For the past 10 months, Amazon Web Services has been running data through its cloud-based software platform on what’s arguably the world’s edgiest edge: a satellite in low Earth orbit.
The experiment, revealed today during AWS’ re:Invent conference in Las Vegas, is aimed at demonstrating how on-orbit processing can help satellite operators manage the torrents of imagery and sensor data generated by their spacecraft.
“Using AWS software to perform real-time data analysis onboard an orbiting satellite, and delivering that analysis directly to decision makers via the cloud, is a definite shift in existing approaches to space data management,” Max Peterson, AWS’ vice president of worldwide public sector, said today in a blog posting. “It also helps push the boundaries of what we believe is possible for satellite operations.”
AWS’ experiment was done in partnership with D-Orbit, an Italian-based company that focuses on space logistics and transportation; and with Unibap, a Swedish company that develops AI-enabled automation solutions for space-based as well as terrestrial applications.
Software tools from AWS — including the company’s machine learning models and AWS IoT Greengrass — were integrated into a prototype processing payload built by Unibap. That payload was then placed on D-Orbit’s ION satellite carrier. The ION spacecraft was one of scores of spacecraft sent into orbit aboard a SpaceX Falcon 9 rocket in January. A few weeks after deploying its satellites, D-Orbit’s ION carrier ramped up data processing operations on the payload using AWS’ software.
During the D-Orbit ION experiment, the team applied various machine learning models to satellite sensor data to identify specific types of objects in the sky, such as clouds and wildfires, as well as terrestrial objects including buildings and ships.
AWS said that its artificial intelligence and machine learning services helped reduce the size of images by up to 42 percent, resulting in increased processing speeds. The AI system could decide in real time which satellite images should be given high priority for downlinking, and which images could be set aside.
The team also modified the process for sending data to and from the satellite, to build in more tolerance for communication delays. The modification made it simpler to manage file transfers automatically, without having to manually process downlinks over multiple ground station contacts.
This isn’t AWS’ only foray into space-based cloud computing: Amazon sent an edge-computing device known as an AWS Snowcone to the International Space Station in April during Axiom Space’s first private space mission. (Axiom Space is also partnering with Microsoft Azure and LEOCloud on a separate project to put cloud infrastructure in orbit.)
D-Orbit’s vice president of commercial sales, Sergio Mucciarelli, said there’s a lot of value in being able to process data in space.
“Our customers want to securely process increasingly large amounts of satellite data with very low latency,” Mucciarelli explained. “This is something that is limited by using legacy methods, downlinking all data for processing on the ground. We believe in the drive toward edge computing, and we believe it can only be done with space-based infrastructure that is fit for purpose, giving customers a high degree of confidence that they can run their workloads and operations reliably in the harsh space operating environment.”
Fredrik Bruhn, Unibap’s chief evangelist for digital transformation, said his company wants to help customers turn raw satellite data into “actionable information that can be used to disseminate alerts in seconds.”
“Providing users real-time access to AWS edge services and capabilities on orbit will allow them to gain more timely insights and optimize how they use their satellite and ground resources,” Bruhn said.
AWS, Unibap and D-Orbit are continuing to test new capabilities on the ION platform — including new approaches for processing raw data on orbit, as well as more refined methods for data delivery. If the experiment bears fruit, it could soon become routine for satellites to make sense of what they’re seeing before they downlink the data.