Detection and Identification of Faults in Clock Ensembles

Kurzfassung

The quality of the positioning and time dissemination services provided by Global Navigation Satellite Systems is tightly coupled with the provision of a stable system time scale. The robustness of time generation is important to ensure the continuity of service without performance degradation. A well-established method to generate a stable and robust time scale is the composite clock approach. Here, the contributions from each clock are combined to generate a weighted average, the so-called Implicit Ensemble Mean. A clock ensemble offers an enhanced robustness with respect to a system depending on a master clock, since the IEM is less sensitive to failures of single units. The time scale generation can be monitored by employing Fault Detection and Identification techniques. These are based on statistical tests designed to detect and identify faults at single clocks. In this work, these techniques are applied to provide statistically rigorous fault detectors for an ensemble of clocks. The clock composition algorithm can be equipped with a number of statistical tests, which are applied to detect and identify abrupt events (e.g. phase and frequency jumps) and slow deviations from the nominal expected clock behavior.