As more and more satellites continue to increase the TRL of highly integrated software defined technologies which is providing enormous amounts of data and spacecraft configurations based on this data. This large data set and dependent outcomes problem is setting the stage for AI/ML to provide multiple solutions. From slight incorporation AI/ML solution in ground stations to help parse large amounts of data to full AI/ML incorporation on orbit with the ability to autonomously reconfigure the spacecraft based on scenarios and commands. A space news article titled ‘Improving Space AI: Ground-to-orbit efforts aim to advance satellite intelligence‘ the author writes about the AI-driven technologies and their incorporation in satellites and ground stations.1
The author asserts that with the incorporation of software defined technologies, from radios to full satellites, provides the operators the ability to reprogram technologies on orbit. This on-orbit re-programmability means that satellites capabilities and configuration is not fully set before launch.
The author also writes that AI/ML incorporation on orbit will have its first hurdle in the fact that real world training data is not available yet so simulated training data will have to be used. This could result in flight data being completely different.
From a cyber perspective this transition period and incorporation of new technology is ripe for injecting false data, false commands, and injecting malware in memory space that re-programability comes with. Component through Satellite developers and satellite operators should ensure cyber security training, best development practices, and best maintenance practices and plans are in place and adhered to in order to reduce the likely hood of cyber attack from happening.
1 – https://spacenews.com/improving-space-ai-ground-orbit-efforts-aim-advance-satellite-intelligence/