Location drift

[davidor]

Whilst out for a walk yesterday, part of the walk was along an old railway track. The location on the Satmap, and the breadcrumb trail, drifted about 200 metres off the actual course over a distance of about half a mile, gradually returning to the correct location. On the return walk three hours later the location was spot-on. The accuarcy was reported as 30 feet in both directions and normally the location on the map is exact. anyone any idea why it should have drifted off as much as it did? At least 8 satelites were linked at the time, so it wasn't down to poor reception.

[jonesbach]

I've had something similar, but on very rare occasions and unfortunately haven't checked it at the time it was on walkabout. I can only assume in your case that perhaps the satellite positions were lower in the sky in the morning and this resulted in a poor fix. It's not just down to the number of satellites, their position/SN will also be relevant.

[Bosun]

If the satellites are spread out over the sky the reception will be better than if they are slosely packed together in a tight sort of "V" shape.

Ian

[planetnine]

A partial explanation of the "accuracy" readings on GPSs:

The reading ususally represents the radius of deviation of the closest 50% of the fixes obtained from the satellites in view (eg 50% fall within a 16ft radius circle).

The signals from these satellites are delayed by the Ionosphere (electrically-charged atmosphere layer 50-250 miles high), which delays GPS signals by minute amounts according to its thickness and activity. This can be thought of as bending the view that your receiver has of the satellite like light bing diffracted by water.

As fixes depend on very accurate signal delay timing, this causes an error. Fixes obtained from closely-spaced satellites will tend to pass through the same bit of ionosphere, acquiring the same (?or very similar) propagation error. Although fixes from closely-spaced satellites may return a "tight" (read small) fix radius, they are likely to be offset by ionospheric propagation error -effectively giving you an offset fix and falsely optimistic "accuracy" figure.

Also, if you've ever done longhand trig on a paper map, you'll know that any inaccuracy in bearing measurement will be amplified by using two trig points that give a narrow angle. It's the same with GPS signals from closely-spaced satellites -any timing errors, etc will be amplified.

Back to propagation errors from the ionosphere: Widely spaced satellites are more likely to pass their sigals through a different part of the layer and so are likely to give a wider spread of fixes, so reducing your "accuracy" figure and letting you know to take the average fix with a pinch of salt. This is also statistically more likely to give a more accurate deduction due to propogation delays than the unidirectionally tainted fix gained from close satellites being pulled off-target by equally delayed signals

Or put more simply, a random spread of errors results in a better position estimate than closely-related similar errors.

If you're walking down an old railway line and an embankment blocks out signal from satellites in a particular direction, or even if one or two fall below the horizon, without their influence the resultant of the errors may shift in a particular direction until the balance is redressed when satellites from that direction become visible again.

Occasionally a satellite will have its own timing or positional error, and losing and regaining that particular signal will cause shifts in a gps trail. Other deviations can be caused by local atmosheric effects, signal reflection electrical interference, gaining and losing EGNOS coverage, etc.

A bit of interesting extra info here...

One thing the EGNOS overlay system does is transmit to your GPS a "map" of propagation delays over the European area catered for. Once your unit has worked out where the source satellite is, it works out what bit of the ionosphere the signal will have passed through and applies a correction taken from the map to the fixes obtained from that satellite.

Overlays like EGNOS also monitor for individual satellite signal errors and offer correction info -or even tells your unit if a satellite goes bad.


I bet you didn't want to know all that...

I've often had little deviations of 10m or thereabouts in a breadcrumb trail that seemed to correct itself further on; 200m was a lot though, bet one sat had a hiccup. I once encountered deviations of quarter to half a mile on a Garmin Nuvi while travelling south to Portsmouth in the early hours -that wakes you up. I was told it was probably a satellite error.
_________________
planet nine
Lincoln, UK.

[davidor]

Thanks for your comprehensive explanations Planetnine.

However: If it was sattelite error due to poor reception, why was the reported accuracy about the same both times? I thought the reported accuracy was a reflection of how good the resolution of the sattelite signals was.

And then, if it was one of the sats having a bad day, then everyone would experience a lack of accuracy - anyone else find themselves 200 metres out on Sunday morning?

Having a fair idea of how gps works, and having a good clear reception from several sattelites (I wasn't in a cutting), it still seems odd to me that it should have been so far out.

[planetnine]

Hello Davidor,

was this Sunday 17th? I'll look at the EGNOS monitor logs and see if they recorded anything.

Remember that the figure you know as "accuracy" just reports how close the trig fixes are grouped together -16ft would mean that half the fixes your unit calculates fall within a circle of radius 16 feet. The resultant fix is an everage of these positions.

What I tried to explain above is that this "accuracy" figure and the actual accuracy of the resultant fix might not always be related: ie you can get a tight pattern and an inaccurate fix and conversely a widely spaced pattern and a more accurate fix.

Saying this, however, the drift of 200m is a bit extreme and seems almost inexplicable. I'll do some more research and ask some questions. Was the drift smooth over a short distance or did it jump out and then back again later? Have you got the raw GPX file from this? what do the speed readings say at the beginning and end times of the fix?


PS I'm actually having problems finding any EGNOS server messages between Friday morning and Monday morning for either of the two main ESA SBAS birds, wonder if there's been some down-time, inclement space weather perhaps?
_________________
planet nine
Lincoln, UK.

[FrequentFlyer]

So why doesn't this seem to happen on my TomTom and CoPilot ?

[lucevans]

[planetnine]

I have a Garmin Nuvi for road navigation and I've had extreme drift issues where the unit tried to lock onto roads about half a mile to the right of where I knew I was. This was part-way through a several hour journey and unit resets didn't correct it. From memory this plagued me for half an hour to an hour or so, and then corrected itself.

I presumed a constellation system issue at the time, but I never found any info about it.

The "road lock" will work for small distances and gradual changes I think, but not this kind of deviation.
_________________
planet nine
Lincoln, UK.

[planetnine]

Another idea was loss and reacquisition of one of the EGNOS signals if Davidor's walk was around the middle of the day.

The highest elevation operational EGNOS satellite is PRN-124 (Artemis); this had a near-unsuccessful launch when one of its Ariane stages malfunctioned and it took ESA a while to "walk" it to its geostationary orbit. The orbit still has a discrepancy and wanders north and south of its place over about a 24-hour period, presently being at its highest elevation just before 1am and its lowest about 1pm. From where I live that's an elevation wander from 34 degrees above the horizon down to less than 18 degrees in the middle of the daytime.

This means it's harder to get an EGNOS signal from PRN-124 in the middle of the day and as the other operational one, PRN-120, is less than 30 degrees elevation, in a different direction (SW) and might not be "visible", that could cause some change in fix.

But again, I don't think it's realistic to expect a 200m deviation from losing and regaining EGNOS (and I don't know if Davidor's Active-10 is even v1.4/EGNOS-active). Also, isn't 30ft "accuracy" is a bit on the low side for outdoors reception unless there is substantial terrain shadowing? I get 16-19ft from mine indoors (or 6ft with EGNOS lock near a south-facing window)

Strange that EGNOS has a big gap in its server data that same weekend though...


PS Where abouts in the UK were you walking Davidor?
_________________
planet nine
Lincoln, UK.

[davidor]

Yes it was the morning of the 17th. I'll see if I can get a screenshot of the trail.

[davidor]

Try and see if I can get this pic inserted: lower track is the am drifted one.



http://www.flickr.com/photos/46849458@N06/4297037463/

OK, that works! Tried using the img tags to display picture here but that didn't seem to work.

No EGNOS signal on this walk. Without EGNOS my normal accuracy is between 16 & 30 ft.

The drifted track was walking west to east. As you can see, the drift started quite suddenly, gradually got worse, then more gradually came back on track. The upper track is the return which was pretty well spot on (the bit on the right where it is very different is because I approached this bit of the walk from a different direction).

[lucevans]

Do you live near any MOD testing ranges? I seem to remember that they were testing GPS jamming kit that was intended to degrade GPS accuracy locally in mid-Wales last September, so maybe they do it in other areas sometimes as well? For the record, I am not a conspiracy theorist!

[FrequentFlyer]

I find all this 'drifting' nonsense amazing...I've been walking with my device now for over two years, and have yet to notice any of this stuff....so where have I gone wrong ? Where should I visit to view 'drifting' at first hand ??

[davidor]

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!