California startup aims to give satellites minds of their own
By Paul Marks|May 7, 2021
If a cyberweapon crept into a satellite’s electronics, the consequences could be devastating: The attacker could discharge the spacecraft’s batteries, perhaps, or vent all its propellant, wrecking its mission. So it’s vital any such intrusion is automatically detected, and the attack malware safely quarantined, as fast as possible. But how? Running intrusion detection algorithms across satellite telemetry from ground stations probably would detect the intrusion too late, given the signal latency.
What’s really needed to reinforce satellite cybersecurity, the San Francisco-startup Loft Orbital argues, is fast, localized computational power installed on the spacecraft itself. The technology could have other benefits: With this data processing power, a satellite could make quick, autonomous decisions about changing its trajectory to sense and avoid space debris. And data acquired by the satellite’s sensors could be processed on board, too.
Such arguments have now swayed both the U.S. Space Force’s Space and Missile Systems Center in California, and AFWERX, the U.S. Air Force’s innovation incubator and accelerator — and on May 5 the organizations awarded Loft Orbital a $1.5 million contract to press ahead and develop a high-performance, reprogrammable, onboard data processor that would lend satellites the number-crunching power they have lacked to date.
“This onboard processor will function as a ‘brain’ on Loft Orbital spacecraft, providing an environment to host applications for sensor data processing, mission autonomy, satellite self-protection, or any other application requiring advanced computing,” the firm said in a statement announcing the Small Business Innovation Research grant.
Loft Orbital’s plan builds on its space-infrastructure-as-a-service business model, says spokesman Mitchell Scher. That model involves buying a commodity satellite bus and then selling payload space on it to operators who want to fly sensors and novel software algorithms — known as massless payloads — in Earth orbit. To boost on-board processing power, Loft will take one of its payload slots to test its new hardware architectures.
But supplying faster processors in space is not as simple as it sounds: Processor power in space is limited to decades-old (for which read glacially slow) generations of microprocessors that have been radiation hardened. “Current processing and memory technologies are susceptible to the effects of the space radiation environment through temporary or permanent upsets caused by interacting with energetic solar or cosmic particles,” says Martin Sweeting, founder and executive chairman of Airbus-owned Surrey Satellite Technology in the United Kingdom, a maker of microsatellites.
But rather than making newer, faster microchips radiation resistant, Loft Orbital is planning to leverage in space a resilient technology that is taking hold in commerce and industry on Earth: edge computing. In this, processing tasks that must be done at speed, and securely, are moved away from cloud data centers to the “edge” — the computer industry’s jargon for processing data locally, on a firm’s premises. But instead of computing on any premises, Loft Orbital will deploy edge processors on satellites. These, the company says, will be ones already proved resilient in applications and military ground vehicles.
That edge approach makes sense, says Sweeting: “The space radiation environment’s impact can be mitigated by resilient architectural spacecraft designs that can include edge computing methods that effectively spread the risk through reconfigurable systems. Not routing everything through a central network helps.”
Loft Orbital is not the first to consider edge computing on orbit. DARPA is exploring the idea, too, proposing an edge-based artificial intelligence system called Pit Boss to control its Blackjack network of agile smallsats, in missile warning applications.
But beating them both was Hewlett-Packard Enterprise of Houston, which already has an edge-based computing system in operation aboard the International Space Station. Called HPE Spaceborne Computer-2, the system is reducing data processing times on ISS experiment data “from months to minutes,” according to the company.
Mark Fernandez, a solutions architect at HPE in Greeneville, South Carolina, thinks the future is bright for such technology. “Having reliable edge computing capabilities in space will unlock new opportunities when more data can be processed faster, and in real time, versus relying on transfer of data to and from Earth,” he says.