Page 1 of 1

GPS Rollover incidents

Posted: Mon Apr 08, 2019 12:35 am
by seat64k
You may be aware of the GPS rollover incident on 6 April. Short description: Date is expressed as a week, and day of the week, both represented by a number. The week is a 10-bit number, so every 1024 weeks (just short of 20 years), it flips over to 0. Devices that don't know this will teleport you back to 1999.

https://twitter.com/ChinaAvReview/statu ... 8919411712

Image

Not sure how real this is, but the flight does appear to be 789 on that day:

https://uk.flightaware.com/live/flight/CES567

Any other events?

Re: GPS Rollover incidents

Posted: Mon Apr 08, 2019 12:48 am
by GalaxyFlyer
Yeah, it happens. There’s been other rollover effects with dates and times. It’s fixed with a NOTAM and a procedure. No one goes back in time. Yawn.

GF

Re: GPS Rollover incidents

Posted: Mon Apr 08, 2019 12:59 am
by northstardc4m
GalaxyFlyer wrote:
Yeah, it happens. There’s been other rollover effects with dates and times. It’s fixed with a NOTAM and a procedure. No one goes back in time. Yawn.

GF


Well dang... and here I was hoping to go back and make a fortune in BitCoin! :P

Re: GPS Rollover incidents

Posted: Mon Apr 08, 2019 5:37 pm
by LH707330
Holy crap, that's a gnarly headwind.

Re: GPS Rollover incidents

Posted: Mon Apr 08, 2019 6:43 pm
by XAM2175
Amusingly, there are two main approaches to avoiding this in future:
* Modernised GPS uses a 13-bit week counter, allowing for just shy of 158 years between roll-overs. Galileo and Beidou are similar.
* GLONASS rolls-over every four years, in the assumption that something that happens so frequently will be accounted for in the design of even the cheapest receivers :lol:

Re: GPS Rollover incidents

Posted: Mon Apr 08, 2019 11:47 pm
by Francoflier
You'd think the GPS receiver on a 787 would be modern (and expensive) enough to have a software workaround for this.

LH707330 wrote:
Holy crap, that's a gnarly headwind.

140 kts? Meh. Try flying westward around eastern Asia or continental US in the winter for that "I might die of old age before we get there" feeling...
;)

Re: GPS Rollover incidents

Posted: Tue Apr 09, 2019 1:46 am
by GalaxyFlyer
It’s not in the FMS, the rollover is in the NAVSTAR system.

Re: GPS Rollover incidents

Posted: Tue Apr 09, 2019 5:08 pm
by Chemist
Francoflier wrote:
You'd think the GPS receiver on a 787 would be modern (and expensive) enough to have a software workaround for this.

LH707330 wrote:
Holy crap, that's a gnarly headwind.

140 kts? Meh. Try flying westward around eastern Asia or continental US in the winter for that "I might die of old age before we get there" feeling...
;)


That's nothing. Try flying westward over the New Mexico Rockies in a Cherokee 140 and seeing 60kts on your GPS, and watching semi-trucks on the interstate below going faster than you.

Re: GPS Rollover incidents

Posted: Tue Apr 09, 2019 6:03 pm
by nws2002
FAA released an InFO about this last week.
https://content.govdelivery.com/attachm ... O19005.pdf

Re: GPS Rollover incidents

Posted: Tue Apr 09, 2019 6:12 pm
by zeke
GalaxyFlyer wrote:
It’s not in the FMS, the rollover is in the NAVSTAR system.


It is an onboard problem, fact is most aircraft (and other GPS receivers) were able to handle the week cycle change without falling over.

The three common receiver offset methods to overcome this are a battery backed up real-time clock, store the reference year in non volatile RAM, or use a date values stored in the receiver software (the software gets update from time to time). All three methods will allow math receiver to select the correct week cycle.

It is downright embarrassing for the 787 to go back to 1999.

Re: GPS Rollover incidents

Posted: Tue Apr 09, 2019 6:13 pm
by LH707330
Chemist wrote:
Francoflier wrote:
You'd think the GPS receiver on a 787 would be modern (and expensive) enough to have a software workaround for this.

LH707330 wrote:
Holy crap, that's a gnarly headwind.

140 kts? Meh. Try flying westward around eastern Asia or continental US in the winter for that "I might die of old age before we get there" feeling...
;)


That's nothing. Try flying westward over the New Mexico Rockies in a Cherokee 140 and seeing 60kts on your GPS, and watching semi-trucks on the interstate below going faster than you.

The worst one I ever dealt with was 74 GS and 125 TAS in a DA40, that turned into a long day.

Re: GPS Rollover incidents

Posted: Tue Apr 09, 2019 6:20 pm
by litz
Chemist wrote:
That's nothing. Try flying westward over the New Mexico Rockies in a Cherokee 140 and seeing 60kts on your GPS, and watching semi-trucks on the interstate below going faster than you.


Or seeing a fully laden freight train stall out on flat, level, track - purely from headwinds.

Re: GPS Rollover incidents

Posted: Tue Apr 09, 2019 7:55 pm
by XAM2175
It should also be noted that the actual positioning function performed by the receiver isn't affected by the roll-over - only the translation to the calendar day is.

Now, the Year 2038 Problem on the other hand should be a real fun one :P

Re: GPS Rollover incidents

Posted: Tue Apr 09, 2019 10:26 pm
by AirKevin
Chemist wrote:
That's nothing. Try flying westward over the New Mexico Rockies in a Cherokee 140 and seeing 60kts on your GPS, and watching semi-trucks on the interstate below going faster than you.

How fast were the trucks going. 60 knots is roughly 70 miles, so they would have had to be going pretty fast. Some of these trucks are governed at 65 or even less.
zeke wrote:
It is downright embarrassing for the 787 to go back to 1999.

Before the 787 was even around.

Re: GPS Rollover incidents

Posted: Wed Apr 10, 2019 12:52 am
by Chemist
AirKevin wrote:
Chemist wrote:
That's nothing. Try flying westward over the New Mexico Rockies in a Cherokee 140 and seeing 60kts on your GPS, and watching semi-trucks on the interstate below going faster than you.

How fast were the trucks going. 60 knots is roughly 70 miles, so they would have had to be going pretty fast. Some of these trucks are governed at 65 or even less.
zeke wrote:
It is downright embarrassing for the 787 to go back to 1999.

Before the 787 was even around.


The semis were very slowly gaining on us.
It was bumpier in the plane than on the interstate, too.

Re: GPS Rollover incidents

Posted: Wed Apr 10, 2019 7:22 am
by WIederling
XAM2175 wrote:
It should also be noted that the actual positioning function performed by the receiver isn't affected by the roll-over - only the translation to the calendar day is.

Now, the Year 2038 Problem on the other hand should be a real fun one :P


You can handle such problems in a sane way or just ignore them ( when writing software.)
Most domains have tended to just ignore the issues ...

On the other hand I don't see why a 2010+ GPS receiver installed on a plane that
will be used for a couple or three decades should stumble over a well known
thing like roll over in regular intervals.
I always do wonder what other less benign oversights sit in the same code basket?

GLONAS approach ( rather like some other solution used by Linux:
start with negative timer value to make errors come up on boot )
seems to be "a Good Thing (TM)" here.

Re: GPS Rollover incidents

Posted: Wed Apr 10, 2019 7:05 pm
by thepinkmachine
AFAIK it is a bit of a problem, as there’s an unexpected glitch affecting the planes that were powered down during the rollover.

On some frames the date remains stuck in the year 99 affecting things like Nav DB validity, CPDLC and even WX radar - they are all fed by GPS clock and reportedly there’s no simple way to reset it...

Re: GPS Rollover incidents

Posted: Thu Apr 11, 2019 2:45 am
by seat64k
thepinkmachine wrote:
AFAIK it is a bit of a problem, as there’s an unexpected glitch affecting the planes that were powered down during the rollover.


That would explain why this isn't a big news story. If it affected planes that were powered on at the time, it would be the entire fleet (assuming this piece of equipment is common).

Re: GPS Rollover incidents

Posted: Thu Apr 11, 2019 8:00 am
by WIederling
thepinkmachine wrote:
AFAIK it is a bit of a problem, as there’s an unexpected glitch affecting the planes that were powered down during the rollover.

On some frames the date remains stuck in the year 99 affecting things like Nav DB validity, CPDLC and even WX radar - they are all fed by GPS clock and reportedly there’s no simple way to reset it...


Didn't the 787 have another unpleasant software roll over problem coming up in the early years?
https://www.engadget.com/2015/05/01/boe ... ftware-bug
and some others:
https://www.google.com/search?q=787+sof ... +roll+over