Re: GPS will fail EVEN SOONER, not
The current GPS data format will fail in approximately 2057, 2079, or 2095 for decelerations of 42, 31, or 25.6 s/cy2, respectively. In terms of deployed systems, that's Real Soon Now. Not to worry. It won't fail. The solution is simply to let delta t sub LS in page 18 subframe 4 roll over every 128 leap seconds. This is the same approach as was used to solve the GPS week number rollover problem. Some people would argue WNRO was neither a solution nor a problem. I think most GPS receiver manufacturers got the week number roll over right in the first place; those that got caught just issued an upgrade in most cases. True, the GPS ICD was updated, but only to explicitly remind users (i.e., GPS receiver firmware writers) to take WN roll over into account. The signal format did not change. So if that was good news, here's the better news: Recall with GPS WNRO there were several years of public notice and the event occurs every 19.6 years. At a leap second rate of one per 500 days we have some 150 years of notice and a LSRO (leap second roll over) event every 180 years. Even with a leap second rate of two per year, we still have 57 years of advanced notice and a LSRO every 64 years. And (just for fun) at today's weird rate of one leap second every 5 years we have 500 years of notice and a LSRO every 640 years. One suggestion, though. Sometime in the next few decades ICD-GPS-200 should be updated with a comment such as delta t sub LS is modulo 128 and may roll over and users must account for this. The signal format does not need to change. /tvb http://www.LeapSecond.com
Re: GPS will fail EVEN SOONER
The W1K rollover for GPS was in 1999, and all that year was spent testing various systems to see how they would fail. It would not be at all surprising if the impending doom of the leap second counter was noticed during a review of other deficiencies in the GPS system. Please see: Some historical notes on the GPS Week Number Rollover http://www.leapsecond.com/notes/gpswnro.htm I think the LEAPSECS group will find the part about the GPS leap second patent quite interesting! /tvb http://www.LeapSecond.com
Re: GPS will fail EVEN SOONER
On Thu 2004-01-01T15:48:01 -0800, Tom Van Baak hath writ: Some historical notes on the GPS Week Number Rollover http://www.leapsecond.com/notes/gpswnro.htm I think the LEAPSECS group will find the part about the GPS leap second patent quite interesting! Wow. I can barely imagine being enough of a conspiracy theorist to suppose that the invalidation of that patent might be a motivating factor for discontinuing leap seconds. -- Steve Allen UCO/Lick Observatory Santa Cruz, CA 95064 [EMAIL PROTECTED] Voice: +1 831 459 3046 http://www.ucolick.org/~sla PGP: 1024/E46978C5 F6 78 D1 10 62 94 8F 2E49 89 0E FE 26 B4 14 93
Re: GPS will fail EVEN SOONER
On Thu 2004-01-01T15:48:01 -0800, Tom Van Baak hath writ: Some historical notes on the GPS Week Number Rollover http://www.leapsecond.com/notes/gpswnro.htm I think the LEAPSECS group will find the part about the GPS leap second patent quite interesting! So here is an obvious exercise for the LEAPSECS reader: There must be some sort of Diophantine equation which combines the values of the 10-bit GPS week counter and the 8-bit GPS Delta-t_LS counter and produces most likely values for the number of 1024 week offsets from 1980 and the number of 256 second offsets from 0. Given the stochastic nature of the decadal oscillations in LOD, how far into the future is it likely that such a Diophantine equation will be valid? Given the link above, this algorithm is undoubtedely also patentable. -- Steve Allen UCO/Lick Observatory Santa Cruz, CA 95064 [EMAIL PROTECTED] Voice: +1 831 459 3046 http://www.ucolick.org/~sla PGP: 1024/E46978C5 F6 78 D1 10 62 94 8F 2E49 89 0E FE 26 B4 14 93
Re: GPS will fail EVEN SOONER
On Wed 2003-12-24T13:33:37 -0800, Steve Allen hath writ: The current downlink data format for the GPS satellites stores the difference between GPS system time and UTC using 8-bits. oops. I've just re-read the GPS Interface Control Document. Those 8-bits are a signed quantity. It can count to 127. The current GPS data format will fail in approximately 2057, 2079, or 2095 for decelerations of 42, 31, or 25.6 s/cy2, respectively. In terms of deployed systems, that's Real Soon Now. Is this the smoking gun for starting the review of leap seconds? The W1K rollover for GPS was in 1999, and all that year was spent testing various systems to see how they would fail. It would not be at all surprising if the impending doom of the leap second counter was noticed during a review of other deficiencies in the GPS system. The review process for the new GPS signal formats is most easily visible at http://gps.losangeles.af.mil/engineering/icwg/ The most recent document still indicates that the delta between GPS and UTC is stored in a signed 8-bit quantity. Is it reasonable that the demise of civil Mean Solar Time should be caused by lack of foresight by the GPS system designers? -- Steve Allen UCO/Lick Observatory Santa Cruz, CA 95064 [EMAIL PROTECTED] Voice: +1 831 459 3046 http://www.ucolick.org/~sla PGP: 1024/E46978C5 F6 78 D1 10 62 94 8F 2E49 89 0E FE 26 B4 14 93
Re: GPS will fail EVEN SOONER
In message [EMAIL PROTECTED], Steve Allen writes: On Wed 2003-12-24T13:33:37 -0800, Steve Allen hath writ: The current downlink data format for the GPS satellites stores the difference between GPS system time and UTC using 8-bits. oops. I've just re-read the GPS Interface Control Document. Those 8-bits are a signed quantity. It can count to 127. The current GPS data format will fail in approximately 2057, 2079, or 2095 for decelerations of 42, 31, or 25.6 s/cy2, respectively. In terms of deployed systems, that's Real Soon Now. Steve, I fundamentally agree with you I think, but you are starting to sound a bit shrill. 53 years is an exceedingly improbable lifetime for any deployed system. Few of todays semiconductors have reliability figures which gives us any reason to think they will work 20 years from now, and in particular, there is plenty of time to augment the GPS signal format several times over before before 2057. Can we turn the alarmist tone down a bit ? Poul-Henning -- Poul-Henning Kamp | UNIX since Zilog Zeus 3.20 [EMAIL PROTECTED] | TCP/IP since RFC 956 FreeBSD committer | BSD since 4.3-tahoe Never attribute to malice what can adequately be explained by incompetence.