Author Topic: Your GPS is lying to you...  (Read 6496 times)

Angle of Repose

  • Newbie
  • *
  • Posts: 41
    • View Profile
Your GPS is lying to you...
« on: April 16, 2014, 04:50:44 PM »
http://blog.oplopanax.ca/2012/11/your-gps-is-lying/

Granted it was an informal survey, but I found it interesting.
"You can't get there from here"

captain paranoia

  • Sr. Member
  • ****
  • Posts: 384
    • View Profile
Re: Your GPS is lying to you...
« Reply #1 on: April 16, 2014, 06:35:06 PM »
It's an interesting discussion, but the biggest uncertainty may be in knowing what the GPS receiver means when it gives an error figure or shows an 'error circle'.  The article assumes it is showing CEP, but we have no way of knowing that...  For instance, here is the description from the Garmin GPSmap 76C receiver:

"In many instances the map displays an 'Accuracy Circle' surrounding the 'Position Icon'.  The GPSmap76C uses both map resolution and GPS accuracy in defining your location to a point within the circle."

Nowhere does it explicitly say anything about CEP, 2DRMS or %confidence, so we can only guess what the receiver means.  Only the Garmin software developers can tell us what their software does, and they don't do that in their manuals...  (Well, in the specifications, they state the generic GPS accuracy figures of "<15m, 95% typical", with a footnote stating this could rise to 100m 2DRMS with Selective Availability).

Then there's the issue that CEP assumes a nice, regular, normal distribution of error in 2 dimensions.  Whereas the shape of the error surface computed from the signals received from a satellite constellation is complex and dynamic, as the constellation and receiver move...  So the figure displayed is a further approximation...  For instance, if the constellation was such that the satellites all happened to be nicely spaced on an azimuthal arc, then the accuracy along that azimuth would be pretty good, but the accuracy perpendicular to the azimuth would be awful; we'd have an error ellipse, not an error circle.

Finally (and perhaps most importantly), what the article doesn't mention is that calculated CEP and SBAS corrections are all well and good, but in a 'hostile environment' such as urban or real canyon, with strong reflections and blocked direct paths, the reported error may be very optimistic; the reported position may be 100s of metres in error, but the 'accuracy' may still look 'reasonable'.

No, you cannot implicitly trust what your GNSS receiver is telling you, and yes, you have to consider when it might be 'lying' to you...

Lost Soul

  • Sr. Member
  • ****
  • Posts: 265
    • View Profile
Re: Your GPS is lying to you...
« Reply #2 on: April 16, 2014, 07:16:01 PM »
Thanks Angle of Repose.  You learn something every day.  The perceived accuracy is worse than I had been making my self believe.  I had always very cynically assumed that GPS manufactures were quoting accuracy to 1 standard deviation which equates to 68% probability.  Knock it down to 50% to tighten up the accuracy number - make it smaller and hey presto deceive the users into believing the gadget is more accurate than it is.

So going forward from the article and noting CPS comments about errors due reflections etc.  Assuming direct line of sight to all visible satellites then accuracy is only valid for 50% of the time for the displayed figure, 95% of the time for 2 times the displayed figure and 99% of the time for 3 times the displayed figure.  Worth bearing in mind.

MoonMan

  • Jr. Member
  • **
  • Posts: 69
    • View Profile
Keeping Track of where Here is in relation to There.