[CUWiN-Dev] 802.11 beacon strength oscillation

[David Young]

On Thu, Jul 21, 2005 at 12:02:03PM -0500, Bill Comisky wrote:
> 
> I've been trying to do some crude QA on our nodes before they go out the 
> door; a sanity check to make sure everythings working (pigtail is 
> connected, etc).  So I've been plugging them in at one end of our lab 
> space, and looking the received beacon strength at another node at the 
> other end of the lab with:  tcpdump -ne -y ieee802_11_radio
> 
> Somes nodes show a steady ~65dB signal level, while a sizable percentage 
> have beacon levels that alternate between ~45dB and ~65dB, maybe every 
> 20-30 packets.  The behavior seems tied to the radio, switching the radio 
> could change the behavior.  I haven't started bandwidth testing to see if 
> there is any difference, but wanted to see if anyone knew if perhaps this 
> was normal (an 802.11 "thing") or could offer an explanation..

There could be a few things going on.  First, beacons are sometimes sent
on alternating antenna ports.  Sometimes that is under the driver's
control (but it is supposed to alternates every 4 beacons, then).
Sometimes the driver lets the hardware choose the antenna---I wonder
which antenna the hardware will choose?

Another possible explanation is that the RSSI is mis-measured.
There's usually a LNA in the receive section that is turned on or
off as part of the first stage of receiver auto-gain control (AGC,
the process where the modem "hunts" for the gain level that saturates
the analog-to-digital converter).  For a lot of wireless cards, it is
not the RF signal strength, but the Rx AGC outcome, that is reported
as RSSI.  If the LNA on/off setting is not accounted for, maybe that
accounts for a 20dB difference.  (I am speaking from my understanding
of how Intersil/ADMtek/RFMD chips work, maybe what I say does not apply
to Atheros.)

Dave

-- 
David Young             OJC Technologies
dyoung at ojctech.com      Urbana, IL * (217) 278-3933