[aprssig] GPS Epoch Rollover Findings

[Greg Troxel]

"Stephen H. Smith via aprssig" <aprssig at lists.tapr.org> writes:

> GPS expresses the current time and date with a?? series of binary
> counters.?? This included a 10-bit binary counter for the current
> week, meaning that it could only count 1024 (decimal) weeks (just
> under 20 years) before rolling over from 1111111111 to 0000000000, and
> starting over.?? This 1024-week cycle is known as an "epoch".??  The
> previous epoch ended on April 6th of this year.?? As of April 6th, a
> 16-bit counter for weeks replaced the 10-bit one.?? This change will
> put off the NEXT epoch rollover for 65,535 weeks - about 1200 years.

This is not quite accurate.  The original on-air format is to send the
lower 10 bits of the week number, which I think is just a number, and
hence 2047->2048 had a rollover (as did 1023->1024).  It has always been
required in the ICD that receivers disambiguate the transmitted week
number field back to an untruncated week number.

I expect that when the W3K arrives (late 2038), there will still be
receivers operating that do not understand the extended format.  (I have
a mid-90s receiver that I updated with a new firmware ROM in 1999 for
W1K, and it was still around for W2K (and did fine).  I think it's
pretty likely it will make it to W3K.)  But I expect W3K to be more of a
discussion among retronavigation enthusiasts than a real issue.

Also, I had the impression that the new transmission format was 13 bits,
not 16, and that it started a while ago.

https://www.spirent.com/blogs/positioning/2018/january/gps-rollover-week
https://www.usno.navy.mil/USNO/time/gps/gps-week-number-rollover

73 de n1dam