Measurable Parameters and tools

[psshergill]

We have 4 12xx APs managed by a 1206 WLC. On the WLC, the only thing that I see are WEP decrypt errors. We are a manufacturing company (steel) and there is certainly lot of electricity used in the mill during arcing process. There have been certain complaints of loss of wireless connectivity but when I test it (only taking my laptop and pinging continuously some other devices), the total packet loss is negligible. The reply times though vary between 1ms to 40ms.

I want to know couple of things:

1. What tools can I use to measure other parameters (like local signal strength, SNR) that can be give me graphical and real time values as I move in various parts of the wireless network.

2. What are acceptable values for these parameters.

3. If there is any additional advice you would like to give, it is always appreciated.

Thanks

[scottmac]

For free tools, Netstumbler is a good place to start. It will give you stats including channel traffic and interference.

Another free tool, Kismet has some other helpful options but it runs under *nix. Probably the easiest way to implement Kismet, if you son't already have a *nix laptop, would be Knoppix.

Knoppix is a fully self-contained Linux system on a CD. You can boot the CD on nearly any laptop and, unless you specifically tell it to, will not touch or harm any of the data that exists on the hard drive.

The Knoppix CD also has some other possibly helpful tools. I believe there is also a similar CD that is built to be wireless focused .... but I don't remember the name.

"Acceptable Values" is too relative to really nail down, especially in a (probably) very noisy area. In addition, some wireless NICs/chipsets are better at discriminating signal from noise. Cisco NICs, and the Atheros chipsets in general, tend to be very good, if not the best overall.

Again, even a good chipset, engineered poorly into a NIC, can give less performance than the same chipset in a better engineered NIC, or a NIC with a better antenna design.

IMO, as part of your survey, it might be worthwhile to try a Cisco NIC versus an Intel NIC/integrated chipset versus Orinoco, etc (budget permitting).

You might be chasing a problem that you think is the AP system, when the problem is just a poorly implemented integated NIC or antenna (or more accurately, one that is not well-suited for your specific environment).

Any other details you can offer might be helpful towards more specific suggestions.

Good Luck

Scott

[robert.wright]

5 points to scott. Backtrack is what i believe you were looking for.. (the project has gone through 3 or so names in the last year or two)

As scott was stating, each nic will have a seperate sensetivity which should be considerd along with driver/firmware versions.

[robfollett]

If we look back at when radio was invented, Marconi used a "spark gap transmitter" to get his telegraph data to travel through the air. Today, spark gap transmitters are illegal, because they produce radiation across nearly the entire radio frequency spectrum. There is a good possibility that this is the effect you are seeing from the arc welding.

If your plant is producing 15-ft diameter pipe or something similar in scale, it is probably not practical to install RF shielding around the arc welder.

You may want to use a spectrum analyzer to find the areas of your building that are least effected by the spark-gap interference, and locate your APs there.

The second factor you may be experiencing is RF reflectivity, assuming your plant is inside a steel building. The AP (and client) can be confused by receiving a signal directly, then receiving some form of the same signal as it is reflected off the walls of the building. Best practice here is to keep APs near the center of the building, away from the steel walls.