Erroneous GPS data

[Legodude]

Does wigle ignore erroneous GPS data that programs such as KisMAC make when there is no GPS present? I accidently uploaded a file with some of this type of data on accident. It seems to have an affect on my stats for it counted from as new logged networks with location.

For example, KisMAC will log the network as Lat: -1.985062S Long: -0.0W

[uhtu]

ignore is a strong word. we do actively filter bad data, and have been constantly advancing that filtering over this project's 8.5 year lifespan.
if you have a particularly troublesome posting, send us the transid to wigle-admin@wigle.net and we'll deal with it.

[Legodude]

Thank you taking the time to answer my question! I don't think there was too much "bad" data.

[gclouse]

I know tis thread is a little old, but it seems to be the place to mention this.

I've been war driving a bit with an Archos 5it (Android 1.6 ) and I noticed that I have a few errant locations that are logged when the GPS doesn't have a satellite fix before the Wigle app is started.The entries for the AP don't have a time stamp, so it should be a simple matter of creating a delete query that removes all data with no time stamp. I have tested the concept on a copy of the database on my PC, I think I might be able to do the same on the Archos with an RFO Basic program. If it works on the Archos 5it, it may or may not work on other devices.

[Surtr]

When investigating hot spots that have gone "missing", I found that erroneous GPS data has caused the hotspots to be "triangulated" to far away from where the actually are. Querying the location data of a BSSID I know the actual location of, there are a lot of pretty accurate observations, but a few that are horribly wrong. Some are completely invalid, such as 158.72918701, -416.63558960. Some are thousands of miles out to sea (38.99082565, -138.9684143). The end result so far is that the hotspot is about 6 miles away from it's actual location.

I have no idea what is causing this to be so far off, how much inaccurate data I (and others) have contributed, or how many hotspots have been thrown off because of this.

Inaccurate data, such as the one containing -416, should just be disregarded. I'm not sure if it would actually weigh towards the averaging for the triangulation, but it's certainly junk. I would suggest just throwing out data that is x amount of miles from an AP with at least a certain amount of other observations, but I have no idea about any unforeseen consequences of that.

[bobzilla]

We do a lot of filtering, and if there are clusters of observations outliers do get thrown out. A longitude less than -180 gets thrown out as that's not on our planet, so that isn't a concern