WiFi Mapping Question

We've all seen the WiFi maps produced by wardrivers such as

http://www.wificharlottetown.org/loc...ttetownmap.jpg

You'll notice that the plotted positions represent where the CAR was when it saw
a particular AP, not the location of the AP itself.

It occurs to me that there ought to be some way to estimate and plot the
location of the access point, using the GPS position coordinates in conjunction
with the signal strength values at those positions. The GPS receiver is spitting
out position updates roughly once per second, and wardriving programs like
Network Stumbler save an RSSI value for each one.

My question is this: given a set of positions and corresponding RSSI values, how
would you estimate the actual position of the AP? What algorithm would you use?

--
Friends don't let friends shop at Best Buy.

"Scott in Aztlan" <(E-Mail Removed)> wrote in message
news:(E-Mail Removed)...
> We've all seen the WiFi maps produced by wardrivers such as
>
>
http://www.wificharlottetown.org/loc...ttetownmap.jpg
>
> You'll notice that the plotted positions represent where the CAR was when
it saw
> a particular AP, not the location of the AP itself.
>
> It occurs to me that there ought to be some way to estimate and plot the
> location of the access point, using the GPS position coordinates in
conjunction
> with the signal strength values at those positions. The GPS receiver is
spitting
> out position updates roughly once per second, and wardriving programs like
> Network Stumbler save an RSSI value for each one.
>
> My question is this: given a set of positions and corresponding RSSI
values, how
> would you estimate the actual position of the AP? What algorithm would you
use?

With only the change in signal strength as you drive down the street, I
would expect two possible locations, one on each side of the car's path.
You'd have to pass through the AP's signal a second time on a different
path. Perhaps just the other side of the street would do.

"MarcRW" <(E-Mail Removed)> wrote in message news:<ThQ8b.37233$S_.24448@fed1read01>...
> "Scott in Aztlan" <(E-Mail Removed)> wrote in message
> news:(E-Mail Removed)...
> > We've all seen the WiFi maps produced by wardrivers such as
> >
> >
> http://www.wificharlottetown.org/loc...ttetownmap.jpg
> >
> > You'll notice that the plotted positions represent where the CAR was when
> it saw
> > a particular AP, not the location of the AP itself.
> >
> > It occurs to me that there ought to be some way to estimate and plot the
> > location of the access point, using the GPS position coordinates in
> conjunction
> > with the signal strength values at those positions. The GPS receiver is
> spitting
> > out position updates roughly once per second, and wardriving programs like
> > Network Stumbler save an RSSI value for each one.
> >
> > My question is this: given a set of positions and corresponding RSSI
> values, how
> > would you estimate the actual position of the AP? What algorithm would you
> use?
>
> With only the change in signal strength as you drive down the street, I
> would expect two possible locations, one on each side of the car's path.
> You'd have to pass through the AP's signal a second time on a different
> path. Perhaps just the other side of the street would do.

Kismet has a feature that helps to figure out this problem. I don't
know how he is doing it... but...

If your sensor antenna is an omni-directional receiver. You can make
some simplifying assumptions:

1) The coverage area is a circle centered on the receiving antenna.

2) The signal strength will follow a curve, from weak to strong to
weak again as you pass the emitter.

3) During the time you heard the emitter, your coverage circle moved
in some pattern along the ground and generated a shape. If you went
straight, the shape is an oval (sort of). If you turned, it might
look more like a kidney bean.

4) You know the speed at which you were traveling and where you were
when each time you got a signal reading. You can therefore
extrapolate the size and shape of the curve generated by your movement
past the emitter.

5) You can then overlay that curve (signal strength vs motion curve)
over the area which you sensed (the kidney bean). If you were turning
when you took your readings, it will only fit on one side of the
?bean? or it will better fit one side than the other side. If you
were going straight, a second pass will generate a second similar
geometry and will likely solve the matter fairly accurately.

Mind you all this is basically wasted work at these frequencies and
power levels. If you drive past something at 50 MPH and hear it for
three seconds, you likely passed within 50' of it and if it is more
than 300 feet from the road it is very unlikely you are going to hit
it at all. I guess I am saying you are looking for a solution to a
problem that is easily solved by keeping your eyes open and paying
attention to the SSID.

Another option is to use a directional antenna and point it toward one
side of the road all the time. Use panel with 8 to 12 dB of gain, a
pringles can doesn't have the front-to-back ratio for the task. When
you hear it, it's at 90 degrees, pretty much every time.

-m-

In message <(E-Mail Removed) >, Michael
Erskine <(E-Mail Removed)> writes
>If your sensor antenna is an omni-directional receiver. You can make
>some simplifying assumptions:
>
>1) The coverage area is a circle centered on the receiving antenna.

Not if you're using a PCMIA/Cardbus card in a laptop. These things are
horizontally polarised and give two lobes, diametrically opposite in the
axis at right-angles to the aerial (normally left and right of the
laptop). One of these lobes, however, is partially screened by the
metalwork of the laptop.

It's always been this way. Vertically polarised dipole is
omnidirectional, horizontally polarised dipole gives two lobes.
Propagation doesn't change with frequency - have a look at horizontally
polarised VHF/FM dipole aerials.

I've just done some trials on this BTW, and I'm in the process of
putting together a webpage.

--
Tony Morgan
Smile in the face of adversity - and adversity will probably
think you're taking the piss and kick the shit out of you.