Bad Data in Texas

Suggestions for WiGLE/JiGLE/DiGLE

6 posts • Page 1 of 1

Postby khmann1 » Tue Jan 13, 2015 12:35 am

Obviously, every once in a while you get some bad data when somebody's GPS goes haywire for example.

However, I think there's really something wrong here, because the satellite view shows it to be the middle of nowhere. ... mapzoom=17

Your thoughts? Perhaps a way to report suspect data...

Postby mahir256 » Fri Jan 16, 2015 4:17 am

viewtopic.php?f=1&t=2238#p8353 explains it.

Ideally any access point without a name (or whatever the period is supposed to signify) should be removed from the database immediately.
There should also be a UI for administrators which allows one to delete APs from the map, so that an entire geographic area can be cleared.

Postby RyanHLouw » Fri Jan 16, 2015 11:03 am

Why should accesspoints without a name be removed? The problem is not the unnamed APs, but the APs generated with FakeAP/similar software. These are named APs.

Postby mahir256 » Fri Jan 16, 2015 12:42 pm

If you zoom in on US 290 between Austin and Houston, all APs with "." as a name are generated by FakeAP. That's why I suggested removing them all.
Other fake names should be evaluated on a geographic basis.

Postby khmann1 » Sat Jan 17, 2015 2:37 am

blank or "hidden" SSIDs are a special kind of problem because each stumbler seems to report them differently. Kismet exports typically show up as " (Cloaked) - no that's not a typo, the quote doesn't get escaped right or something. Some clients as <no ssid> or <hidden ssid> or (null). WiFiWhere on iOS chooses to report the MAC address as plain uppercase text which looks cool on the map at first but quickly becomes visually distracting. sqlite3 or csv uploads can omit the ssid field and so seem to get stored as a "true" null which is the "black dots" you see.

Some beacons (from Sonos systems and other mesh network) do not include an SSID element at all in such case the SSID truly doesn't exist - not even as a null network. anyway, the point is you can't judge a "fake" by it's SSID or presence thereof. there's a lot of data and obviously "sketchy" data is undesirable, but is clearly better than no data.

A technical approach to data validation gets complicated, but at some point could be necessary due to: iOSlike hotspot random MAC address, in my area several independent (different company) Cisco enterprise deployments are re-using MAC addresses - this may be a trend, might be nice to classify "mobile" mac addresses, at some point the database gets big and stale and will need to be filtered,summarized.

In the specific case of US290, it is very likely that some of the generated macs are in-fact that of real networks in other parts of the world, hopefully the transactions can be rolled back. I'm curious to watch how this goes down... I want accurate wireless geolocation data, I'd happily volunteer to "mod" uploads or work on rogue detection algorithms.

Postby ithink314 » Thu Feb 05, 2015 12:12 am

At least the quality of service indicator is still OK? If those start turning green, then it's an even bigger problem, and accomplishment(?).

6 posts • Page 1 of 1

Return to “WiGLE Project Suggestions”

Who is online

Users browsing this forum: No registered users and 6 guests