ClearSpace to use EPFL software to protect satellites
A convolutional neural network developed in the Embedded Systems Lab (ESL) in the School of Engineering will be implemented by EPFL spin-off ClearSpace to shield satellite communications from interference by solar radiation.
All satellite communication systems are vulnerable to hackers by definition, and are protected with both hardware and software. However, because they reside outside the protection of the Earth's atmosphere, satellites are perhaps exposed to the mightiest and most unpredictable hacker of all: the Sun.
"The Sun's radiation can affect the data being processed in any space system," explains Rubén Rodriguez Álvarez, a PhD student in the Embedded Systems Lab (ESL) led by David Atienza. "The radiation causes bits of binary data to flip, polluting your data with interference, and you might not even know that this is happening."
One way to protect a system from radiation would be to install reflective shields, but this is exactly the kind of heavy-duty solution that satellite engineers are keen to avoid. An idea from recent ESL graduate Flavio Ponzina, now a postdoctoral researcher at the University of California, seemed to provide a better way using software. The technology, E2CNN, provides a solution to the problem of radiation interference without an increase in overhead.
"We take a neural network, prune it carefully, and then send multiple copies of it down separate channels, for the sake of redundancy," explains Ponzina. "If radiation or any other kind of interference affects the information in one of these channels, the other ones will be given credibility over the one with the corrupt data.”
I think it is really beneficial for EPFL scientists to see that the research they do can be applied to the real world by collaborating with companies like ClearSpace that work on the deployment of aerospace systems
In two case studies applying E2CNN to positional data for satellite navigation, the researchers found that there was no increase in processing speed, but there was an increase in accuracy — even without radiation interference. The work was presented at the SPAICE Conference 2024 in September.
EPFL spin-off ClearSpace intends to use E2CNN technology on its pioneering mission to remove debris from Earth’s orbit, ClearSpace-1, which take off sometime in the next few years. Now that Ponzina has left EPFL, Álvarez is continuing the project, implementing the software on field programmable gate arrays (FPGAs): processors that can change shape according to the requirements of the application. In this case, they adapt wonderfully well to the sending of neural networks down multiple parallel channels.
Jacques Viertl of ClearSpace is clear on the benefits of this system. "We asked for an implementation on FPGAs because of the redundancy they offer. Whenever you deal with images there is a trade-off between quality and overhead, but in the case of the visual recognition of targets by ClearSpace-1, the processing of the information will be protected by the multiple channels being used, at no extra cost in processing power.”
Atienza adds that for the ESL, the collaboration ensured that the lab’s research in edge AI and smart embedded systems can be applied to space technology. “I also think it is really beneficial for EPFL scientists to see that the research they do can be applied to the real world by collaborating with companies like ClearSpace that work on the deployment of aerospace systems,” he says.
Using ensemble learning to improve radiation tolerance of CNNs in space applications
This article orignally appeared on the website of the EPFL EcoCloud center: ClearSpace will use Minority Report technology from EPFL