[time-nuts] Technical Paper: GPS Receiver Impact from the UTC Offset (UTCO) Anomaly of 25-26 January 2016
magnus at rubidium.dyndns.org
Wed Nov 2 17:28:43 EDT 2016
On 11/02/2016 09:11 AM, Martin Burnicki wrote:
> There's now a paper with an analysis of the problem available:
Many thanks for that link.
Some background can be found in my report:
Martins email with details to the list is also good background, which
well reflects the high degree of input that Meinberg provided as input
to the USCG CGSIC/NAVCEN.
When I wrote my report, I did not have all the hard facts available, but
these folks have access to data which may be hard to find even if published.
In my report I propose that algorithms shall be described to make GPS
receivers more resilient, and that the GPS control segment shall include
data checks to validate uploads. It is with great satisfaction that I
find that both these is included in the recommendations that their paper
I have seen one proposed algorithm for improving resilience, and I
provided feedback on it. The basic flaw the proposed algorithm had a
focus only on the issue of data. This is indeed a good thing to
discriminate and look at. However, doing it single-minded would
potentially make the situation worse. Announcements also needs to be
discriminated based on wildly incorrect data. The article still focuses
on the data-set fit interval, which also needs to be honored but also
seems to acknowledge the parameter range invalidation. I think it can be
Details aside, we can look forward to an improved GPS system and
improved GPS signal specification. For the later, we shall start to ask
receiver-vendors for their time-plan to upgrade receiver firmware to
include those improved algoritms, and this is essentially the third
recommendation in my report.
There is a certain amount of politics in this naturally, some of it
quite understandable in that they want to defend the views of the GPS
system. Some wordings is maybe toning impact down a little more than I
think it wise, but for many aspects several important services was
indeed not affected, such as navigation (if correctly implemented using
only GPS-time in receivers). I think that 36 years of operation, a
problem every blue moon is not really a big problem, it is if it is not
taken good care of. While I am not yet fully satisfied, I see this as a
good step in the right direction.
More information about the time-nuts