Trilateration

[csteddy]

Hi,
I'm wondering why the accuracy of the location of the source of each SSID (usually a router) doesn't increase over time with greater sampling? I would have thought that my own SSID would be pinpointed to my house, rather than 30 or 40m away?

Regards,

Craig

[bobzilla]

It should, if there are observations from all sides. If all observations are on the north side of a device, it will always appear to be north of its actual location.

[csteddy]

The Android app itself seems to be quite accurate, but not the website map. I'll do some investigation and give a few examples.
btw. How to you deal with outlier samples (eg. a router is moved, or a wacky GPS co-ord)?

[bobzilla]

There's an iterative filtering that biases towards newer clusters.