Quote:
|
Originally Posted by syn-ack
Scurge -- is there a way to make a scatter plot that weights all collected signals? IE: good signal strenght would make a larger circle, and S/N would be coloured from bad to good?
|
If you are referring to gradient/contour plots, I looked into them some time ago. The suggestion is good and I would have loved to have incorporated, but the math looked over-whelming and extremely time consuming.
After receiving numerous sample logs from other users and seeing the data most people were recording I didn't think it was worth the effort.
Currently Knsgem basically takes the GPS data for each network and checks to see what data exist within a circular sector with a 10 degree arc. The result is 36 circular sectors which are used to group the data, like the spokes of a wagon wheel. Knsgem then calculates the distance from the center to every point within each sector and flags the most distant point within that sector. It next does a rolling average from sector to sector, not a true scatter plot. A true scatter plot winds up looking like a hull.
What I've found is that most of the data is clumped together in just a couple of sectors. What one might expect when a wardriver does a drive-by. There's a ton of data within the approach sectors followed by a bunch 180 degrees in the departure sectors, but not enough to do a good contour plot.
The problem is further compounded by the fact that many large corporate campuses use the same bssid in many of their wireless devices. Then add all the wifi printers like "hpsetup" that re-use the same bssid in many of their products. Which is why I added the ability to limit the max distance of networks, to keep a network from having the appearance of having 1000 square mile coverage.
At the time the problems seemed exhausting so I took a more reasonable approach and designed knsgem for the hobbyist.
I do have a few other things that I'll be adding in the near future. One of which may come close to a contour plot but via a simpler method.