Nebula Public Library

The knowledge bank of ESA’s R&D programmes

SaasyML: Onboard Machine Learning Software As A Service For Experimenters

Programme Reference
Start Date
End Date
SaasyML: Onboard Machine Learning Software As A Service For Experimenters

The 'SaaSy ML' project will provide experimenters with on board Machine Learning (ML) functionalities that any app can subscribe to via a Software as a Service (SaaS) app hosted on the Satellite Experimental Processing Platform (SEPP). The ML features provided by the SaaS app will cover both training and prediction operations. The JSAT open-source java library for ML will be re-used thus making accessible over 100 training algorithms onboard a flying mission. Re-using the JSAT library will ensure the development to focus on the idea's novel aspect: a spacecraft SaaS app for in-orbit ML.

Past experiments have successfully implemented ML onboard ESA's OPS-SAT satellite but despite their innovations they have not offered any re-usability beyond the scope of the experimenters’ own research objectives. SaaSy ML's service-oriented approach will spare the experimenters the complexities of having to implement their own data provisioning and ML solutions so that they can focus instead on their experiments’ objectives. Further novelty will also be introduce onboard a spacecraft by designing a plugin architecture that will allow experimenters to inject custom code that address specific ML needs (e.g. calculating target labels/classes during supervised learning training operations). Lessons learned from developing a data provisioning service will be applied so that the SaaS design also includes a data subscription service to feed training data directly into an experimenter’s app. An app will be able to register to SaaSy ML's data feed and pull selected training data from any of the spacecraft’s instruments or its OBSW datapool. An additional novelty will be that of concurrent serviceability in a space segment software. The app will be designed to take advantage of JVM’s Thread Pool implementations as well as the SEPP’s dual core processor so that non-blocking ML training and prediction operations can run in the background while multiple researcher apps interact with SaaSy ML.

Technology Domain
Competence Domain
Executive summary
Final presentation