify sensed signals of interest without the need to activate the satellite’s power-hungry radio. Orbital edge computing satellites collect the energy that they use from their environment and only operate after collecting enough energy, requiring specialized software and hardware that react to energy availability. Flying in formations distributed around Earth’s orbit, ChipSats and CubeSats can share the work of collecting and processing sensor signals, allowing a constellation of orbital edge computing nanosatellites to support missions that are impossible with today’s satellites.

The future for ChipSats and CubeSats is bright, with plans for more launch missions and increasingly capable sensing and computing hardware. Using the support for orbital edge computing developed in Lucia’s lab, constellations of nanosatellites will one day be able to sense their environment from above and use sophisticated machine learning to answer key questions about our world.