GPS accuracy

The gear needed for wardriving

4 posts • Page 1 of 1

Postby FredR » Sun Nov 14, 2021 3:10 pm

when I watch the map with only my data, I can see that there are little dots in place I never went. Also some strings of points in linear alignment in odd places.
Also I have data points in the middle of a ponds.
Do you know if there is something that can be made to fix that and filter odd data ?

Postby arkasha » Sun Nov 14, 2021 9:28 pm

these issues happen because of several reasons:
1. the long lines sound like GPS issues - these are (unfortunately) common in older devices.
2. points are trilaterated - the centers of the signal are estimated according to an algorithm. Bad GPS coordinates can contribute to points being located in odd spots, as can other users observing them in conflicting places.

The short version of how to fix it:
1. observe the points from multiple places and at multiple times, and they'll snap to the right locations
2. if you discover a particular device has GPS problems, you might consider replacing it. this can be diagnosed by turning on "display routes" in the WiGLE WiFi Android app, or mapping your results before uploading with other packages!



Postby TROMOS » Wed Dec 08, 2021 12:32 am

Another major factor that causes odd locations to appear on the map is also due to trilateration, but it isn't GPS accuracy that is to blame. I have several small clusters of points in countries that I have never been to! Usually in major cities.
The explanation is simple. I discover a mobile hot-spot. It is recorded at my location. The tourist carrying the mobile then returns home and someone else logs the hot-spot. The point is now modified to be halfway(*) between the first and second sightings and could well be in a river or mid-ocean. Over time, as more data comes in, the point will shift closer to where the hot-spot actually spends most of its time.
There will be an increasing amount of this as there is a huge growth of mobile hotspots, especially on public transport, cars and mobile phones.

(*) Probably not exactly halfway as I believe trilateration also uses things like signal strength for improved accuracy.

Postby arkasha » Wed Dec 08, 2021 1:00 am

mostly - we also do cluster detection - points should "snap" not drift in a perfect world. When you se things in weird places, it generally means:

1. they were reported there, improbable though it may be
2. they're PRECISELY POISED between two clusters, or have a very small number of observations.
3. they haven't been seen in a while and were last trilaterated before cluster detection was implemented.

4 posts • Page 1 of 1

Return to “Net Hugging Hardware and Software”

Who is online

Users browsing this forum: No registered users and 7 guests