netidee Projects: LEO Trek
As part of the LEO Trek project at TU Wien, open building blocks for the edge-cloud-space continuum and a toolkit for executing, simulating, and optimizing AI and serverless workflows were developed.
LEO Trek (Low Earth Orbit Trek) is aimed at researchers, platform engineers, and developers of distributed systems. It includes modular components such as Stardust (simulation), HyperDrive (SLO-aware scheduler), ChunkFunc (resource optimization), FedCCL (federated learning), Databelt (state management), and Gaia (serverless runtime with automatic CPU/GPU selection). All modules can be used independently and combined.
A key application scenario is disaster relief, where large amounts of Earth observation satellite data must be quickly combined with in-situ data. Since downloading this data is time-consuming, processing it directly in low Earth orbit enables faster analysis and the timely provision of critical information to emergency responders.
May 2026
More information