The GSOC satellite telemetry analysis framework

Abstract

Given the huge amount of data a modern spacecraft sends back to Earth, monitoring its status without the aid of automatization has become an almost unfeasible task. To assist engineers, the Automated Telemetry Health Monitoring System (ATHMoS) is being developed at the German Space Operation Center. It employs a semi-supervised machine learning approach to automatically detect novel behavior or outliers in satellite telemetry. The algorithms and models of ATHMoS are further designed to detect slowly varying trends in telemetry to rise awareness of potential problems of a satellite at an early stage. Since expert knowledge of engineers is indispensable for the correct assessment of certain behaviour of telemetry data and for contextualizing it, the automatically trained ATHMoS models can be augmented by engineers. To ease the interpretation of ATHMoS results and to give engineers a powerful and comprehensive platform for their investigations, ATHMoS is embedded in the new Visualization and Data Analysis (ViDA) framework. ViDA further allows engineers to access and plot all telemetry data sent to ground and to run customized analyses.