LEO Trek is a research and development project conducted at the (DSG) at TU Wien.
The project team is composed by Asst. Prof. Nastic, Cynthia Marcelino, and Thomas Pusztai.
LEO Trek aims to provide open and reusable building blocks for computing across the Edge, Cloud, and Space 3D continuum.
LEO Trek is an open source toolkit for orchestrating, simulating, and optimizing serverless and AI workflows in the Edge Cloud Space 3D Continuum. It is composed of multiple modular software components that can be used independently or combined depending on the use case.
-
Stardust
A scalable simulator for the 3D Continuum
https://github.com/polaris-slo-cloud/stardust-go/ -
HyperDrive
An SLO-aware serverless scheduler for the 3D Continuum
https://github.com/polaris-slo-cloud/hyper-drive/ -
ChunkFunc
A workflow resource optimizer accounting for input size and cost performance trade offs
https://github.com/polaris-slo-cloud/chunk-func/ -
FedCCL
A federated learning framework
https://github.com/polaris-slo-cloud/fedccl -
Databelt
A state management framework for serverless workflows in the 3D Continuum
https://github.com/polaris-slo-cloud/databelt/ -
Gaia
A serverless runtime for automated CPU and GPU selection for serverless AI
https://github.com/polaris-slo-cloud/gaia/
Each module is self contained and includes its own documentation, build instructions, and usage examples.
LEO Trek targets:
- Researchers studying distributed systems, serverless computing, and edge or space computing
- Platform engineers designing resource aware and SLO driven execution platforms
- Developers and third party contributors building extensions, schedulers, runtimes, or optimization mechanisms
The toolkit enables reproducible evaluation, system prototyping, and extension of state of the art approaches for heterogeneous and dynamic environments.
LEO Trek is structured as a collection of modular components covering simulation, scheduling, optimization, state management, and hardware acceleration within the 3D Continuum. Users can run simulations and performance evaluations, while developers can extend individual components or integrate them into existing platforms.
Each component is provided as a standalone open source repository and contains all information required to build, configure, use, and further develop the software.