[Cu-wireless] Wireless map updated (again)

[David Young]

Thanks for the new map, Stephane.

Will your software detect an IBSS, or only a BSS? I am surprised that
you have never detected my adhoc network at home.

BTW, a really slick browser for 802.11 maps can be programmed in Java
using the Jazz API (http://www.cs.umd.edu/hcil/jazz/).

I throw out this suggestion for a complementary coloring of the
map. Essentially, color a pixel green by the probability that the "green"
station is located there, given the strength of green's signal at every
GPS location, and the uncertainty in your GPS location. Use some simple
model of radio propagation to get the probability distribution function:
e.g., inverse square law. (Multipath and other phenomena probably make
that model WAY too simple.) Use high-intensity green for very likely
locations of the green station, low-intensity for unlikely locations.

I wonder if you can infer microwave propagation characteristics of a
landscape from characteristics of the landscape's satellite photographs,
and that way produce a superior probability distribution for a station's
location. I mean, suppose you have some function you can apply to an
NxN pixel square in your satellite photo that will tell you that there
is, with probability P, a tree in that square, and with probability Q,
there is a building, P plus Q not exceeding 1. =) By applying some other
function to P and Q, you will figure the likely power that microwaves
entering that square from one side will emerge from the other three
sides of the square.

I can probably think of a dozen other ways to produce a more sophisticated
probability distribution, but the end result is the same: you solve
for the zillion parameters in your probability distribution function
that maximize the log-likelihood of your collected GPS/signal strength
measurements. Your home PC might be adequate to solve for the parameters;
if not, well, that's what a supercomputer is for.  But once you've solved
for the parameters, all that's left is to color pixels.

Dave

On Sat, Apr 27, 2002 at 12:08:27AM -0500, stephane at shimaore.net wrote:
> 
> Hi,
> 
> I udpated the "wireless" map with data I collected last week-end (a little
> bit around Race Street as Sascha requested, and some along Green Street).
> I also did a major revamp of the way the data is presented. The page now
> shows wireless networks and (new!) GPS data; this gives you an idea of
> where I've been driving -- and potentially, where there are no open access
> points. Showing all the data while keeping the images readable prooved a
> little bit challenging; hopefully the compromise I came up with makes some
> sense.
> 
> Still same place: http://mapper.shimaore.net/Mapper/wireless/index.html
> 
> HTH,
> Stéphane
> 
> 
> _______________________________________________
> Cu-wireless mailing list
> Cu-wireless at lists.groogroo.com
> http://lists.cu.groogroo.com/cgi-bin/listinfo/cu-wireless

-- 
David Young             OJC Engineering from the Right Brain
dyoung at onthejob.net     Urbana, IL * (217) 278-3933