Great fun. Scribble a map of your home or office with any paint program. Walk around the room while left clicking the mouse to mark your position on the map and record the signal strength. Right click to stop recording. Results look something like:
I ignored the instructions that claimed it only works with XP and Vista (32 bit) and installed it on a W2K desktop. No problems (so far).
It works using my built in AR5007 wireless card in Vista but would not work using USB dongles with ZD1211 and RT2870 chipsets. There was an added problem with neither of these being able to get an internet connection after installing the Ekahau s'ware, had the usual
169.254.xx.xx addresses. When I tried to remove the Ekahau driver from the adapter properties Vista locked up. I had to remove the Ekahau program to get the USB adapters to work properly, strangely there was no problem with the AR5007 card.
That was badly worded. the Ekahau driver was listed in the "Network Connection Properties" for the wireless connection as one of the items that were used by that connection, in my Vista laptop it was shown just above "Internet Protocol Version 6(TCP/IPv6)".
Did a good job of guessing where the APs are in my office, but I have to wonder how useful it really is. It's all very well knowing how well you can receive a signal from an AP, but surely what matters is what sort of latency/packet loss you see whilst connected to any given AP [as you don't really know how well the AP can hear you]. Granted it's a bit more work and would take longer, but it would be worth it in the end.
My guess(tm) would be that it added something to the LSP (layered service provider) winsock2 chain that wasn't removed. Try LSPFix:
or other repair utility. I think this one works with Vista, but I haven't tried it.
If deperate, you can reset the whole network mess:
If nothing works, do a system restore to a point before things went awry.
I'll try it on a Vista machine in a few days. Remind me if I forget.
I screen grabbed a local road map from Google maps and used HeatMapper to build a wi-fi map of the neighborhood. I went for a walk with my laptop, left clicking the mouse merrily down the beaten path. 22 access points were heard and somewhat located. My neighborhood is on a hillside full of reflections and in a forest full of attenuations. The signal strength color pattern therefore appeared rather weird. I'll try it again next weekend, hopefully with better results.
Doing the uninstall from the "network connection properties" did work for the AR5007. I had previously tried a system restore but that didn't work. I had tried the original version of this program but that didn't work at all with the AR5007 and I think you only got 15 minutes of time to record the signal levels. I had hoped that the USB devices would have worked so that I could see if there was any variation in signal pattern when using these.
Since I couldn't remember exactly what it was called I re-installed the Ekahau s'ware. In the "Wireless Network Connections Properties" it is listed as installing "Ekahau User Protocol Driver for NDIS6" and this is what has caused me problems, even on uninstalling the Ekahau program this driver is not removed but can be by using the uninstall function in "Wireless Network Connections Properties".
Well, not so wonderful. It seems to work ok for indoor mapping, but seem to get rather confused when faced with an outdoor environment. Results are far from usable. As I walked around the loop, the order of the access points seem to be in the correct order, but the found AP locations are useless.
I was thinking the problem might be the laptop, so I tried a total of
3 XP laptops with similar results. This plot was generated on an Asus eee PC 701.
I also created similarly useless results with multiple AP's at my house. If I walked in a circle, the results were similar to the photo. However, if I followed the instructions, and sorta walked in a semi-drunken manner, with lots of cross-over points, the plots looked more like in the examples.
Dunno. I would have to see the raw collected data to determine what's happening. Also, I would need some clue as to the algorithm used. In the outdoor case, my guess(tm) is that the huge variations in signal strength are rendering the data useless. There may also be a problem with the large number of SSID's collected. In the indoor case, it seems to be a problem with the algorithm used, or possibly because the room is small enough that I might get equal signal strength almost anywhere in the room. If it does something really crude, like average or peak detect the indicated signal strength of ALL the AP's heard, that might explain the almost uniform signal level indicated. It may also be that my wireless devices are returning garbage values. That's dubious as I tried 3 different laptops, but still possible. Lots of possibilities, but nothing definitive, yet.
Unfortunately, the software does not offer any means to filter the results, limit the test to a specific set of SSID's, or lock it on a specific channel. It's difficult to tell what's happening without filters. The full version called "Ekahau Site Survey" does these and much more.
The problem is that it costs $2,200 and $2,700 for the pro version. Ouch.
I'll play with it some more later and see if I can deduce how it works.
Jeff, I'm still curious about that Airsleuth product line which uses the Symphony HomeRF2 adapters in a mode that permits fine grain spectrum analysis and RSSI; there seems to be a site survey mode with graphics, although no map overlay. Perhaps you could convince the authors to permit a no-cost evaluation (I don't have the cachet for those perks).
I wondered if it recorded the peak value of each AP at each location but needed a high signal level, say -50dBm, for at least one AP to provide some form of reference. Looking at your survey I would say you probably did your first check outside your home/office and recorded a low signal. Have you tried doing your first check inside your office where there is a high signal level and then going outside for the next check?
I think it's dumber than that. I've been trying to mentally reverse engineer what they're doing to produce my bogus results. After tossing all the complexicated bad guesses, I have yet another guess(tm).
Methinks, that they're averaging *ALL* the signal levels at any given point. They measure the signal level from the assortment of hot spots heard, but only record the average value. They also might do some data smoothing with adjacent locations. In a perfect environment, with only a few AP's, that would produce a very usable result. It worked fairly well inside my palatial office, where I could only hear
4 AP's. However, when faced with my outdoor environment, and a myriad of AP's, the average value will be fairly constant. The result is a uniformly constant signal level. Mehthinks this is highly clever, but not particularly wonderful for all applications. It would require data filters to fix the problem.
It's my home and you're correct. The results I posted was test walk #4. Test #2 started inside my house, where I was close to two AP's. I decided that this might affect the results, so I started test #4 from the road level, so that everything was on equal footing. That made it worse resulting in the uniform signal level.
Yes. I got a rather hot (red color) spot starting at my house. However, it radiated to well over half the map, which isn't much of an improvement.
That's a surprise, I expected you to have had a nice green spot inside and then going to red as you moved further away outside. The small amount of playing I did showed green down to -75dBm, light green.
Oops. I never tried that. I just ran the program with and screen captured the results after it was done, without selecting a specific AP. I'll try another circular walk around the neighborhood and see what happens.
Yeah, but there's a problem with small screens like my Asus eeePC 701. I can't get to the "Screenshot" button unless I first open the left hand AP list window. Even if I'm quick, I usually end up doing mouse over on one of the other other icons that ring around the map. Hitting the AP icon and doing a subsequent screen capture is more than tricky. A keyboard shortcut for screenshot would be useful (along with an "are you sure" when accidentally hitting the right mouse button).
I managed to highlight and grab 3 nearby AP's. All the plots are useless:
(I also reduced the size of the PNG's from about 220KB to 50KB by converting the overkill 24 bit color to 8 bits). Plots of the other
24 odd SSID's are equally useless.
The "Testnet" SSID, being located in the middle of the map, was an accident. I prematurely hit the right mouse button about 1/5th of the way through the course. Amazingly, it's located in the correct location. Nothing else is even close. Some of the SSID's are located off the map and miles away (I live on the side of a valley).
I'm still mystified as to what's happening. I'm doing something to confuse the program. Perhaps it's having trouble dealing with 24 SSID's. I guess more trial and error will eventually deduce the cause.
Perhaps it's the size of the area you are trying to cover rather than the number of AP's. My limited testing showed that I needed to use a large number of hits in an AP's location and even then the display was not perfect. I did not use a map but used the grid display and my home AP as the start point. The resulting map for my home AP covered the whole of the grid but the coverage maps for the other AP's were restricted to within a few squares of the the route I had taken and where I had obtained readings and did not cover the whole grid. Did you notice that as you conduct the survey the line that is drawn is the colour of the signal strength, this is replaced by a solid green line on the completion of the survey.