Log analysis for anomaly prediction in ground segment space systems

The increased complexity of modern software systems has contributed toward expanded and challenging maintenance and debugging processes of the potential failures. The importance of these processes increases with the level and segment at which these systems operate: space, finance, and healthcare sectors are a few examples of systems where faults should be avoided at all costs and actions should be taken to deem the systems usable in case of software failures. Ground segment systems for satellite and mission operations of the space sector belong to this category, where the potential failures must be avoided as early as possible and specific action taken to redeem or prevent them. Anomaly prediction is an engaging concept that can help the maintenance and control of such systems by predicting when they enter fault state. However, it poses a challenging task as the complexity of the systems increases and multiple components or sub-systems are involved in the chain. The ground systems are representative of these complex systems and its operatives can benefit from the prediction and notification of impending anomalies or failures. This can lead to successful plan of actions to be undertaken to avoid problems before they occur and to reduce the maintainability costs if the alert of anomalies and problems could be automated. Within the ground segment space systems only root-cause analysis was employed, where failures of systems were analysed to determine the cause of the errors. The innovation of this approach is to predict ground system failures by learning a model of the system behaviour.

Recommended citation: Petreski, H., Veskos, P., Santos, R., Psaroudaki, E. & Prasad, R., (2022). Log analysis for anomaly prediction in ground segment space systems. In 2022 25th International Symposium on Wireless Personal Multimedia Communications (WPMC) pp. 447-452. IEEE. doi: 10.1109/WPMC55625.2022.10014733.
Download Paper