- Starion Technologies BV has been awarded additional funding to test a very low power artificial intelligence (AI) processor onboard the European RF Spectrum Monitoring Service (ESMS) SmallSat.
- The new onboard data processing unit, STERNA from EDGX, harnesses NVIDIA GPU technology to process vast volumes of data in orbit, vastly reducing latency, minimising downlink costs, and enabling faster, mission-critical decision-making.
Starion Technologies BV and the ESMS consortium have been awarded additional funding from the Belgium Science Policy Office (Belspo) to test a new onboard computing technology on the Belgium-funded ESMS SmallSat, which is potentially due for launch in 2027. Funding from Belspo is being provided through the European Space Agency’s Advanced Research in Telecommunications Systems (ARTES) 4.0 programme.
The ESMS SmallSat will provide an innovative monitoring service to support regulated use of the RF spectrum, delivering data to the relevant authorities and authorised users to enable them to identify illegal or unauthorised activities on land, at sea or in the air.
The new AI technology has been developed by EDGX, a Belgian company preparing a low Earth orbit (LEO) SmallSat mission to demonstrate its advanced onboard data processing unit. At the core of the system is EDGX’s GPU-based computing platform, designed to deliver high-performance AI processing directly in orbit while operating within the strict power constraints of SmallSats.
Conventional onboard processing solutions often struggle to combine performance with power efficiency. EDGX’s platform addresses this challenge by enabling intelligent data filtering and processing in space, significantly reducing the need for downlink capactiy and improving mission responsiveness.
For this mission, EDGX will also evaluate a low-power neuromorphic module as an expansion card to the system, exploring further optimisations in energy management within the overall system architecture.
Alastair Pidgeon, Vice-President Business Development and Technology at Starion, said: “We are very happy to trial this new technology on the ESMS mission. EDGX’s new technology has the potential to be transformative for satellite missions. Some satellites can collect up to 40 terabytes of data, far more than can be downloaded. So instead of transmitting raw data, the new AI processor can extract snippets of information and download that instead.
“To support this, the ESMS team is exploring how they can further compress and optimise the downloadable data. One of the roles of ESMS is to locate where any signal interference originates, and the AI processor could help us obtain this data more quickly from the satellite.”
Because ESMS is already in development, the consortium is collaborating with EDGX to test the AI processor on the ground through experimental campaigns prior to the launch. These tests will continue over the coming months to refine the technology and evaluate its performance. The processor also incorporates some conventional processing capability, which may provide additional benefits to the satellite.
EDGX plans to fly-test the AI processor, which is roughly the size of a pack of playing cards, on a launcher early this year, mainly to verify its resilience in space. ESMS will be the first operational mission to use this technology.
