Calculating exact location of APs - Page 2

Skirecs · 01-05-2005

Quote:

Originally Posted by Tuckie

wouldn't gathering more points make the real location more accurate? Having say 100 readings and then averaging them through some form of triangulation would at least compensate somewhat for physical barriers?

Triangulation wouldn't like a GPS, because GPS measures time difference from time signal is sent and time signal is recieved, and that is how diastance ffrom an object is calculated. With out the transmit and recieve time it would be very difficult to do this, and the short distancdes would require very accurate clocks.

True more points equals more accuracy, but start off small and work our way up.

No offense, but jsut taking 100 readings and somehow avgeraging them would not give a location.

Skirecs · 01-05-2005

In the picture, the cicle represents a set level, say 50 dB or whatever, whenever that level is attain, a mark is made, those are the four points.
Thoretically they are on a crcle surronding the AP, so it wouldn't be hard to find a circle regression.

jdc2003 · 01-05-2005

In reading your posts, on thing you might try is making a directional antenna (plans in Wireless Hacks published by O'Rielly). That way you can know that direction the signal is coming from. My bad news is that although the signal may seem to be from an omni-directional AP, very few are truely that, they may be close but not completely. The antennas on laptops and pdas are slightly directional as well (I loose the signal all the time on mine by turning it slightly). Sorry for the bad news (correct me if I'm wrong on any point)

-jdc2003

Skirecs · 01-05-2005

That would pose a problem, on top of all the obstacles disturbing the theoretical perfect circle of signal strength, the non-omni-directionalness of the AP would have to taken into account, or just disregarded. Depending on the difference in the axes of the theoretical ellipse of signal strength, it could just make the calculations off by a few feet, which for now, I think would be acceptable.

If there are any scripters that could help me with my idea, I would greatly appreciate it. I have no experience in coding in VB, but am willing to help with the math, because I believe it is possible.

Even for a triail run, get the points, and do that math manually, it's a start.

Airstreamer · 01-05-2005

Quote:

Originally Posted by Skirecs

That would pose a problem, on top of all the obstacles disturbing the theoretical perfect circle of signal strength, the non-omni-directionalness of the AP would have to taken into account, or just disregarded. Depending on the difference in the axes of the theoretical ellipse of signal strength, it could just make the calculations off by a few feet, which for now, I think would be acceptable.

If there are any scripters that could help me with my idea, I would greatly appreciate it. I have no experience in coding in VB, but am willing to help with the math, because I believe it is possible.

Even for a triail run, get the points, and do that math manually, it's a start.

The big boys (Like the FCC and other TLA agencies) use a phased array antenna structure to get real time directional info as they move around in an area. As you move, you get a vector in the direction that the signal is coming from. True, you do get reflections, but if you were to plot out several hundred reads as you moved around in an area, the intersection of the vectors will pretty much localize where the signal is coming from.

It is basic triangulation run 'on the fly' with some pretty sophisticated gear, but a good RF engineer and a computer programmer working together could do the same. There is commercial equipment on the market that the average citizen can get, but it is pretty expensive.

Steve

ShadowFX · 01-21-2005

Quote:

Originally Posted by Airstreamer

The big boys (Like the FCC and other TLA agencies) use a phased array antenna structure to get real time directional info as they move around in an area. As you move, you get a vector in the direction that the signal is coming from. True, you do get reflections, but if you were to plot out several hundred reads as you moved around in an area, the intersection of the vectors will pretty much localize where the signal is coming from.

It is basic triangulation run 'on the fly' with some pretty sophisticated gear, but a good RF engineer and a computer programmer working together could do the same. There is commercial equipment on the market that the average citizen can get, but it is pretty expensive.

Steve

Now you’ve gone and done it. You have p…ed off my wife. She just got me into a routine of doing some work around the house during our hibernation time (-28°C up here). But now that I’ve read this thread, I can’t think of anything I’d rather do than solve this problem.

First off, as was pointed out by jdc2003 the AP antenna pattern is not a circle, however that doesn’t preclude you from thinking of it as one, if you are standing still.

Now, the two receive setup is most interesting. NS can find you the beacon, but it is not likely that its time stamp is accurate enough. However, if you switch to sniffing packets, and time stamping them with your GPS clock, you can compare known packets’ timestamp, and that will give you the distance delta from the AP to each of the two receivers. Given the baseline of your triangle, and the delta of the two remaining sides, you can determine a straight line upon which the AP must be located. Now, move one receiver and sniff another beacon packet (or packets) and calculate a second straight line. The AP should be on the intersecting point of these two lines, give or take the 20 meter cube accuracy of GPS. The more packets you get the more accurate you will be.

Just a thought.

Airstreamer · 01-22-2005

Quote:

Originally Posted by ShadowFX

Now you’ve gone and done it. You have p…ed off my wife. She just got me into a routine of doing some work around the house during our hibernation time (-28°C up here). But now that I’ve read this thread, I can’t think of anything I’d rather do than solve this problem.

First off, as was pointed out by jdc2003 the AP antenna pattern is not a circle, however that doesn’t preclude you from thinking of it as one, if you are standing still.

Now, the two receive setup is most interesting. NS can find you the beacon, but it is not likely that its time stamp is accurate enough. However, if you switch to sniffing packets, and time stamping them with your GPS clock, you can compare known packets’ timestamp, and that will give you the distance delta from the AP to each of the two receivers. Given the baseline of your triangle, and the delta of the two remaining sides, you can determine a straight line upon which the AP must be located. Now, move one receiver and sniff another beacon packet (or packets) and calculate a second straight line. The AP should be on the intersecting point of these two lines, give or take the 20 meter cube accuracy of GPS. The more packets you get the more accurate you will be.

Just a thought.

This is getting a bit off from the topic of scripts....but...

The time frame you are talking in is down in the 10s to 100s of nanoseconds region. GPS clock signals over a serial or even USB interface are not going to have the resolution you need, if I understand what you are proposing.

That is why there is specialized circuitry designed to do the phase analysis to come up with the direction vector.

Just the distance the sync signal would have to travel from the GPS unit to both the receivers would introduce delays that would mess up your geometry.

Steve

ShadowFX · 01-22-2005

Quote:

Originally Posted by Airstreamer

This is getting a bit off from the topic of scripts....but...

The time frame you are talking in is down in the 10s to 100s of nanoseconds region. GPS clock signals over a serial or even USB interface are not going to have the resolution you need, if I understand what you are proposing.

That is why there is specialized circuitry designed to do the phase analysis to come up with the direction vector.

Just the distance the sync signal would have to travel from the GPS unit to both the receivers would introduce delays that would mess up your geometry.
Steve

Each unit has its own GPS

Good point, one of the problems of doing research for NRC, is the access to neat equipment.
So, as you pointed out, I am off topic so I'll bugger off, but not without first thanking everone for giving me something in to which to sink my teeth.

Shawn

Airstreamer · 01-22-2005

Quote:

Originally Posted by ShadowFX

Each unit has its own GPS

Good point, one of the problems of doing research for NRC, is the access to neat equipment.
So, as you pointed out, I am off topic so I'll bugger off, but not without first thanking everone for giving me something in to which to sink my teeth.

Shawn

PM'd ya. Now you've got MY interest up!

Steve

icefreez · 02-18-2005

You also have to take into consideration hareware and software delays. You would have to set your GPS and your wireless card to scan at the fastest rates possible.

I know on my wireless card I have driven by a location where I know an AP is located, and NetStumbler will not detect that AP visually untill I have almost passed it. This would hinder any kind of triangulation script and throw the center off by quite a bit depending on how fast your driving.

Also any kind of averaging would not work in a triangulation script. You would definantly have to calculate the mean or throw out several locations out of the general range of data, to try and compensate for the obstructions inbetween you and the AP.

beakmyn · 02-18-2005

Quote:

Originally Posted by icefreez

You also have to take into consideration hareware and software delays. You would have to set your GPS and your wireless card to scan at the fastest rates possible.

I know on my wireless card I have driven by a location where I know an AP is located, and NetStumbler will not detect that AP visually untill I have almost passed it. This would hinder any kind of triangulation script and throw the center off by quite a bit depending on how fast your driving.

Also any kind of averaging would not work in a triangulation script. You would definantly have to calculate the mean or throw out several locations out of the general range of data, to try and compensate for the obstructions inbetween you and the AP.

Does this also mean there will be a Fudd factor?

Thank you. I'll be here all week. Try the Veal.

nashr · 02-18-2005

Quote:

Originally Posted by beakmyn

Does this also mean there will be a Fudd factor?

Thank you. I'll be here all week. Try the Veal.

View Elmer's Forums here.

Sorry, but you'll have to click the Dialectize button. I can't get it to go directly to the results.

rotinom · 04-18-2005

I'm actually playing around w/ GPSMap which is a part of the Kismet project. It uses a variety of map sources (such as satellite maps, mapquest maps, etc) to generate maps similar to what you're speaking of.

The big problem I've seen so far, is that you do need a lot of data to get the most accurate picture. (For instance, in the attached picture, it found a bunch of ap's in the front of my building, just because I walked around it, not all the way down the block ,etc..)

If you really want to located ap's, you're best bet is w/ a directional antenna to locate it.

Airstreamer · 04-18-2005

Quote:

Originally Posted by rotinom

I'm actually playing around w/ GPSMap which is a part of the Kismet project. It uses a variety of map sources (such as satellite maps, mapquest maps, etc) to generate maps similar to what you're speaking of.

The big problem I've seen so far, is that you do need a lot of data to get the most accurate picture. (For instance, in the attached picture, it found a bunch of ap's in the front of my building, just because I walked around it, not all the way down the block ,etc..)

If you really want to located ap's, you're best bet is w/ a directional antenna to locate it.

Yep. I came in from the east side of Bellevue, Wa. drove down around ONE block and then headed back out on the same street I came in on. When plotted with Jigle, it showed a thick line made up of a bunch of APs' going around in a perfect square!

Steve

G8tK33per · 04-19-2005

Quote:

Originally Posted by rotinom

I'm actually playing around w/ GPSMap which is a part of the Kismet project. It uses a variety of map sources (such as satellite maps, mapquest maps, etc) to generate maps similar to what you're speaking of.

The big problem I've seen so far, is that you do need a lot of data to get the most accurate picture. (For instance, in the attached picture, it found a bunch of ap's in the front of my building, just because I walked around it, not all the way down the block ,etc..)

If you really want to located ap's, you're best bet is w/ a directional antenna to locate it.

Well slap my ass and yank out the gerbil...and all this time I've been using omni's!!

01-05-2005	#18 (permalink)
jdc2003 Registered Member Join Date: Jan 2005 Posts: 1	directional and "omni" directional antennas In reading your posts, on thing you might try is making a directional antenna (plans in Wireless Hacks published by O'Rielly). That way you can know that direction the signal is coming from. My bad news is that although the signal may seem to be from an omni-directional AP, very few are truely that, they may be close but not completely. The antennas on laptops and pdas are slightly directional as well (I loose the signal all the time on mine by turning it slightly). Sorry for the bad news (correct me if I'm wrong on any point) -jdc2003

Thread Tools
Show Printable Version Email this Page
Display Modes
Linear Mode Switch to Hybrid Mode Switch to Threaded Mode

01-05-2005	#19 (permalink)
Skirecs Registered Member Join Date: Dec 2004 Posts: 13	That would pose a problem, on top of all the obstacles disturbing the theoretical perfect circle of signal strength, the non-omni-directionalness of the AP would have to taken into account, or just disregarded. Depending on the difference in the axes of the theoretical ellipse of signal strength, it could just make the calculations off by a few feet, which for now, I think would be acceptable. If there are any scripters that could help me with my idea, I would greatly appreciate it. I have no experience in coding in VB, but am willing to help with the math, because I believe it is possible. Even for a triail run, get the points, and do that math manually, it's a start.

02-18-2005	#25 (permalink)
icefreez Banned Join Date: Dec 2004 Location: South Dakota Posts: 32	You also have to take into consideration hareware and software delays. You would have to set your GPS and your wireless card to scan at the fastest rates possible. I know on my wireless card I have driven by a location where I know an AP is located, and NetStumbler will not detect that AP visually untill I have almost passed it. This would hinder any kind of triangulation script and throw the center off by quite a bit depending on how fast your driving. Also any kind of averaging would not work in a triangulation script. You would definantly have to calculate the mean or throw out several locations out of the general range of data, to try and compensate for the obstructions inbetween you and the AP.