Register | Sign In


Understanding through Discussion


EvC Forum active members: 65 (9162 total)
5 online now:
Newest Member: popoi
Post Volume: Total: 915,808 Year: 3,065/9,624 Month: 910/1,588 Week: 93/223 Day: 4/17 Hour: 1/1


Thread  Details

Email This Thread
Newer Topic | Older Topic
  
Author Topic:   GPS Devices Might Stop Working This Weekend -- Week Number Rollover
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


(2)
Message 1 of 22 (850186)
04-02-2019 8:21 PM


I should start by stating that as a software engineer, retired about a year ago, I worked very closely on a daily basis with several GPS receivers for over 20 years, so I know both personally and professionally that this is a real thing.
You may consider this a public service announcement, which is how Today I Found Out describes having been approached by Tom Tom to make this video, "How Does GPS Actually Work and Why Many GPS Devices are About to Stop Working":
That prompted me to open my GPS Week Numbers list on my phone (which I've rarely done since retiring over a year ago) to verify that, indeed, the GPS Week Number will roll over from 1023 back to zero this weekend, 07 April 2019. Like I said, I know that this is a real thing because I went through it the first time for 22 August 1999 -- kind of a dress rehearsal for the Y2K scare.
Basically, your GPS receiver determines your 3-D position by triangulating off of four satellites (optimally, but three could still work) and the precise time (contained in the satellite signals). Time is expressed in Week Number (WN), the number of weeks since the beginning of GPS time (06 Jan 1980), and Time Of Week (TOW), the number of seconds into the week. Given WN and TOW, the receiver calculates the exact date and time -- there's also a precise one-pulse-per-second signal, 1PPS, that the receiver syncs to to determine fractions of a second.
The problem stems from the WN counter, which is only 10 bits long. That means that it can only count 1024 weeks ranging from 0 to 1023 before the count rolls over (hence the term) back to zero. 1024 weeks amounts to about 19 years 8 months. The standard explanation for such short-sightedness is that nobody had expected the system to be around for so long, especially since it had started out as an experiment.
Part of how the GPS system works is that the orbital parameters for the 31 operational GPS satellites are downloaded into the receiver, so that it can calculate which satellites should be where in the sky at that given time. The receiver then uses that information to acquire satellite signals to use for its position solution. That means that if the receiver doesn't know the right date and time, then it will think that the satellites are where they are not which will completely mess up its position calculations. Bottom line: the receiver cannot work.
The first Week Number rollover, 22 August 1999, kind of took us by surprise. The receiver manufacturers came up with fixes that worked. I just assumed that they knew what to do from this point forward and so, coupled with the knowledge that it would happen again, they would know to always include those fixes in the new designs. Or maybe the manufacturers of products that use GPS receivers (eg, Tom Tom and Garvin) are just nervous and have their own fixes to detect and correct wrong dates coming out of the receiver.
At any rate, we're hearing buzz about the need to be sure that your firmware is up to date. Personally, I doubt that we will see many GPS-based devices fail, but if yours does then you will know what happened and then you can update your firmware. Or you can respond proactively and update before this weekend.
There are a couple more things to think about. This second Week Number rollover will occur at midnight UTC going into Sunday, 07 Apr 2019. Since I'm in the Pacific Standard Time time zone (UTC-8), it will happen on Saturday at 4PM (1600h). Everybody can apply their own time zone's offset for when the rollover will occur.
Second, testing whether our cellphone/mobile survived the rollover or not can be problematic. On the train in Germany, I would run the Maps app on my phone. I wasn't getting enough satellite signals to update my location, so it would just freeze until we passed through a train station at which time it would talk with the station's Wifi network, update to the station's location, and then freeze there. It turns out Wifi signals include location data whether you can connect or not. So if your phone is connected to Wifi, then it will get its location from that regardless of whether the GPS is working. It also gets time information from the cell towers (and maybe also from Wifi?) and might also get location information as well. Though it looks like switching to Airplane Mode will disconnect you from both the cell towers and Wifi, so that should allow you to test your GPS receiver.
I hope that this can be of some help. Even more, I hope that after rollover all your GPS-based devices continue to function without a hitch or hiccup.

Replies to this message:
 Message 2 by Percy, posted 04-03-2019 10:27 AM dwise1 has replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


(2)
Message 3 of 22 (850203)
04-03-2019 12:32 PM
Reply to: Message 2 by Percy
04-03-2019 10:27 AM


Like I tried to say, we've known for over two decades that this day would come and most manufacturers should have been competent enough to factor this day into their designs, you never know until it's put to the test. BTW, there are such things as GPS simulators which can simulate satellite signals for any date and any location in order to test GPS receivers, but they can be expensive so not every company owns one; it took several years before our company finally got one (so they've already tested all their products).
BTW, in New Hampshire (UTC-5) the rollover should happen at 7PM on Saturday, so you could check it then.

This message is a reply to:
 Message 2 by Percy, posted 04-03-2019 10:27 AM Percy has replied

Replies to this message:
 Message 4 by Percy, posted 04-03-2019 5:17 PM dwise1 has not replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


Message 5 of 22 (850303)
04-05-2019 10:35 PM
Reply to: Message 2 by Percy
04-03-2019 10:27 AM


Been thinking about this a bit more.
Cell phone towers use precise frequency signals (eg, 10 MHz) and a precisely timed 1PPS (one pulse per second, on the second within a few nanoseconds) in order to support time-division multiplexing of your phone signals to send and receive them over a high-volume trunk line (made high-volume by that multiplexing). A cell tower gets those signals from a device, a rack drawer, like I had worked on for 20 years as a software engineer. That kind of box uses GPS as a time base with which to generate those signals. If that box loses GPS, then it loses that time base and, when the cell tower loses that box's outputs or they degrade too far, then the cell tower network will no longer be able to function.
That's the possible cell phone failure scenario I was thinking of before, but upon further reflection I remembered that it won't happen immediately -- keep in mind that I haven't thought about any of this for over a year since I retired. One of the features of a properly designed box is that it will do its best to maintain those output signals in the event of losing GPS -- I cannot speak for our competitors whose boxes are in most cell towers, but I do not doubt for an instant that they have designed their boxes properly as we have.
When a box loses GPS, it goes into "coast mode" during which it uses what it had learned to maintain the accuracy of its output signals. I forget the more exactly times, but it should be able to keep that up for several hours to even a day or two. At the same time, it should send a notification message to the computer that's monitoring it (which should exist) that something has happened and somebody had better respond to fix the problem.
So, does the owner of the cell tower know about this problem? He should. I do know that my old company has been receiving a lot of questions from its customers about this Week Number rollover.
So in terms of watching for cell phone service disruption because of this, we should not see it immediately and, assuming that cell phone network companies respond to problems when they do arise, we should never see it. But if this takes those companies by surprise, we could still see it through degraded or non-existent cell phone service.
BTW, even if the worst happens, the cell towers will still know their locations and (assuming that location data is transmitted from cell towers like it is from WiFi) then your phone would still receive proper location data. Unless something like an epically massive earthquake were to happen, but then Week Number rollover would be the least of our problems.

This message is a reply to:
 Message 2 by Percy, posted 04-03-2019 10:27 AM Percy has seen this message but not replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


Message 6 of 22 (850340)
04-06-2019 3:33 PM


Forgot Daylight Savings
I forgot that we're on daylight savings right now in the US, so the times I gave are off by one hour.
Midnight UTC going into Sunday is Saturday at:
  • 5PM Pacific Standard Time
  • 6PM Mountain Standard Time
  • 7PM Central Standard Time
  • 8PM Eastern Standard Time
Sorry about that.

Replies to this message:
 Message 7 by Tanypteryx, posted 04-07-2019 12:35 PM dwise1 has replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


(1)
Message 8 of 22 (850414)
04-07-2019 1:23 PM
Reply to: Message 7 by Tanypteryx
04-07-2019 12:35 PM


Re: Forgot Daylight Savings
Mine too, even my Garmin which is about 5 years old (a pleasant surprise). As I said, the industry has known about this since before the first rollover in 1999 and, according to an NBC NEWS report on my phone's news feed, The Air Force and CISA (DHS's Cybersecurity and Infrastructure Security Agency) had been warning the GPS industry about this for the past two years. I fully expected most GPS receivers to work because this is a long-known problem. At my old job where we design and build GPS-based devices, they tested all their products with their GPS simulator and all units passed except for some older units using an older receiver which had been fixed successfully for the first rollover (22 Aug 1999).
Basically, a receiver without a fix would suddenly think that it was 22 Aug 1999 again, so it would start outputting wrong time/date values. Because the receiver uses that time data to predict where the satellites should be, it would very likely fail to acquire satellite data and therefore be unable to get a location fix. That would in turn impair the system using that receiver.
This article in Nature (https://www.nature.com/articles/d41586-019-01048-2?utm_so...) describes some of the data gathering stations that could be affected. In addition, cell phone towers use GPS-based sub-assemblies (our product), so failures there could have affected phone coverage.
This is a real thing. The only unknown is whether some receivers didn't get fixed.

This message is a reply to:
 Message 7 by Tanypteryx, posted 04-07-2019 12:35 PM Tanypteryx has seen this message but not replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


(1)
Message 9 of 22 (850453)
04-08-2019 2:34 PM


An Odd Coincidence
I noticed an odd coincidence, but I cannot think of how it could be related.
My phone (HTC M8 One, Windows 8.1) is paired with my car (2015 Honda Civic Hybrid Touring) through BlueTooth. They reconnect every single time I start the car. For that matter, in order to use the phone (it runs in safe driving mode) I have to stop the car and turn the car off.
I was helping teach a dance class when the rollover happened on Saturday, so the moment I stepped out of the studio I put my phone in Airplane Mode to disconnect it from cell towers (I don't connect to the studio's Wifi) and tested it (Maps worked as did a GPS app I have loaded). Then I turned Airplane Mode off and my phone reacquired cell towers. I did all of this will out of range from my car, which was also turned off at the time.
When I started the car, it didn't connect to my phone. Throughout the next day, same thing; the car even asked if I wanted to connect a phone. Finally this morning on the way to the gym, I pulled over and reestablished the connection. Now it works fine as before.
The timing is a weird coincidence suggesting that it's somehow connected to the Week Number rollover. But I cannot think of any way that GPS would affect BlueTooth.

Replies to this message:
 Message 10 by Percy, posted 04-08-2019 5:16 PM dwise1 has replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


(1)
Message 11 of 22 (850486)
04-09-2019 12:59 AM
Reply to: Message 10 by Percy
04-08-2019 5:16 PM


Re: An Odd Coincidence
Tried that test. No effect: the phone connected with the car as it should.
Tonight I was well out of range of the car. When it happened the first time, I thought I was out of range, but maybe I was in range. I can try that tomorrow.

This message is a reply to:
 Message 10 by Percy, posted 04-08-2019 5:16 PM Percy has seen this message but not replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


Message 14 of 22 (850621)
04-11-2019 10:20 AM
Reply to: Message 13 by PaulK
04-11-2019 8:48 AM


Re: A Failure in New York City
According to an NBC report, that very change is being implemented.

This message is a reply to:
 Message 13 by PaulK, posted 04-11-2019 8:48 AM PaulK has not replied

Replies to this message:
 Message 16 by Percy, posted 04-11-2019 10:57 AM dwise1 has replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


Message 15 of 22 (850625)
04-11-2019 10:54 AM
Reply to: Message 12 by Percy
04-11-2019 8:32 AM


Re: A Failure in New York City
Maybe you already said, but what was the change that means the next rollover won't be for another 157 years?
I thought I had already mentioned that, which I read in an NBC report. Guess it was to another group.
The original design, immediately set in concrete, was 10 bits for the week number count, limiting it to 1024 weeks, which is about 19.5 years. NBC reported that the system is being upgraded to a 13-bit counter which increases the limit 8-fold (23) to 8192 weeks, about 157 years. I'm not sure exactly how this upgrade is being implemented, since the entire system must continue to work seamlessly throughout the upgrade.
Regarding the NYT article you quote, I expected that most failures would be in systems that use legacy receivers. When my former company tested their GPS-based products, it was the few that used an old receiver (because the product was an old, early one) that failed the test. I suspect that NYCWiN used some GPS equipment from an early project.
Told all y'all that this was a real thing.

This message is a reply to:
 Message 12 by Percy, posted 04-11-2019 8:32 AM Percy has seen this message but not replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


Message 20 of 22 (850664)
04-11-2019 3:19 PM
Reply to: Message 16 by Percy
04-11-2019 10:57 AM


Re: A Failure in New York City
We're so byte-oriented now that we forget that so many earlier computers had different length data words; there's also no standard definition for "byte". For example, in network documentation (RFCs) the term for 8-bit words is "octet".
Refer to the ICD-GPS-200c document for the format of the data pages in the subframes of the messages transmitted by the satellites -- also for the use of that data and how all that works. You've got short groupings of bits with parity bits and other bits interspersed. There's no such thing as contiguous bytes (ie, octets) in that format.
Also, there's a lot of data being transmitted in a low-power signal. For the entire message, all 5 subframes, to be transmitted from one satellite takes about 12 minutes. Having extra bits in those messages that serve no purpose would slow things down even more.
Gotta rush to the next class now.

This message is a reply to:
 Message 16 by Percy, posted 04-11-2019 10:57 AM Percy has seen this message but not replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


Message 21 of 22 (850665)
04-11-2019 3:23 PM
Reply to: Message 19 by ringo
04-11-2019 11:46 AM


That's Selective Availability (SA) at work adding random error to the signal on the commercial frequency, thus deliberately reducing its accuracy. It was done basically for the reason you guessed: we wanted to deprive the enemy of a valuable navigation tool. Of course, military GPS units used a different frequency which did not have SA. Unfortunately, in Desert Storm there weren't enough military GPS units, so soldiers' family mailed them commercial GPS units which were subject to SA. As a result, SA was turned off temporarily for the duration. After that the FAA, wanting aircraft to be able to use GPS, campaigned to have SA turned off permanently. That finally happened in 2000 since the military could turn of commercial availability for any theater.
There are still limits built into commercial receivers on altitude and velocity for the reason you give, so that one could not be used to guide ordnance. Those the Russians have their own GPS-like system called GLONASS, so they don't need access to our system.
Edited by dwise1, : Added references and more information to first paragraph, basically rewriting it.

This message is a reply to:
 Message 19 by ringo, posted 04-11-2019 11:46 AM ringo has seen this message but not replied

  
dwise1
Member
Posts: 5930
Joined: 05-02-2006
Member Rating: 5.8


Message 22 of 22 (850751)
04-13-2019 6:33 PM


New YouTube Video on the Subject
Why didn't GPS crash?
On the Isle of Wight, he plugs batteries into his old Garmin for the first time in about a decade. I assume that this is on Sunday, since he's in daylight whereas the actual rollover itself would have happened around 0100 hours his time (taking Summer Time into account, DLT for us). He was able to acquire satellites, but the date was 22 August 1999. I would have thought that the receiver wouldn't have been able to acquire satellites.

  
Newer Topic | Older Topic
Jump to:


Copyright 2001-2023 by EvC Forum, All Rights Reserved

™ Version 4.2
Innovative software from Qwixotic © 2024