WLAN is RF and thus again voodoo. I only know I have similar puzzling results whenever I compare any 2 randomly chosen WLAN-Xceivers. 10cm of movement may change situation totally, and the sending out of beacons isn't very comparable between different APs. To make things worse, not only the beacons are sent a few times/sec, but also the DUT-xceiver has to scan 11~14channels. So one DUT may just hit a good beacon not spoiled by interference from another AP on same channel, while the other DUT might be unlucky to read any beacon of same AP, just 200mSec later, when it's about to scan same channel. Then it seems there are a multitude of proprietary filter algos in the WLAN-chipset FW, that have great influence on what's reported to and by the driver, and probably there are differences in the drivers themselves as well.
It's mere chaos and you won't get reproduceable results, you can just do statistical analysis over a huge number of samples. Reported dB and % values are usually not standardized between drivers, and I've seen a lot of bogus values being reported. Best test I found for WLAN performance is actual throughput resp Quality of Signal (though the later isn't usually reported in a reasonable way so you could compare different WLAN-chipsets) cheers jOERG
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ hardware mailing list [email protected] http://lists.openmoko.org/mailman/listinfo/hardware
