SpaceX has disrupted the traditional aerospace playbook by deploying over a million consumer-grade Linux computers on their Starlink satellites. Rather than opting for the conventional $10,000 radiation-hardened chips, SpaceX embraced commercial silicon running real-time Linux. This approach has not only reduced costs but has also proven remarkably effective in a high-stakes environment like space.
Traditional aerospace has relied on radiation-hardened CPUs that are both costly and slow—akin to computers from the 1980s like 386 or 486 machines. These components use thick traces and large transistors to withstand cosmic radiation, but they come with a hefty price tag and aren’t particularly speedy. SpaceX, however, chose a different path by leveraging consumer-grade computing power and implementing a backup strategy to manage potential issues like cosmic-ray-induced bit flips.
Based on content from Speedify
At the heart of Starlink’s success is its innovative three-computer voting system. Each satellite hosts three computers for critical computations. This redundancy allows Starlink to tackle cosmic ray bit flips by having the computers compare calculations and vote on the most consistent results. If one computer malfunctions, the remaining two can override the errant output, ensuring the system remains functional. This setup is a clever middle ground, providing reliability without the complexity and cost of more extensive systems like those used by the space shuttle, which employed five computers for navigation tasks.
Starlink’s system also supports an agile development process with weekly software updates, a rarity in the space industry. By updating a small number of satellites initially and progressively rolling out changes, SpaceX minimizes risks of bricking any part of their fleet as they zoom around Earth at 17,000 mph. This method demonstrates a fast-paced, iterative approach to satellite maintenance and evolution that stands in stark contrast to the bespoke, named systems of the past.
An intriguing aspect of SpaceX’s approach is how it navigates the use of Linux under the GPL license. By keeping the Starlink source code private, they’re able to maintain proprietary advantages because the GPL’s source code distribution requirement activates only when the software is sold, not when it’s merely launched into orbit. While they contribute back certain patches to the open-source community, their vertical integration means that they are in control of both product and process, crafting even the tools to build their factories.
The cosmology of bit flips—changes in memory value due to radiation—is a frequent concern for satellites. With a multitude of systems and ample memory, such events can happen anywhere from ten to over a hundred times daily. Systems like ECC (Error-Correcting Code) memory enhance reliability by adding error correction to the mix, a standard in server hardware and high-end desktops but even more crucial in the unyielding domain of space.
SpaceX’s Starlink initiative not only highlights a triumph in satellite technology but also spotlights how innovations in commercial computing can set the pace in space exploration. The clever blending of Linux’s robustness, modular software updates, and smart system redundancy create a model that others in the industry are sure to study closely.
