There is already a week-based calendar in use

classic Classic list List threaded Threaded
3 messages Options
Reply | Threaded
Open this post in threaded view
|

There is already a week-based calendar in use

Amos Shapir-2
 On Saturday, 6 April, 2019, 23:25 UTC, the GPS week counter field will roll over.
It seems that the GPS system uses a calendar based on a 10-bit counter of week numbers; this means that it has to restart its count every 1024 weeks, or 19.5 years.  It had started to count in Jan. 1980, and had already restarted once, in 1999, so this is the second time this happens.

There were some issues with GPS systems misreading the date the last time it happened, but there are a lot more GPS systems in use now.
--
Amos Shapir
 
Reply | Threaded
Open this post in threaded view
|

Re: There is already a week-based calendar in use

Martin Burnicki
Amos Shapir wrote:
> I saw this on http://catless.ncl.ac.uk/Risks/30/92#subj13 :
>
>  On Saturday, 6 April, 2019, 23:25 UTC, the GPS week counter field will roll over.
>
> It seems that the GPS system uses a calendar based on a 10-bit counter
> of week numbers; this means that it has to restart its count every 1024
> weeks, or 19.5 years.  It had started to count in Jan. 1980, and had
> already restarted once, in 1999, so this is the second time this happens.

Yes, and this is exactly defined in the GPS interface control document.

> There were some issues with GPS systems misreading the date the last
> time it happened, but there are a lot more GPS systems in use now.

Problem related to the GPS weeknumber rollover don't only occur on these
specific rollover dates.

Some manufacturers of GPS devices have implemented a simple comparison
to determine the era of a 1024 week cycle, e.g. if the week number is
above 870 then it's the first era, and if it's below 870, it's the
second era.

Of course this approach fails if the week number in the second era
reaches 870, and week 871 is converted to a date in the first era, which
results in a date 1024 weeks in the past.

Unfortunately you often don't even know which week number has been used
as limit in the implementation, so you can have a bad surprise at a
random weekend wintin a 1024 week cycle.

If you are interested, I've collected some information also here:
https://www.meinbergglobal.com/download/burnicki/gps_week_number_rollover.pdf

Martin
--
Martin Burnicki

Senior Software Engineer

MEINBERG Funkuhren GmbH & Co. KG
Email: [hidden email]
Phone: +49 5281 9309-414
Linkedin: https://www.linkedin.com/in/martinburnicki/

Lange Wand 9, 31812 Bad Pyrmont, Germany
Amtsgericht Hannover 17HRA 100322
Geschäftsführer/Managing Directors: Günter Meinberg, Werner Meinberg,
Andre Hartmann, Heiko Gerstung
Websites: https://www.meinberg.de  https://www.meinbergglobal.com
Training: https://www.meinberg.academy
Reply | Threaded
Open this post in threaded view
|

Week counts RE: There is already a week-based calendar in use

Ryan Provost-2

Based on the consecutive week count, it would be week 2,048 by that time

 

Today (Thu 2018-11-22) Is 5th day week 2028 (Tomorrow Fri 23rd will be 6th day), and this following Sunday 2018-11-25 (based on epoch on Sun 1980-01-06) will be 2049th week

 

However on the millennium epoch of week numbering (since 2000-01-01) This week is week 985 and next week is week 986, despite day count number 6900 for today (Thu 2018-11-22) and week 1000 slated to occur starting on Saturday 2019-03-02 in which also is day 7000 at the same time so a dual milestone! Also hence the 70-day cycle of 7 “metric weeks”/“deci-weeks” (or 10-day periods whatever you name them) or 10 standard weeks it will be cycle number 100!

 

Sent from Mail for Windows 10

 


From: East Carolina University Calendar discussion List <[hidden email]> on behalf of Martin Burnicki <[hidden email]>
Sent: Thursday, November 22, 2018 11:32:03 AM
To: [hidden email]
Subject: Re: There is already a week-based calendar in use
 
Amos Shapir wrote:
> I saw this on http://catless.ncl.ac.uk/Risks/30/92#subj13 :
>
>  On Saturday, 6 April, 2019, 23:25 UTC, the GPS week counter field will roll over.
>
> It seems that the GPS system uses a calendar based on a 10-bit counter
> of week numbers; this means that it has to restart its count every 1024
> weeks, or 19.5 years.  It had started to count in Jan. 1980, and had
> already restarted once, in 1999, so this is the second time this happens.

Yes, and this is exactly defined in the GPS interface control document.

> There were some issues with GPS systems misreading the date the last
> time it happened, but there are a lot more GPS systems in use now.

Problem related to the GPS weeknumber rollover don't only occur on these
specific rollover dates.

Some manufacturers of GPS devices have implemented a simple comparison
to determine the era of a 1024 week cycle, e.g. if the week number is
above 870 then it's the first era, and if it's below 870, it's the
second era.

Of course this approach fails if the week number in the second era
reaches 870, and week 871 is converted to a date in the first era, which
results in a date 1024 weeks in the past.

Unfortunately you often don't even know which week number has been used
as limit in the implementation, so you can have a bad surprise at a
random weekend wintin a 1024 week cycle.

If you are interested, I've collected some information also here:
https://www.meinbergglobal.com/download/burnicki/gps_week_number_rollover.pdf

Martin
--
Martin Burnicki

Senior Software Engineer

MEINBERG Funkuhren GmbH & Co. KG
Email: [hidden email]
Phone: +49 5281 9309-414
Linkedin: https://www.linkedin.com/in/martinburnicki/

Lange Wand 9, 31812 Bad Pyrmont, Germany
Amtsgericht Hannover 17HRA 100322
Geschäftsführer/Managing Directors: Günter Meinberg, Werner Meinberg,
Andre Hartmann, Heiko Gerstung
Websites: https://www.meinberg.de  https://www.meinbergglobal.com
Training: https://www.meinberg.academy