NavList:
A Community Devoted to the Preservation and Practice of Celestial Navigation and Other Methods of Traditional Wayfinding
Re: Possible GPS Date Rollover problems week of April 6, 2019
From: Peter Monta
Date: 2019 Apr 7, 15:55 -0700
From: Peter Monta
Date: 2019 Apr 7, 15:55 -0700
Paul Hirose writes:
A week number rollover occurs every 1024 weeks, and this will be the second rollover in GPS history, so add 2 * 7 * 1024 days to obtain JD 2 458 580.5 + 19s TAI = 2019 April 7 00:00:19 TAI.
The newer GPS navigation-data format, CNAV, allocates 13 bits to the week number, so that extends the rollover interval to 8192 weeks. The first one happens on January 6, 2137, at the start of GPS week 8192. Mark your calendars.
CNAV is still in testing mode, I think, so very few receivers are using it as their primary source for ephemeris data. (About two-thirds of the GPS constellation is currently transmitting CNAV, using the L2C signal.) Galileo seems to have a 4096-week rollover. I haven't checked the GLONASS or BeiDou ICDs, but if their data is on a different cadence, perhaps taking the least common multiple of all the GNSS data together could extend the unambiguous interval further. (It is vanishingly unlikely that any GNSS firmware will go to that trouble.)
Speaking of CNAV and celestial navigation, some time back the CNAV earth-orientation data was mentioned on NavList. Recently there was an article somewhere (GPS World?) mentioning that the primary users of that data would be space vehicles. Makes sense. If you're in LEO, let's say, then you want to know where to point your camera, and that depends on earth rotation. Effectively this GPS metadata is defining an inertial coordinate system by proxy.
Cheers,
Peter
