Location drift

[planetnine]

All GPS in in the mind, it's just too damn amazing and clever to really exist!

In fact you're all a figment of my imagination, caused by eating too much cheese before bedtime...

Looks like the sort of thing I encountered on my Portsmouth drive Davidor. Must be some kind of heavy external influence, and I wouldn't discount the MOD theory too quickly.

Thanks for posting this, whereabouts was this if you don't mind me asking?
_________________
planet nine
Lincoln, UK.

[davidor]

Hi Planetnine. The walk was on the 'wheelock line', a disused railway line near Sandbach in Cheshire. Starts at 53.15N 2.39E. Continues on as the 'salt line' and makes a nice 12mile round walk by returning along the Trent & Mersey Canal.

[FrequentFlyer]

I wasn't having a go at you..just the general topic and so called theory's....(I must go and walk around some MOD sites next week.)

[AllyCat]

Hi,

When I first saw this topic I assumed that the OP had suffered one of the occasional "bad fixes" that GPSrs sometimes exhibit, or maybe even corruption of the ephemeris data. My "best" example of this was when my GPS informed me that I was travelling at 100mph in the middle of the English Channel, when I was actually sitting in a barn in Surrey.

However, now that davidor has posted a track of the event, I think his description is both fair and accurate (except the GR should be W and not E !). So here is my explanation for the scenario, although if anybody can produce actual satellite data for the location and time, then I'm happy to be proved wrong (or right). A "faulty" satellite or local radio interference is a possibility, but not IMHO the most likely cause.

Firstly, bear in mind that the GPS system does not give a "guaranteed" accuracy but specifies an error circle radius (and not as small as some would claim) for 95% of the time. Now, the situation described in this thread is probably from the remaining 5% of the time, and might even be a "0.1% event", or worse.

The GPS satellites do not fly in a fully "synchronised" constellation, because any orbital modification to "station keep" would have to take the satellite out of commission until its (incredibly precise) orbital parameters could be recalculated. So there is not a constant number of satellites "visible" in the sky, sometimes there can be as many as 12, other times only 6 - 8 (it tends to vary over a 24 hour cycle).

Their orbital planes cross, so sometimes two satellites can be at a very similar location, making one almost redundant for the geometric calculation. Also any satellites very high in the sky (above the GPS) can contribute only very little data for the horizontal position calculation. Furthermore, the satellite orbits are inclined (to the equator) so they don't go to higher latitudes than the UK . Therefore there is a "hole" in the coverage to the north (if you have the VisualGPS software and a GPSr connected on your PC try running it for 12+ hours to see how the orbital trails build up).

Looking at the photo provided by davidor, it appears that he may have been walking between hedges or trees, and the corresponding OS map shows a few small cuttings along the (railway) track. Therefore, the reception of signals from satellites close to the horizon might have been poor or non-existent.

So we could have had a situation where there were perhaps only 6 or less satellites visible, with a few of those of limited usefulness due to their elevation or co-location. Very likely none in a northern quadrant and maybe none in the southern quadrant either (it might be significant that the obvious error is in the northing - we don't know about the eastings which might have been very good).

I don't know about the Satmap, but most of my GPSrs don't actually "give up" on trying to calculate a position until the "DOP" is more than about ten times worse than a "good" fix. So I don't think that the recorded track is necessarily "faulty", but just rather "unfortunate".

Cheers, Alan.

PS: It's not obvious which direction you were walking in, but my guess is that the problem was when heading west?
_________________
Garmin GPS72H/76/60/45, Etrex H, Mapsource v6.5.
Acer N50,HP114,Loox N560,Dell x50,CF/SD cards to 4/32GB.
RoyalTek,Holux236,Navman B10 & Copilot(Globalsat) BT GPS,TomTom5/6.
Memory Map (v5.4.2 & v5.1.3 OS & Euro), GPS gate,OSGPSconverter.

[davidor]

Thanks for your reply Alan, and yes, of course the gr should have been west!

I was walking west to east in the morning when the drift occured, the return being mid afternoon.

The north-south error is obvious from the difference between the two tracks. There was an east-west error, but less than the N-S error.

It is interesting that the deviation started quite suddenly, as can be seen in the top left of the track picture, where there were some trees alongside the track but it was fairly open. Also, if it was the result of cover interfering with the reception, I would expect the displacement to alter as the trees etc varied but it seems a more consistent drift. Normally I find the Satmap very good at getting a signal through cover, and even get a fix indoors.

I find this interesting as, whilst I have had the odd misplaced location before, this is the first time I've noticed an extended displacement. Usually the error is after starting up and after walking a few yards it comes to the right location.

[AllyCat]

Hi David,

Sorry, reading through the whole thread again, I see that I overlooked a few of the details that you gave the first time. And I wonder if FrequentFlyer is more familiar with the "lock to road" used by all car satnavs?

There must be better software around, but I've just fired up my Garmin PCX5 (DOS !) and GPS45 (both circa 1995 and now working on XP), so I can now look at the satellite geometry and DOP for any time in the recent past. What was the exact time that your tracklog drifted?

On 17th January I can't see any major problems with the DOP but at around 10.30 in the morning there seem to be 6 satellites nearly in an E-W line with no satellites predominantly to the north and south. However, I understand that DOP is a root-mean-square of (predicted) x and y errors, so a good easting shouldn't be able to completely "hide" a bad northing.

My guess is that one or two of the "important" (N-S) satellites were temporarily "lost" at the time. I don't know which GPS engine the Satmap uses, but I've often seen SiRF chips briefly "drop" a satellite signal for no obvious reason. But the GPS signals are so incredibly weak that probability theory plays a large part in the system behaviour. So, as I suggested before, it might just have been an unfortunate "1 in a thousand (or million) event".

Cheers, Alan.
_________________
Garmin GPS72H/76/60/45, Etrex H, Mapsource v6.5.
Acer N50,HP114,Loox N560,Dell x50,CF/SD cards to 4/32GB.
RoyalTek,Holux236,Navman B10 & Copilot(Globalsat) BT GPS,TomTom5/6.
Memory Map (v5.4.2 & v5.1.3 OS & Euro), GPS gate,OSGPSconverter.

[davidor]

Hi Alan

Time was about 11:00 - 11:30 on the Sunday morning.

[AllyCat]

Hi David,

Well, I've checked the satellite positions over that time period and can't see any sign of an "unfortunate" geometry, or even an undue dependence on any one particular satellite in the constellation.

Clutching at straws now, but did anyone with you have any electronic equipment that might have produced interference? That might even include another GPS, because my Polstar GPS actually destroys reception on my Garmin, if placed within a few feet of it. But, although interference might produce a complete loss of tracking, your drift in position is much more difficult to explain.

Perhaps I shouldn't be contributing in this thread as I don't actually have a Satmap, but I'm wondering if it has an unusually large amount of filtering or "track smoothing" (since it is aimed primarily at walking and cycling). I've always been surprised that most handheld GPSrs are so "unintelligent" that they will log a walking track for hours, and then (on rare occasions) happily log a "rogue" trackpoint perhaps a mile away, corresponding to a speed of hundreds or even thousands of mph.

Currently on the Memory Map forum there is a user complaining that their Adventurer produces a very "ragged" track log compared with the Satmap. To me, the shown Adventurer tracklog looks typical of a high sensitivity chipset struggling under very difficult conditions, and with a short logging interval.

So my hypothesis is that your Satmap GPS engine might have calculated one "bad" location (maybe due to the trees you were passing), for example a mile to the south. But the "intertia" of some form of software filtering caused this "kick" to appear as only the gradual drift in recorded direction. In practical terms, the filter algorithm might take the average positions and velocities over (say) the previous ten minutes (with some appropriate weighting) so that the "bad" position has relatively little immediate effect (compared with a mile error in position), but its effect continues (decaying away) until it "falls out" of the calculation some minutes later.

This assumed filtering action might actually be part of the Satmap firmware, but equally I believe some of the data used by advanced GPS engines is subject to slow filtering so a "bad" bit of data that slips through the validity checking could produce a similar effect.

Sorry no definitive answer, so it may have to remain an unexplained event, unless anyone else has any information or ideas?

Cheers, Alan.
_________________
Garmin GPS72H/76/60/45, Etrex H, Mapsource v6.5.
Acer N50,HP114,Loox N560,Dell x50,CF/SD cards to 4/32GB.
RoyalTek,Holux236,Navman B10 & Copilot(Globalsat) BT GPS,TomTom5/6.
Memory Map (v5.4.2 & v5.1.3 OS & Euro), GPS gate,OSGPSconverter.

[4ndynorfolk]

I think that 1 second refresh is too frequent for walking over terrain.

The distance actually covered in 1 second rarely exceeds the estimated tracking error.

The 4 second interval in advanced power mode is more suited to a ground speed of 3 or 4 km/hour.

It is a shame that selecting advanced power mode disables other useful features.

[davidor]

This is interesting (and frightening) ...

http://news.bbc.co.uk/1/hi/sci/tech/8533157.stm

We see you’re using an ad-blocker. We’re fine with that and won’t stop you visiting the site.
Have you considered making a donation towards website running costs?.
Or you could disable your ad-blocker for this site. We think you’ll find our adverts are not overbearing!