Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?
Check out the new SourceForge HTML5 internet speed test! No Flash necessary and runs on all devices. ×

Discrepancy Detected In GPS Time 187

jones_supa writes that on Tuesday, 26th January, Aalto University's Metsähovi observatory located in Kirkkonummi, Finland, detected a rare anomaly in time reported by the GPS system (Google translation). The automatic monitoring system of a hydrogen maser atomic clock triggered an alarm which reported a deviation of 13.7 microseconds. While this is tiny, it is a sign of a problem somewhere, and does not exclude the possibility of larger timekeeping problems happening. The specific source of the problem is not known, but candidates are a faulty GPS satellite or an atomic clock placed in one. Particle flare-up from sun is unlikely, as the observatory has currently not detected unusually high activity from sun.
This discussion has been archived. No new comments can be posted.

Discrepancy Detected In GPS Time

Comments Filter:
  • ...disregard it if it continues to exhibit faulty timing.

    'course, there really should be a way to correct time in a GPS satellite, if only to avoid making them completely disposable (then again, maybe there is a mechanism to correct/self-diagnose timing issues on-board? One would think/hope so...)

    • by chuckymonkey ( 1059244 ) <{moc.liamg} {ta} {notrub.d.selrahc}> on Tuesday January 26, 2016 @02:05PM (#51375687) Journal
      I think they have to be updated pretty much constantly. Their orbits aren't perfect and they have to be adjusted for relativity in their orbits.
    • ...disregard it if it continues to exhibit faulty timing.

      Sure, you just have to update the configuration of all GPS devices on earth... it is more realistic to turn it off.

      • by Strider- ( 39683 ) on Tuesday January 26, 2016 @02:24PM (#51375875)

        Actually GPS receivers on earth are in a constant state of being updated. Part of the transmission from the satellite includes a continuous update of the orbital data for the GPS constellation, and other related data. Also, in North America, the WAAS system downlinks atmospheric correction data in real-time so that the GPS receiver can compensate for changes in the ionosphere.

        • by Obfuscant ( 592200 ) on Tuesday January 26, 2016 @03:17PM (#51376399)
          I'll expand on that a bit. The orbital data is called the ephemeris, and it takes (or used to take when I had to deal with such things) about ten minutes to download. This is a mandatory bit of information for high accuracy GPS since the actual location of the space vehicles (SV) have to be known with high accuracy.

          The status of each SV is also part of the datastream, and all it takes to "turn off" an errant SV is to set the flag in the data stream that says it is unusable.

          WAAS doesn't know about atmospheric corrections. What WAAS does is use a network of fixed ground stations that detect deviations in position and generate data to correct those deviations. The assumption is that the WAAS receiver isn't moving, so any deviations are from propagation errors. This is the same kind of thing that has been used by surveyors and other high accuracy GPS users for decades. At the highest level of accuracy it is called realtime kinematic GPS, and it uses both the correction data and actual carrier phase information to give centimeter level accuracy. There is also "differential", which makes use of the correction data to get multi-cm level accuracy. Both were very big issues when selective availability was on.

          • Re: (Score:2, Funny)

            by Anonymous Coward

            Ephemeris only takes a few seconds to download. I think you meant Almanac which takes ~7 minutes and describes orbit with relatively low accuracy.

          • by Strider- ( 39683 ) on Tuesday January 26, 2016 @04:20PM (#51377003)

            In regards to WAAS, I think you're talking about something else. WAAS was developed for the FAA to allow the use of GPS in all stages of flight, including precision landing.

            It's based on a network of high precision ground receivers which are used to calculate two sets of correction information. The first is intended for all receivers in the WAAS footprint (basically North America), and consists of estimates of the error in the satellite position, and clock errors. The other breaks the continent up into a grid, and provides local estimates for errors in the ephemeris, clock errors, and ionospheric delay.

            • It's based on a network of high precision ground receivers which are used to calculate two sets of correction information.

              It sounds similar enough to what Obfuscant posted to be the same system. In either case, the high precision ground station is used to increase the accuracy by correcting various errors. He just simplified it more.

              Though looking it up - Ephemeris is 30 seconds per satellite, Almanac takes a bit longer.

            • In regards to WAAS, I think you're talking about something else. WAAS was developed for the FAA to allow the use of GPS in all stages of flight, including precision landing.

              No, I was describing WAAS. A network of ground stations determining corrections. I didn't deny it was implemented by FAA.

          • by daknapp ( 156051 )

            WAAS doesn't know about atmospheric corrections.

            That is incorrect. WAAS stations create a model for the ionospheric propagation delay over the entire network and use that to provide corrections for receivers located anywhere in the area covered by the network.

            WAAS also provides corrections for ephemeris and clock errors.

      • by Seraphim1982 ( 813899 ) on Tuesday January 26, 2016 @04:24PM (#51377051)

        >Sure, you just have to update the configuration of all GPS devices on earth...

        If only they were capable of receiving signals from some sort of satellite.

    • by Anonymous Coward

      Was systemd installed on any of the satellites recently? Maybe it was installed unintentionally, while upgrading the satellite from Debian 7 to Debian 8?

    • Well there are corrections that are broadcast from the satellites. If you have a GPS that allows a user to upload a new almanac manually you can get them here [uscg.gov] instead of waiting for the over the air update. Also there is the WAAS system that broadcasts corrections that take some of the error out of the system. Also this is a huge error, as in off by kilometers off. That said I don't think either of those would be able to correct for something like this. A time off by 13 milliseconds means position could be
      • by PRMan ( 959735 ) on Tuesday January 26, 2016 @03:01PM (#51376241)
        As someone who works in GPS software, you see these kinds of off-by-miles readings from time to time. If it was one time and not constant, we already have software in place to ignore the anomalous reading.
        • by Greyfox ( 87712 ) on Tuesday January 26, 2016 @05:28PM (#51377563) Homepage Journal
          Oh yeah, don't do any root cause analysis or anything, just keep ignoring all those gravity waves jittering the clocks on our orbiting atomic clock satellite network :-P
          • The "root cause" is that remote sensing is inherently noisy and you learn to live with it. All the cool kids now how to use a Kalman filter.

            • Nothing new either. Kalman filters have been used for avionics with mixed and low quality sensors for a long time: https://en.wikipedia.org/wiki/... [wikipedia.org]
              • Exactly. I know that Greyfox was making a joke, but anyone who has worked with sensors knows that intermittent anomalous readings are normal. Any sensor which doesn't produce anomalous readings almost certainly does the filtering internally.

        • I presume you're talking about RAIM: https://en.wikipedia.org/wiki/... [wikipedia.org]

          Basically if your GPS receiver can hear more than the minimum (4 satellites usually) at once, the simultaneous equation solutions are overdetermined. This makes it possible for software to detect and ignore outliers in the solution set.

          I heard of a GPS failure back in the mid 90's that caused the entire Los Angeles area CDMA cellular network to stop working - because US CDMA (as opposed to UMTS CDMA) is exceedingly sensitive to timing e

      • The time discrepancy is 13.7 microseconds [not milliseconds]. I don't know how that translates to position accuracy, but, to me, even that seems a bit large for something derived from an atomic clock.

        • I meant to type microseconds and the calculation was done using microseconds.
          • Several AC's replied to me about speed of light being [roughly] one foot per nanosecond [which I had forgotten]. So, 13.7 us is 13,700 ns, or 13,700 feet, or 2.5 miles [just as you said]. Wow! I know that GPS receivers [try to] use several satellites. Can they compensate for this without an almanac update [automatic or manual]? Or, if they use the faulty one, what happens? Would they try to average it in or reject it as too far off the average of the others?

            • If the solution can be oversolved, more than enough satellites to provide position, they will drop the ones that offer the least accurate positions. So in most cases it would drop that one since it would appear that the signal had traveled a couple of miles further. If however it can only really lock on to 3 (2D lock) or 4 (3D lock) satellites you would likely see huge issues with accuracy. With only the minimum ones in view there really wouldn't be any way for the GPS to know one was off without an update
    • by Snotnose ( 212196 ) on Tuesday January 26, 2016 @03:21PM (#51376433)

      there really should be a way to correct time in a GPS satellite

      1) Press and hold the Set Time button until the indicator lamp lights (5 seconds)
      2) Press + or - until the correct time is reached
      3) Release Set Time button.

    • ...disregard it if it continues to exhibit faulty timing.

      I'm not sure of the specs of block III, but in the case of block II, each satellite has three atomic clocks each (i.e. two hot spares) and the constellation as a whole has a few hot spare satellites (my memory is telling me four, but this may be wrong). However, this may not be a single-satellite failure.

    • I believe that is the responsibility of the Ground Segment http://www.navipedia.net/index... [navipedia.net]

  • It's a hacker or NK trying to start a war.

  • I think the clock went through a pocket of dark matter and the time dilation caused the time discrepancy.

    • After reviewing the current state of Earth, in general, and of US politics and their recent "debates", in specific, Judiciary Pag [wikipedia.org] has realized that the Earth population will not be satisfied alongside the existence of the rest of the Universe, and has sentenced Earth and its sun sealed in a Slo-Time envelope within which time will pass almost infinitely slowly until the end of the Universe, thus serving the dual purpose of protecting the Universe from Earth, and allowing us Earthlings to enjoy a solitary e

    • by balbeir ( 557475 )
      It's a glitch in the simulation.
    • The machine that is running the simulation we experience as our universe had a data fault, and we had to back up to the last savepoint.

  • They where checking if they could skew the accuracy of GPS and if anybody would notice when they did. Oops, somebody noticed.

    Well, that's my theory.. Don't ask me who "they" are because I left my tinfoil hat at home today.

    • Moving at the speed of light, 13 microseconds is only 4 kilometers, at the speed most normal things move, you're looking at less than 10 millimeters.

      • by Strider- ( 39683 )

        With a good GPS fix, the receiver should throw out the 13 microsecond delayed data as being anomalous. If you only had a 3 satellite fix though, it could cause a significant positioning error.

      • by bondsbw ( 888959 )

        Sure, if "most normal things move" around 0.77 km/s.

        • What? [wolframalpha.com]

          • by bondsbw ( 888959 )

            I don't understand your confusion:

            1) 500 knots is 1/3 of 0.77 km/s. 3.3 mm is 1/3 of 10 mm.
            2) 10 mm is the number you provided, so it's the number I calculated. 3.3 mm is significantly less than 10 mm, so if you meant that, why not say "around 3 mm" or "less than 4 mm" or "3.3 mm"?
            3) I don't know too many "normal things" that move even at 500 knots.

      • Well considering that the time stamp from each satellite is what is used in the calculations, see pseudoranges [wikipedia.org], and those travel at the speed of light your position could be off by some large, easily 100s of meters, amount.
        • by TWX ( 665546 )
          Imagine the entertainment as thousands of self-driving cars suddenly try to correct their positions without notice...
          • by PRMan ( 959735 )
            This happens occasionally already. We throw that kind of data away, because it's clearly not valid for some to travel a mile in a millisecond.
      • by Shatrat ( 855151 )

        It's also about 10% of a Sonet frame. Fortunately most telecom clocks don't blindly repeat GPS time, but instead use it to gradually steer their internal rubidium and quartz clocks so there wouldn't be an abrupt change in the output.

  • Y'heard it here first! Don't tell me I didn't tole ya!

    Going back in time by a mere 13.7uS doesn't seem very exciting though.
  • by Anonymous Coward

    It's about to get exciting, folks!

  • Damn, my $80 dash cam uses GPS time. I'm screwed.
  • Mr. President, they are using our own satellites against us and the clock is ticking.

  • by phkamp ( 524380 ) on Tuesday January 26, 2016 @02:21PM (#51375831) Homepage
    This is the second time a bug in the firmware of Motorola Oncore GPS receivers have manifested itself. There is a bug relating to a 32 bit wide bitmap, and DoD just took the GPS satellite numbered 32 out of the constellation and that seems to be the cause. I have data for two such receivers showing the anomaly and for one different receiver seeing no trouble at all.
  • This is why there's redundancy built into the system, more satellites than are strictly needed for operation. If one's clock goes out-of-spec, you notice that it's not agreeing with the rest of the constellation and drop it from your sources. If it's a transient glitch it'll come back in-spec and come back into use, if it's a permanent problem they decommission it and schedule a replacement. Redundancy makes the difference between a major crisis and a minor annoyance.

  • Tiny? (Score:3, Insightful)

    by alvieboy ( 61292 ) on Tuesday January 26, 2016 @02:24PM (#51375877) Homepage

    "(...) reported a deviation of 13.7 microseconds. While this is tiny (...)"

    Tiny ? It's huge.

    If such an error occurs every hour, the total accumulated error would be more than 7 seconds. It's tiny if you look at it individually (well, not so tiny - your 2GHz CPU clock has a period of 500ps (picosseconds) - that's 0.0000005 microseconds).

    The atomic clock period (based on Cs-133) is 108.78278 picosseconds. So this is very very large.


    • 1us = 1000ns = 100000 ps hence 500ps=0.5ns=0.0005us=0.0000000005s=5e-10s

      but yes, not a tiny error given expected GPS accuracy.

      pulse-per-second signal derived from GPS should be accurate to tens of ns...

    • Interesting post, but 500ps is 0.0005 microsec, not 0.0000005 microsec. Look at it another way: 500ps is 1/2 ns, i.e. 1/2 of 0.001 usec.
  • Time-Nuts... (Score:5, Interesting)

    by sillivalley ( 411349 ) <sillivalleyNO@SPAMcomcast.net> on Tuesday January 26, 2016 @02:30PM (#51375933)
    Lots of folks on the time-nuts mailing list have GPS-based systems to maintain not only precision time, but also precision frequency standards, and many of them saw and recorded this one.
  • OK finally (Score:5, Funny)

    by argStyopa ( 232550 ) on Tuesday January 26, 2016 @02:49PM (#51376105) Journal

    Reading this, I really feel like I'm living in the future:
    "The automatic monitoring system of a hydrogen maser atomic clock triggered an alarm which reported a deviation of 13.7 microseconds."

  • by 140Mandak262Jamuna ( 970587 ) on Tuesday January 26, 2016 @03:17PM (#51376391) Journal
    When the atomic clocks in GPS satellites have a discrepancy, you don't report a discrepancy. You report you have done some experiments that suggest faster than light travel across some 30 km apart in the Swiss Alps. By the time they track it down and attributed to some discrepancy in some atomic clock, you got your headlines, the 15 minutes of fame.
  • God is messing with space time getting ready for the rapture!
  • Chuck Norris did some push ups so aggressivly that in contrast to his normal work out he liftet his body upwards and did not push the earth down,
    thus generated a gravitational wave hitting the gps satellite.

    The other explanation is:
    It is a black hole and we all are doomed.

  • new physics? (Score:5, Interesting)

    by peter303 ( 12292 ) on Tuesday January 26, 2016 @03:46PM (#51376673)
    Clock speeds are sensitive to the structure of the gravitational field. Maybe other aspects of the Universe.

    GPS times have all sort of noise. Some geophysicists use this "noise" to figure things like the atmospheric temperature and density. The GPS signal wavefront bend slightly then. You can tomographically invert for spatial location of the travel time anomalies to locate temperature and density changes. There are papers on this every year at the American Geophysical Union meeting.

    Microsecond size anonalies are huge and may have more mudane causes like software.
  • It's not a problem unless all four corner days are off and thus the four corners of the cube.

    • by narcc ( 412956 )

      You educated evil human have not the education or rationale to comprehend Nature's Simultaneous 4-Day Cube. who can't understand 4-day simultaneous cubic time. You are probably brainwashed, indoctrinated, educated stupid and cannot comprehend Nature's Harmonic Simultaneous 4-Day

      No single corner human can occupy or experience more than a single corner at the same time during a 4-corner rotation within the 4/16 creation principle. Earth sphere rotates within an invisible Time Cube.

      Educated people are the evi

      • by clovis ( 4684 )

        I don't think that I can ever comprehend the 4 simultaneous years within a single rotation of Earth about the Sun.
        What makes it hard for me is that I now know that I am educated stupid.
        I understand that this is the truth, but I cannot occupy more than a single corner at the same time.
        This is so hard for me - the not occupying or experiencing.

  • The activation of the Large Hadron Collider with full power this past year has ripped a hole in the space time continuum and has been jumping the planet forwards in time by nanoseconds, now milliseconds. If the rift is not contained it will eventually grow exponentially from seconds, minutes, hours, days, months, years, centuries, millenniums to eventually eons, and may forward us in time past the moment of when Sol had collapsed killing us all!!!

  • Aliens. Wait, Bigfoot. WAIT, ALIEN BIGFOOTS!!!!11!1!!!1
  • The troublesome bird was SVN-23, one of the oldest GPS birds, launched in 1990!
    It was the last of the Block IIA birds, and had an expected 8 year lifetime, which it beat by quite a few years!
    It featured a combination of cesium and rubidium clocks -- two of each. Now decommissioned -- http://www.navcen.uscg.gov/?Do... [uscg.gov]
    Read more of this bird's interesting history -- http://www.schriever.af.mil/ne... [af.mil]
  • by Whip ( 4737 ) on Tuesday January 26, 2016 @07:47PM (#51378361)

    It looks like the actual problem was a bad data upload; Specifically, some satellites were transmitting incorrect parameters for UTC offset correction. https://www.febo.com/pipermail... [febo.com] is the posting from a gentlemen at Meinberg that has the details. http://www.usno.navy.mil/USNO/... [navy.mil] has more information about the time offset parameters (A0 and A1) and how they interact with GPS and UTC time.

    According to another message (https://www.febo.com/pipermail/time-nuts/2016-January/095686.html), PRNs 2, 6, 7, 9, and 23 got hit. It is interesting to note that the satellite that was taken out of service this morning (PRN 32) is not in this list. It looks like the decommissioning of PRN32 was quite possibly scheduled (see http://gpsworld.com/last-block... [gpsworld.com]), and even if not, a failure of that specific satellite could not have caused multiple satellites to start broadcasting incorrect offset data.

    I'm really looking forward to the postmortem on this.

  • Even if you're using a slowpoke processor with 1 MHz clock, that's over a dozen clock cycles.

    And if you're measuring the distance that radio signals travelled, that's a whopping 4 km.

  • Is this one of those "that's funny" events that lead to world shattering discoveries?

    Or just a bug...?

  • Just patted a passing bird(in orbit)

"Well I don't see why I have to make one man miserable when I can make so many men happy." -- Ellyn Mustard, about marriage