Accuracy of GPS information while stumbling

Chris_Schear · 09-08-2002

You'll have to forgive this fairly uninformed set of questions concerning GPS information while stumbling, I've been scratching my head attempting to figure out what would be the logical conclusion. I figured I would just ask...

Alright, so you have your prefered stumbling setup. Maybe you have a home-brew antenna, maybe you don't. Regardless, you have purchased a handheld GPS from BestBuy and want to get the Long/Lat of access points in your area.

When NetStumbler detects an access point, does the CURRENT Long/Lat at the time of detection get reported as the "location" of the AP? If so, this would not be accurate. Especially if you have some addon antenna action going on, the access point may actually be several blocks away (or greater). The Long/Lat dumped into the data is relative to where YOU were when you detected the signal, not where the ACCESS POINT is.

Additionally, for any given signal, does the Long/Lat update in the NetStumbler log if the signal gets greater? Let's say I'm driving down the road and I detect an AP. Boom - The GPS records my current Long/Lat. Now, if I drive two blocks further down the street and the signal is getting stronger, will NetStumbler dump *new* GPS information into the datafile as it would be a more accurate representation of where the access point is?

How accurate is the GPS information while war driving? I would think the Long/Lat information cannot accurately pinpoint where an access point is unless you followed each and every signal until you reach a point where the signal doesn't get any stronger. This is assuming NetStumbler updates the Long/Lat recorded for any access point if the signal increases, perhaps someone can straighten me out.

I can stand in the middle of a fountain near downtown and detect 5 access points with no antenna. If I had my GPS connected, the logs would indicate those 5 access points existed where I was physically standing, with regards to longtitude and latitude, would it not?

I'd be interested in hearing discussions concerning the accuracy of data with GPS.

DaClyde · 09-08-2002

...judging by the mapserver at http://mapserver.zhrodague.net/. If you zoom in on most APs on the map, you see the signal strength indicated by red or green dots of varying intensity depending on where the signal is picked up. I guess if you wanted the most accurate fix possible, you'd set NS to the highest scan speed and anytime you found an AP, drive around the area to get as many signal contact points as possible.

WiGLEs' map server has added polygons to show the boundaries of the AP signals detected.

rogerRabbit · 09-08-2002

You are, of course, correct - the GPS gives the location of its antenna. Marius did post an answer as to how NS handles multiple readings; I'm pretty sure it is the highest SNR that drives the co-ords - but you could search to check.

There's the prospect of using NS's scripting to read out a source's signal strength + your GPS co-ords and how they change over time. You could then take back-bearings from those to derive the source's lat/lon.

And you don't mention the error inherent in the civilian GPS system. If you don't have a good number of satellite contacts your circular error of probability can be of the order of the range of some APs!

jjkaczor · 09-08-2002

Quote:

Originally posted by rogerRabbit
There's the prospect of using NS's scripting to read out a source's signal strength + your GPS co-ords and how they change over time. You could then take back-bearings from those to derive the source's lat/lon.

And you don't mention the error inherent in the civilian GPS system. If you don't have a good number of satellite contacts your circular error of probability can be of the order of the range of some APs!

The error has been "toned" down, if not removed for the last several years, as well public WAAS satellites compensate for that error.

Actually you wouldn't need to backtrack if you had multiple readings from a single-AP, with differing GPS coordinates, export to "text", not summary, then use some tools/scripting to "triangulate" the signal origin based on a combination of signal strength, GPS coordinates of the same MAC-address. Search google for "triangulation".

Chris_Schear · 09-08-2002

Imagine, for any discovered access point and any relevant NetStumbling data contained about said access point, that Lat/Long information and Signal Strength are both cateloged. As you drive down the street, a signal slowly grows to a peek and then bleeds off. This, in essence, defines the area of coverage for that access point. Imagine NetStumbler collecting data and being able to export this type of information to a graphical map. For any "icon" of a discovered access point, you could view a graphical representation of WHERE to WHERE you saw that access point's signal.

Did you receive a usable signal from "Access point #1" from 53rd street until University Ave? (example) Being able to log and display the coverage area, given signal strength and Lat/Long data - would be a very useful set of information.

Even being able to drive down the road, hearing NetStumbler "ping" when it discovers an access point - logging your current Lat/Long information, and then logging the Lat/Long information of that last point when you saw that access point, would provide scope. It would answer the question, "Ok, so I know where an access point is, what are the boundaries of where this point was seen?" I know there is an access point named "ISTS_Wireless" seen on my normal drive home. With a GPS hooked up to your stumbler, I should be able to map out (graphically) a "cloud" of area where that access point was visible.

Seeing icons on a map is one thing, having the option to click a "view coverage area" feature, given recorded Long/Lat daat and see a graphical representation of the signal, is another. The images below are a couple of examples of this. I've faded out the bulk of the image so the focal point is clear. The darkness of the green would signify the strength of the signal. Darker = stronger, obviously.

NOTE: These images are photoshopped to portray the intended feature. Of course, a more 'zoomed in' example would be more relevant.

EXAMPLE #1:

( Normal NetStumbling, without 'range features' )

( Same data with optional 'range overlay' feature )

- - - - - - - - - -

EXAMPLE #2:

( Normal NetStumbling, without 'range features' )

( Same data with optional 'range overlay' feature )

Thorn · 09-09-2002

All of this has been posted before, many times. I would suggest use of the big button on top of the page marked "SEARCH"

To give a quick summary:

1) The Lat/Lon are reduced at the output screen and the summary to those readings that give the strongest signal. So there is a sort of lazy man's triangulation going on. You can see this occur as you move. (Hopefully someone else is driving.)

2) This Lat/Lon updated continuously. The assumption is that the closer you are the higher the signal strength. This works well most of the time. The problem with this is that you may reach a point where signal is stronger due to a clearer line of sight, but where you are actually further away.

As to the mapping/signal imaging, its been done:
http://www.ittc.ku.edu/wlan/procedure.shtml

Chris_Schear · 09-09-2002

I can assure you, searching for "gps AND accuracy OR triangulation" will result in to returns addressing the specific questions I am inquiring about. While you chose to focus on assumptions made concerning the nature in which the NetStumbler application updates Long/Lat data, such context was not a singular point of interest.

Additionally, as perviously stated, "I'd be interested in hearing discussions concerning the accuracy of data with GPS.". What I am not interested in is entertaining the seudo-elitist, search gestapo mentality, which is so common place here. Perhaps, you devote hours of your day perusing every last inch of this message board, in all its individual forums, reading every last post umpteen times - but that may not be the case for each and every individual browsing. That potential fact, aside, vBulletin's implementation of regular search expressions has long since left much to be desired. I would suggest many things for you, none of which lie within scope of this discussion.

To say mapping and signalling, "has been done", is a misnomer, to say the least. When you can point out a downloadable addon, which enables the end user to signal-ranged, Long/Lat data through some generally accepted "standard" of interfaces, say, Microsoft Streets...I would contend the solution"has been done". When you can import NetStumbler data into your selected mapping application and script a "range finder" addon, I would agree the horse is dead.

Regardless, the context of this post was centered around the discussion of real-world application of GPS and its accuracy. Active discussions do not take place through messageboard search parameters. If you choose not to participate, that would be completely acceptable. Your "did you search" retorts are neither valid nor warranted. If you elect not to contribute, you may quite easily click on bye.

mentat · 09-09-2002

Quote:

Originally posted by Chris_Schear
If you elect not to contribute, you may quite easily click on bye.

The thing is a lot of us have been here for quite a while. We have contributed and spent quite a while discussing this "new" subject you wish to discuss. Take for instance the thread (found by searching for "GPS lat lon file") http://forums.netstumbler.com/showth...S+lat+lon+file

It discusses plotting network ranges with RabioMobile, something I have not personally tried but which looks pretty sweet.

Or there's this thread found by searching for "triangulation" http://forums.netstumbler.com/showth...=triangulation
which introduces WiGLE.net. If you actually went there you'd see that the latest version implements the ability to plot all the individual signals you've received.

And fairly recently this thread, found by "NS file format", http://forums.netstumbler.com/showth...NS+file+format
discusses how to access the "extended log" of every signal with it's lat lon.

So you see, search does work. I wasn't sneaky when I typed in these queries. I didn't know what I'd find, but low and behold, the information you were looking for is there. And look further, I'd bet Thorn had contibuted to at least one of those threads, as have I. So, before you say that search doesn't work and if someone's not going to contibute, they should leave, have a look at who's contibuting and apply the statement to yourself. Thanks.

Chris_Schear · 09-09-2002

I never stated "Search didn't work", I said it was ineffectual. Additionally, I never stated anyone should "leave", I said they should click past the thread without depositing their two cents of retorts. I would kindly ask you not place words into my mouth. I'm perfectly cabable of inserting my own foot when necessary, I need not any assistance from yourself.

Thanks.

Thorn · 09-09-2002

Chris_Schear,
Several points:

Quote:

Originally posted by Chris_Schear
I can assure you, searching for "gps AND accuracy OR triangulation" will result in to returns addressing the specific questions I am inquiring about. While you chose to focus on assumptions made concerning the nature in which the NetStumbler application updates Long/Lat data, such context was not a singular point of interest.

When you started the thread, you entitled it: "Accuracy of GPS information while stumbling" Perhaps this is what lead me to the "assumption" that this was the discussion at hand. As to it being a singular point of interest, I would like to just remind you that this is a NetStumbler forum, specifically a FAQ area. General discussions regarding accuracy of GPS aren't relevant here except in the context of the use of GPS with NS.

Quote:

Originally posted by Chris_Schear
Additionally, as perviously stated, "I'd be interested in hearing discussions concerning the accuracy of data with GPS.". What I am not interested in is entertaining the seudo-elitist, search gestapo mentality, which is so common place here. Perhaps, you devote hours of your day perusing every last inch of this message board, in all its individual forums, reading every last post umpteen times - but that may not be the case for each and every individual browsing. That potential fact, aside, vBulletin's implementation of regular search expressions has long since left much to be desired. I would suggest many things for you, none of which lie within scope of this discussion.

As I stated above: General discussions regarding accuracy of GPS aren't relevant here except in the context of the use of GPS with NS.

The search function is there to help prevent the repeat of the same questions and information being posted ad naseum. The fact that vBulletin has shortcomings in this regard does not diminish the fact that it may be used sucessfully to extract information from these forums.

As to the personality remarks, I shan't dignify them with a response.

Quote:

Originally posted by Chris_Schear To say mapping and signalling, "has been done", is a misnomer, to say the least. When you can point out a downloadable addon, which enables the end user to signal-ranged, Long/Lat data through some generally accepted "standard" of interfaces, say, Microsoft Streets...I would contend the solution"has been done". When you can import NetStumbler data into your selected mapping application and script a "range finder" addon, I would agree the horse is dead.

Feel free to program such an add-on. I'm quite sure that any number of people on this board would be interested in seeing it, myself included. The link I provided would give you ample information to start.

Quote:

Originally posted by Chris_Schear Regardless, the context of this post was centered around the discussion of real-world application of GPS and its accuracy. Active discussions do not take place through messageboard search parameters. If you choose not to participate, that would be completely acceptable. Your "did you search" retorts are neither valid nor warranted. If you elect not to contribute, you may quite easily click on bye.

Regardless of your opinion as to whether my responses are valid or warranted, I do have a responsibility here to help maintain this section of the board and keep it on track. Please keep the discussion to GPS with NS.

Grey Wolf · 09-09-2002

Here is my take on this:

1st the gps signal is not updated continuously. But at intervals, if your lucky you can get an update every sec, if your not so lucky (you have an older/cheaper gps it may be longer). So the slow you drive the better.

2nd looking at your example, you have drawn a nice circle around an ap. So do you want the program to calculate how far the signal should reach, based say on the gain of your antenna? Or do you want real bounderies? If you want real ones how do you figure to get them? At each Ap will you stop, and say walk away from the AP till you lose the signal, in say 4 directions, or 16 or 32?

Or are you only interested in how far along street X you can pick up the AP? If this is the case you may want to look at TThomas program. He had an option where every AP location was printed, not just the strongest. This would result in a smear of AP being printed for each AP on a street. By looking at the smear you can get the point of where the AP first came in range, and where you lost it. In my city, the AP are dense enought that in most cases, this is useless do to the overlapping. From looking at your maps I think this would also be true for many of your AP.

As you pointed out, it would be possible to mount directional antennas on your car and pick up APs a block, or two away. With the right kind of antenna you could get them miles away. But for stumbling what would be the point? The longer the range the greater the error is in your location. And doesn't the same hold true for your signal strength? If you're using the built in antenna, the AP will have one kind of coverage area, if your using the 5 dbi external antenna you'll get a different range, if you use the pringle antenna another range. If you happen to have a satilite antenna still another range. As far as I know, the location is the last one received, at the strongest signal point.

I think most of us have decided it's best to have the strongest point recorded, if you want to come back to it later this is where you'll most likely find it again. Lots of things can effect the range, is it raining? How humid is it? Did you put the antenna in the exact same place? Did the AP get moved, in the house? Is another AP working now in the area? The list of things that can change is endless.

Grey

mentat · 09-09-2002

I still recommend the latest version of JiGLE from the WiGLE project or RadioMobile to plot ranges of access points. (Those points were not responded to in your post. Did you follow any of the links?)

Chris_Schear · 09-09-2002

All of them. Wigle was "unavailable" until just recently.

Grey Wolf · 09-09-2002

Quote:

Originally posted by mentat
I still recommend the latest version of JiGLE from the WiGLE project or RadioMobile to plot ranges of access points. (Those points were not responded to in your post. Did you follow any of the links?)

I've tried, and tried, and I can't get JiGLE to run for me. I get a black (comand prompt type screen) that pops up and is gone to fast to read. I've got Java installed, any thoughts as to what I might be doing wrong?

Grey

Thorn · 09-09-2002

Quote:

Originally posted by Grey Wolf

I've tried, and tried, and I can't get JiGLE to run for me. I get a black (comand prompt type screen) that pops up and is gone to fast to read. I've got Java installed, any thoughts as to what I might be doing wrong?

Grey

Grey,
It's been some months since I've run JiGLE, but I ran into the same problem. To solve it, just open a command window and run it from there.

09-08-2002	#1 (permalink)
Chris_Schear Packetmonkey Join Date: Aug 2002 Location: WDM, Iowa Posts: 243	Accuracy of GPS information while stumbling You'll have to forgive this fairly uninformed set of questions concerning GPS information while stumbling, I've been scratching my head attempting to figure out what would be the logical conclusion. I figured I would just ask... Alright, so you have your prefered stumbling setup. Maybe you have a home-brew antenna, maybe you don't. Regardless, you have purchased a handheld GPS from BestBuy and want to get the Long/Lat of access points in your area. When NetStumbler detects an access point, does the CURRENT Long/Lat at the time of detection get reported as the "location" of the AP? If so, this would not be accurate. Especially if you have some addon antenna action going on, the access point may actually be several blocks away (or greater). The Long/Lat dumped into the data is relative to where YOU were when you detected the signal, not where the ACCESS POINT is. Additionally, for any given signal, does the Long/Lat update in the NetStumbler log if the signal gets greater? Let's say I'm driving down the road and I detect an AP. Boom - The GPS records my current Long/Lat. Now, if I drive two blocks further down the street and the signal is getting stronger, will NetStumbler dump new GPS information into the datafile as it would be a more accurate representation of where the access point is? How accurate is the GPS information while war driving? I would think the Long/Lat information cannot accurately pinpoint where an access point is unless you followed each and every signal until you reach a point where the signal doesn't get any stronger. This is assuming NetStumbler updates the Long/Lat recorded for any access point if the signal increases, perhaps someone can straighten me out. I can stand in the middle of a fountain near downtown and detect 5 access points with no antenna. If I had my GPS connected, the logs would indicate those 5 access points existed where I was physically standing, with regards to longtitude and latitude, would it not? I'd be interested in hearing discussions concerning the accuracy of data with GPS.

09-08-2002	#2 (permalink)
DaClyde Stumbling Luchadore Join Date: Aug 2002 Location: Alabama Posts: 73	NS seems to log multiple contacts... ...judging by the mapserver at http://mapserver.zhrodague.net/. If you zoom in on most APs on the map, you see the signal strength indicated by red or green dots of varying intensity depending on where the signal is picked up. I guess if you wanted the most accurate fix possible, you'd set NS to the highest scan speed and anytime you found an AP, drive around the area to get as many signal contact points as possible. WiGLEs' map server has added polygons to show the boundaries of the AP signals detected.

09-08-2002	#3 (permalink)
rogerRabbit Peripatetic Stumbler Join Date: Jun 2002 Location: Motueka Posts: 219	Re: Accuracy of GPS information while stumbling You are, of course, correct - the GPS gives the location of its antenna. Marius did post an answer as to how NS handles multiple readings; I'm pretty sure it is the highest SNR that drives the co-ords - but you could search to check. There's the prospect of using NS's scripting to read out a source's signal strength + your GPS co-ords and how they change over time. You could then take back-bearings from those to derive the source's lat/lon. And you don't mention the error inherent in the civilian GPS system. If you don't have a good number of satellite contacts your circular error of probability can be of the order of the range of some APs!

09-09-2002	#6 (permalink)
Thorn Did you do the math? Join Date: Apr 2002 Location: Villa Straylight Posts: 10,096	All of this has been posted before, many times. I would suggest use of the big button on top of the page marked "SEARCH" To give a quick summary: 1) The Lat/Lon are reduced at the output screen and the summary to those readings that give the strongest signal. So there is a sort of lazy man's triangulation going on. You can see this occur as you move. (Hopefully someone else is driving.) 2) This Lat/Lon updated continuously. The assumption is that the closer you are the higher the signal strength. This works well most of the time. The problem with this is that you may reach a point where signal is stronger due to a clearer line of sight, but where you are actually further away. As to the mapping/signal imaging, its been done: http://www.ittc.ku.edu/wlan/procedure.shtml __________________ Thorn "I'm The Doctor. I'm a Time Lord. I am from the planet Gallifrey in the constellation Kasterborous. I'm 903 years old and I am the man who is going to save your lives and all 6 billion people on the planet below... You got a problem with that?"

09-09-2002	#9 (permalink)
Chris_Schear Packetmonkey Join Date: Aug 2002 Location: WDM, Iowa Posts: 243	I never stated "Search didn't work", I said it was ineffectual. Additionally, I never stated anyone should "leave", I said they should click past the thread without depositing their two cents of retorts. I would kindly ask you not place words into my mouth. I'm perfectly cabable of inserting my own foot when necessary, I need not any assistance from yourself. Thanks. __________________ Regards... http://packetmonkey.net

09-08-2002	#5 (permalink)
Chris_Schear Packetmonkey Join Date: Aug 2002 Location: WDM, Iowa Posts: 243	Imagine, for any discovered access point and any relevant NetStumbling data contained about said access point, that Lat/Long information and Signal Strength are both cateloged. As you drive down the street, a signal slowly grows to a peek and then bleeds off. This, in essence, defines the area of coverage for that access point. Imagine NetStumbler collecting data and being able to export this type of information to a graphical map. For any "icon" of a discovered access point, you could view a graphical representation of WHERE to WHERE you saw that access point's signal. Did you receive a usable signal from "Access point #1" from 53rd street until University Ave? (example) Being able to log and display the coverage area, given signal strength and Lat/Long data - would be a very useful set of information. Even being able to drive down the road, hearing NetStumbler "ping" when it discovers an access point - logging your current Lat/Long information, and then logging the Lat/Long information of that last point when you saw that access point, would provide scope. It would answer the question, "Ok, so I know where an access point is, what are the boundaries of where this point was seen?" I know there is an access point named "ISTS_Wireless" seen on my normal drive home. With a GPS hooked up to your stumbler, I should be able to map out (graphically) a "cloud" of area where that access point was visible. Seeing icons on a map is one thing, having the option to click a "view coverage area" feature, given recorded Long/Lat daat and see a graphical representation of the signal, is another. The images below are a couple of examples of this. I've faded out the bulk of the image so the focal point is clear. The darkness of the green would signify the strength of the signal. Darker = stronger, obviously. *NOTE: These images are photoshopped to portray the intended feature. Of course, a more 'zoomed in' example would be more relevant.* EXAMPLE #1: ( Normal NetStumbling, without 'range features' ) ( Same data with optional 'range overlay' feature ) - - - - - - - - - - EXAMPLE #2: ( Normal NetStumbling, without 'range features' ) ( Same data with optional 'range overlay' feature )

09-09-2002	#7 (permalink)
Chris_Schear Packetmonkey Join Date: Aug 2002 Location: WDM, Iowa Posts: 243	I can assure you, searching for "gps AND accuracy OR triangulation" will result in to returns addressing the specific questions I am inquiring about. While you chose to focus on assumptions made concerning the nature in which the NetStumbler application updates Long/Lat data, such context was not a singular point of interest. Additionally, as perviously stated, "I'd be interested in hearing discussions concerning the accuracy of data with GPS.". What I am not interested in is entertaining the seudo-elitist, search gestapo mentality, which is so common place here. Perhaps, you devote hours of your day perusing every last inch of this message board, in all its individual forums, reading every last post umpteen times - but that may not be the case for each and every individual browsing. That potential fact, aside, vBulletin's implementation of regular search expressions has long since left much to be desired. I would suggest many things for you, none of which lie within scope of this discussion. To say mapping and signalling, "has been done", is a misnomer, to say the least. When you can point out a downloadable addon, which enables the end user to signal-ranged, Long/Lat data through some generally accepted "standard" of interfaces, say, Microsoft Streets...I would contend the solution"has been done". When you can import NetStumbler data into your selected mapping application and script a "range finder" addon, I would agree the horse is dead. Regardless, the context of this post was centered around the discussion of real-world application of GPS and its accuracy. Active discussions do not take place through messageboard search parameters. If you choose not to participate, that would be completely acceptable. Your "did you search" retorts are neither valid nor warranted. If you elect not to contribute, you may quite easily click on bye.

09-09-2002	#11 (permalink)
Grey Wolf Registered Member Join Date: Apr 2002 Location: Cincinnati, Ohio Posts: 345	Here is my take on this: 1st the gps signal is not updated continuously. But at intervals, if your lucky you can get an update every sec, if your not so lucky (you have an older/cheaper gps it may be longer). So the slow you drive the better. 2nd looking at your example, you have drawn a nice circle around an ap. So do you want the program to calculate how far the signal should reach, based say on the gain of your antenna? Or do you want real bounderies? If you want real ones how do you figure to get them? At each Ap will you stop, and say walk away from the AP till you lose the signal, in say 4 directions, or 16 or 32? Or are you only interested in how far along street X you can pick up the AP? If this is the case you may want to look at TThomas program. He had an option where every AP location was printed, not just the strongest. This would result in a smear of AP being printed for each AP on a street. By looking at the smear you can get the point of where the AP first came in range, and where you lost it. In my city, the AP are dense enought that in most cases, this is useless do to the overlapping. From looking at your maps I think this would also be true for many of your AP. As you pointed out, it would be possible to mount directional antennas on your car and pick up APs a block, or two away. With the right kind of antenna you could get them miles away. But for stumbling what would be the point? The longer the range the greater the error is in your location. And doesn't the same hold true for your signal strength? If you're using the built in antenna, the AP will have one kind of coverage area, if your using the 5 dbi external antenna you'll get a different range, if you use the pringle antenna another range. If you happen to have a satilite antenna still another range. As far as I know, the location is the last one received, at the strongest signal point. I think most of us have decided it's best to have the strongest point recorded, if you want to come back to it later this is where you'll most likely find it again. Lots of things can effect the range, is it raining? How humid is it? Did you put the antenna in the exact same place? Did the AP get moved, in the house? Is another AP working now in the area? The list of things that can change is endless. Grey __________________ ~the packets are out there~ waiting....

09-09-2002	#12 (permalink)
mentat Registered Member Join Date: May 2002 Location: CT Posts: 476	I still recommend the latest version of JiGLE from the WiGLE project or RadioMobile to plot ranges of access points. (Those points were not responded to in your post. Did you follow any of the links?)

09-09-2002	#13 (permalink)
Chris_Schear Packetmonkey Join Date: Aug 2002 Location: WDM, Iowa Posts: 243	All of them. Wigle was "unavailable" until just recently. __________________ Regards... http://packetmonkey.net

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