Measuring link quality

Hello,


the 802.11 standard specifies that link quality should be (for DSSS modulation) calculated from the correlation value obtained when code lock is achieved between the local PN code and the incoming PN codes.

In practice however I am convinced that Wi-Fi cards' producers do not calculate it in this way but rather by the percentage of correctly recieved bits or as the difference between signal strength and noise level.

I am currently doing some research on the possibility of using a 802.11 signal in a Wi-Fi LAN for locational purposes and I would like to consider all the information available to estimate a terminal's position. That includes, signal strength, noise level and link quality. However I cannot use link quality if I am not sure of how it is calculated.

Does anyone know how I can find a Wi-Fi card whose vendors specify their method of determining the link quality? Or a card from which I can extract a link quality according to the definition of the standard?

The PN code is adjusted depending on how much corruption resistance is required, from what I can tell most vendors base their connect speed on the RSSI (Receive Signal Strength Indicator) which is hash of actual signal strength, missed beacons, re-transmits due to corruption etc. I have found the best way to check the quality of a link is to do a large FTP file transfer and move around with a laptop using a bandwidth metering program. The max throughput you will get on a 802.11g link is about 22Mbps, as you move away it often says it is connected a 54Mbps but the actual throughput drops with re-transmits and maybe larger PN code (if it adjust dynamically). There is no standard for how vendors measure their RSSI and some seem to want to hold on to that 54Mbps connect rate even when it should be lower for a more reliable link. Most likely for marketing purposes.
Is this any help to you?

One thing I have noticed is that many cards can still report excellent signal strength even if the RF side of things is not up to standard resulting in tons of dropped/resent data. A classic example is the wireless configuration service included with Windows XP. I assume this could be more likely due to bad interpretation by the software of what the card is actually trying to report?