Feature Request: Zero Strength-Logging

Want to see something on NetStumbler.com/.org? Post your idea here! (Note - Posting does not guarantee that your idea will be implemented)

Feature Request: Zero Strength-Logging

Postby clentz » Thu Mar 06, 2003 4:27 am

I'm using Netstumbler to collect data for wireless mapping, and I am trying to map the range (extent) of coverage of individual APs on a large network. I know it would be a ridiculous amout of output, but It would make my day if there was an option by which Netstumbler could record zero signal strength values for APs it has seen already, but isn't currently seeing at a given position.

This type of information cannot be easily created through post-processing because of GPS wander: Though it might seem like I could insert zero SNR values at every location where I don't see a specific AP, It may have been recorded in the text file at a location that is only a few hundreths of degrees different from the location in question... The reason this is important is I'm looking for "holes" in wireless coverage, so I'd be looking for zero/low values in the middle of a field of high ones.

Also... if Netstumbler could output Lat/Longs where it sees no APs at all, it would make it a lot easier for me to keep track of where I've been stumbling. (and I could Map those areas as having no wireless networks).

Perhaps I should look into scripting something on linux, but I'm not looking to reinvent the wheel. And I really don't want to run linux on my laptop (I hear it ruins batteries and overheats CPUs, but that's a different issue)

Thanks for reading the novel...

Chris
clentz
Mini Stumbler
 
Posts: 8
Joined: Fri Feb 21, 2003 1:02 am
Location: Hanover, NH

Re: Feature Request: Zero Strength-Logging

Postby agentgrn » Thu Mar 06, 2003 6:07 am

First, don't hold your breath on feature requests. Marius has three child processes running around which keep his load average hovering around 1.0, I figure.

Comments inline...

Originally posted by clentz
I'm using Netstumbler to collect data for wireless mapping, and I am trying to map the range (extent) of coverage of individual APs on a large network. I know it would be a ridiculous amout of output, but It would make my day if there was an option by which Netstumbler could record zero signal strength values for APs it has seen already, but isn't currently seeing at a given position.
If an AP isn't seen at all, then nothing's recorded...so if you're mapping points based on a full text dump, then any space that's empty has no signal. Depending on method you're using to interpret the analysis, the zero-signal space can be filled in. More below.

For the rest of us wardrivers, that could potentially amount to a very, very large NS1 file fill of almost entirely empty data. It kinda makes the feature impractical.

Originally posted by clentz
This type of information cannot be easily created through post-processing because of GPS wander: Though it might seem like I could insert zero SNR values at every location where I don't see a specific AP, It may have been recorded in the text file at a location that is only a few hundreths of degrees different from the location in question... The reason this is important is I'm looking for "holes" in wireless coverage, so I'd be looking for zero/low values in the middle of a field of high ones.
This is something referred to as GPS wiggle, and if you have a decent receiver, you're accurate to within 10 feet. If 10 feet is enough to make an impact in performance, then your signal is too weak to begin with and it might be a good time for some yagis.

Originally posted by clentz
Also... if Netstumbler could output Lat/Longs where it sees no APs at all, it would make it a lot easier for me to keep track of where I've been stumbling. (and I could Map those areas as having no wireless networks).
There are MUCH more efficient ways to do this. I use Street Atlas 9.0 to track everywhere I go, then put an outline on the map when I'm done...it makes for a pretty nice reference. I also use GPSPlus and MP2k2 to track the current wardrive I'm on since SA9.0 doesn't do that very well once the cute arrow fade away.

Originally posted by clentz
Perhaps I should look into scripting something on linux, but I'm not looking to reinvent the wheel. And I really don't want to run linux on my laptop (I hear it ruins batteries and overheats CPUs, but that's a different issue)
I'm not sure running Linux is going to help you...you might want to try Cygwin for all your shell needs, though. And Linux doesn't ruin batteries per se...it's just not known for its efficient power management.

To do what you're looking to do, you'd have to cook up a script that compares the .ns1 file with a GPS log file and fills in the data space appropriately...but then you still need to get the data separate when you import it into your mapping program, which could be a bit more of a challenge.

Depending on what you're making your maps in, I'd go with MP2k2 and GPSPlus for your navigation trails. Use Cygwin or Linux to parse the AP you're looking to map out of a text dump. Import the data points on top of your nav trail.

Sorry to come across as harsh at parts...dealt with a super noob earlier.
-A.G.-
agentgrn
 
Posts: 517
Joined: Sun May 26, 2002 6:44 am
Location: Worcester, MA

Re: Re: Feature Request: Zero Strength-Logging

Postby clentz » Fri Mar 07, 2003 2:34 am

Originally posted by agentgrn
If an AP isn't seen at all, then nothing's recorded...so if you're mapping points based on a full text dump, then any space that's empty has no signal. Depending on method you're using to interpret the analysis, the zero-signal space can be filled in. More below.


No... this isn't true in my case. I'm not interested in mapping general locations - I'm mapping exact field coverage of specific APs. "Every space that's empty" is potentially space that I haven't taken readings at / space where NetStumbler didn't record data. As I stated in my initial post, I can't assume a zero value for an AP even if I do have a location and signal strength for other APs. Here's an example from the text dumps:

N 43.7000117 W 72.2803583 ( Greenwave Wireless ) BBS ( 00:40:96:45:71:ed ) 19:53:11 (GMT) [ 26 81 55 ] # ( ) 0021 0400 0
N 43.7000067 W 72.2803567 ( Greenwave Wireless ) BBS ( 00:40:96:45:83:00 ) 19:53:12 (GMT) [ 10 63 53 ] # ( ) 0021 0400 0

Note that there is no entry for the second AP at the first coordinates, and vice versa. However, the coordinates are so close, that it is unreasonable to assume that neither can be seen at the the other's recorded location. There are several means of post-processing to correct for this where appropriate (populate zero values in realistic locations), but it would be nice if NetStumbler could do this itself.

the rest of us wardrivers, that could potentially amount to a very, very large NS1 file fill of almost entirely empty data. It kinda makes the feature impractical.


I thought I acknowledged this in my post, and that's why I recommended it be an option that you turn on. For mapping in heavily-wirelessed urban areas, the file size increase would not be significant anyway.

This is something referred to as GPS wiggle, and if you have a decent receiver, you're accurate to within 10 feet. If 10 feet is enough to make an impact in performance, then your signal is too weak to begin with and it might be a good time for some yagis.


Thanks for the tip, but I'm mapping for research purposes, and I'm interested in finding dead spots in coverage.

There are MUCH more efficient ways to do this. I use Street Atlas 9.0 to track everywhere I go, then put an outline on the map when I'm done...it makes for a pretty nice reference. I also use GPSPlus and MP2k2 to track the current wardrive I'm on since SA9.0 doesn't do that very well once the cute arrow fade away.


That's kind of what I'm forced to do now - use the raw GPS track data and merge it with the NetStumbler information. This merging is time consuming - because of GPS wiggle, I can only do it after I have rasterized my vector data. Hence the feature request.
clentz
Mini Stumbler
 
Posts: 8
Joined: Fri Feb 21, 2003 1:02 am
Location: Hanover, NH

Postby Thorn » Thu Mar 13, 2003 5:58 am

If you're already not familar with it, you might want to look at University of Kanas' Wireless Network Visualization Project. It's similar to some of what you seem to be looking to do, although I'm not sure it's accurate enough for you to map dead spots.
Thorn
Stop the TSA now! Boycott the airlines.
Thorn
 
Posts: 10340
Joined: Sat Apr 13, 2002 3:00 am
Location: Villa Straylight

Postby clentz » Thu Mar 13, 2003 6:56 am

Thanks, I looked at their pages a while ago. They don't really post their techniques, and they didn't respond to my email... Plus, I think they're just taking netstumbler data and rasterizing using Inverse Distance Weighted equations - Without populating zero values, this could potentially put false signal values where none existed. (ie, positions measured with zero signal aren't recorded, and therefore aren't part of the data set, so values might be interpolated over zero-strength positions).
clentz
Mini Stumbler
 
Posts: 8
Joined: Fri Feb 21, 2003 1:02 am
Location: Hanover, NH

Postby sparafina » Thu Mar 13, 2003 7:37 am

I'm not sure what y'all exactly mean by "GPS wiggle." I have seen that term used in conjunction with Selective Availability, which has be turned off.

If you are referring to general error, then that might be the wiggle you are referring to. GPS does not give you an absolute position, position is calculated from a sphere of possible positions. The longer you sit on that position, the more data you get, the better the calculated position.

The same would hold true for finding coverage areas. If you want to precisely map areas, you would need a GPS receiver capable of doing sub-meter positioning fixes either on the fly or through post-processing. Another way would be to overlay a grid of known points/location and sit on each point for say at least 15 minutes. Either way will need to sit on a point, driving by will not suffice for the level of precision you seem to want.
User avatar
sparafina
Mini Stumbler
 
Posts: 1428
Joined: Tue May 07, 2002 2:12 pm
Location: TX,MD,NY


Return to Suggestions/Requests

Who is online

Users browsing this forum: No registered users and 2 guests