Just after 8 am AEDST (UTC+11) on Wednesday, 2 April 2014, all our CORS started showing issues with tracking. Pretty soon it became obvious that this was related to GLONASS.
The Trimble Pivot software provided a bit more clue into what was happening. Pivot uses both broadcast and predicted ephemeris and under normal operating condition the difference between these are usually in the order of several meters. However, during the outage the differences were extremely large as shown in the following screenshot.
Such anomaly indicates invalid broadcast ephemeris which then made the receivers unable to track the GLONASS satellites properly.
This is further confirmed by Russia’s Federal Space Agency’s own monitoring system, showing most of the satellites broadcasting Illegal Ephemeris.
It took a while for GLONASS operators to fix the issue. This was apparently because they had to wait until those satellites became visible above Russia again so they can upload the correction from the ground stations.
By about 5 pm AEDST (UTC+11) all GLONASS satelites were shown with valid broadcast ephemeris again in the Trimble Pivot software.
The American Surveyor reported that, “it was nothing nefarious. An engineer simply uploaded an incorrect ephemeris and as soon as it was noticed, they started rectifying the situation by uploading the correct one.”
Altus Positioning Systems provided a technical explanation that was reported by GPS World. “Our analysis reveals the GLONASS integration algorithms skipped an interval of around 1.5 minutes at the control centre software At 21:00 UTC April 1, all GLONASS satellites received an orbit state (ephemeris) which was clearly several minutes ahead of the current orbit shape without actually changing the applicable reference time stamp. In other words, future orbit-position, velocity and accelerations were assigned to a current reference time stamp. This led to incorrect orbit positions for all GLONASS satellites and subsequent problems with receiver using GLONASS measurements.”