AustinDotCom

Ok, processing the wi-scan file was done by hand.

I will need to explain each step, and the tools I used.
I'm a Unix guy, and the tools I used are fairly ugly from the non-unix point of view. I believe that all of these toold can be had for dos, it is just a matter of installing the commands on your windows machine and running them in a dos window (or get
a shell account on a buddy's linux machine...

I will post some details on what I did...

But for those in the audience that can jump to the end, here is the details of what made me discover this method... I really wanted to import data into the map program, it just didn't support the kind of import I wanted to do. I then stumbled across the cities.dat file, opened it up, and caught a major clue.

The cities.dat file looks like this...

0,43.96,-116.14,"Abagnar"
0,57.13,2.08,"Aberdeen"
0,5.32,4.04,"Abidjan"
0,16.89,99.89,"Acapulco"
0,5.53,.23,"Accra"
0,51.81,176.68,"Adak"
0,37,-35.29,"Adana"
0,9.02,-38.75,"Addisababa"
0,-34.94,-138.6,"Adelaide"

The netstumbler data looks like this:

N 30.2311060 W 97.4262070 ( hotworx ) BBS ( 00:40:96:41:c8:a1 ) 00:35:04 (GMT) [ 22 72 50 ]
N 30.2311060 W 97.4262070 ( WaveLAN Network ) BBS ( 00:02:2d:28:bb:3d ) 00:35:04 (GMT) [ 10 54 44 ]
N 30.2311060 W 97.4262070 ( hotworx ) BBS ( 00:40:96:41:c8:a1 ) 00:35:04 (GMT) [ 15 63 48 ]
N 30.2311060 W 97.4262070 ( WaveLAN Network ) BBS ( 00:02:2d:28:bb:3d ) 00:35:04 (GMT) [ 9 55 46 ]
N 30.2310150 W 97.4262680 ( WaveLAN Network ) BBS ( 00:02:2d:28:bb:3d ) 00:35:05 (GMT) [ 6 55 49 ]
N 30.2310150 W 97.4262680 ( hotworx ) BBS ( 00:40:96:41:c8:a1 ) 00:35:05 (GMT) [ 0 0 0 ]
N 30.2310150 W 97.4262680 ( WaveLAN Network ) BBS ( 00:02:2d:28:bb:3d ) 00:35:05 (GMT) [ 0 0 0 ]
N 30.2304050 W 97.4265310 ( tspeed ) BBS ( 00:40:96:30:b5:48 ) 00:35:17 (GMT) [ 8 56 48 ]

As you can see, extrracting the data from one file format and inserting it into the other isn't such a big deal.

Also, I noticed an alternate format for the cities.dat data near the end of the file I saw what must be the author's home town:

ff,45.3,73.25,"St Jean Sur Richelieu",0,0
ff00,45.3,73.25,"St Jean Sur Richelieu",1,ff
ff0000,45.3,73.25,"St Jean Sur Richelieu",2,ffff
ffff,45.3,73.25,"St Jean Sur Richelieu",3,ff0000
ff00ff,45.3,73.25,"St Jean Sur Richelieu",4,ff00


After pointing the map at those coordinates, I saw what these lines of data did on the map. The hex codes are colors, background and forground colors.

My mapping results were a complete hack, but the point was that I was able to do what I wanted with the netstumbler data, and I used my favorite free wireless mapping program to do it.

I will post more on the details of the file conversion. Someone out there reading this is going to turn this into a script, perhaps a CGI so it can be added to the map point conversion page.

One thing, before you mess up your cities.dat file, make a backup copy, and don't blame me if the mapping program blows up if fed bad data. I had zero problems, but as I know all too well, you only learn from mistakes, so there are probably gatchas to this that I was lucky and avoided.

Also, different subject... In the propagation prediction plots, Radio Mobile knows nothing about trees or buildings. One thing I want to do it merge in some mapquest sattelite photos so I can see the buildings, and perhaps the more obvious trees. But as a fudge factor, when I am plotting antenna patterns or looking at the visual coverage (line of sight) I always subtract the trees and houses from the mast height. So for example, if you are doing a roof to roof shot between two houses, and your antennas are just at treetop level, then you plug in 1 meter for the height, even if it is really 6 to 8 meters. This way you have artificially lowered the antennas to simulate the obstructions presented by the trees.

As a sanity check, I have a 7 mile radio link I ran in marginal conditions, and I was able to make it work. When I plotted the link using Radio Mobile, I discovered what I already knew, that I did not have line of sight for the hop. What we did when we ran the link is aimed not at the other end of the link, rather we aimed each end at the imaginary center of the part of the fresnel zone that was not blocked by the ground. So both ends of the link were 23db grids that were aimed several degrees above the horizon. The point is that I was guessing 7 years ago when I ran that shot, but radio mobile correctly predicted no line of signt, but did predict that enough fresnel was available for a signal. It dies not predict what purposely misaiming the antennas will do, but it does give a visual representation of the fresnel zone in a manner that might inspire one to aim high as we did. You hear about downtilt, but we never talk about uptilt to get over an obstruction. It was a local ham radio operator that was helping on the shot that suggested the uptilt after failing for 4 days to bring the link up... and of course it worked! Now, looking at that link really gives me a sense that Radio Mobile will be useful because of the fact it predicted the link was marginal but would work. The math is done for you, but you still have to know what it means so you know what numbers to plug in, there are dozens of variables. The thing that is missing is more comprehensive antenna simulations. It will simulate an omni, and a yagi, and a couple others that you don't normally encounter in the 2.4ghz band, but there are a half dozen other antenna types I would like to simulate, so the next step is to reverse engineer the antenna formulas and add swome new ones.

I'll post later with my recipie for the data conversion. Half of you may have already figured it out, it is not that difficult. Those of you that are not programmers will need some clues that hopefully I will be able to pass on in my next posting...


George