WiGLE.net

Hi, i've been putting off uploading my scans for a while because of a few reasons. The one I wanted to ask about here is that I was hoping to go back and improve the accuracy of each AP point from multiple reading datapoints using triangulation and signal strength. It doesn't seem like this is done in kml files, but I noticed that the csv exports contain multiple points. When I look at the kml file on google earth its obviously not any more improved accuracy than when I made the scan because there is just a cluster of APs at one point every X feet.

Has anyone else tried to improve the accuracy of their data in this way? I can write a program to do it, but it would obviously save time if something has already been written and well tested. Does the wigle.net database already do this?

Yes, WiGLE takes multiple observations into account, see the triangulation section of the faq: https://wigle.net/faq

I am responding to Bobzilla's comment, "Yes, WiGLE takes multiple observations into account, see the triangulation section of the faq: https://wigle.net/faq"

I previously tried to post a comment about WAAS (Wide Area Augmentation System), which amounts to a kind of corrected GPS. After searching for 'Android WAAS', for instance, I have been entirely unable to find any indication that WAAS-corrected GPS is employed in smartphones. This is odd, because WAAS is broadcast by GPS, and it is probably also transmitted over the Internet.
In any case, if enough information is collected and stored about GPS readings, it should be possible to later correct stored readings with those WAAS corrections, even months or years later. The issue is which satellites' outputs are used to generate the GPS location in question.
While I didn't expect a smartphone to have an accuracy equal to a standalone GPS unit, I have seen commentary on the Internet that the inherent electronic accuracy of a smartphone's GPS isn't dramatically different. But, if WAAS correction isn't ordinarily available, that would certainly account for any perceived inaccuracy of smartphone-derived GPS readings, in comparison to dedicated-GPS accuracy.
Therefore, I encourage the operators of WiGLE to ensure that all readings of GPS contain not merely the as-computed position, but also the specific satellites whose values were used to compute those positions. Those positions might later be corrected to improve the accuracy of the entire database.

Does accuracy get improved by other users observations?

in the centralized database, we apply the trilateration techniques described elsewhere in this forum to the network record you see in the search interface/maps/APIs. We don't modify the individual observations.