On Fri, 11 Nov 2005 08:24:46 -0500, bjs555 <(E-Mail Removed)> wrote:
>On Thu, 10 Nov 2005 09:52:02 -0800, Jeff Liebermann
><(E-Mail Removed)> wrote:
>
>>Netstumbler will report the signal strength and S/N ratio of both ends
>>of the test link. The DWL-2100AP supports SNMP. You'll find the
>>signal strengths somewhere in the MIB tree. If you're going to use
>>such reporting tools, be sure to change the data rate from auto to
>>some fixed speed as the signal strength will vary with the wireless
>>data rate.
>That's a really good point about making the data rate a fixed speed.
>Suddenly my measurements make more sense 
>Thanks,
>Bruce
Yep. I was getting an awful lot of "circular" antenna patterns until
I discovered the obvious. However, you're not done yet. I noticed
that fixing the connection speed doesn't necessary disable some of the
other speed mutilating features. I testing a Dlink DI-624 and found
that if I set the speed to 12Mbits/sec OFDM, I should expect that
802.11b compatibility would also be disabled, since it should never
connect at 802.11b speeds. Nope. It continues to listen for 802.11b
clients that will never be able to connect. I also found that leaving
the super-G, turbo, or whatever modes enabled left it also sniffing
for those, which will never connect. Therefore, turn OFF all advanced
features before trusting the signal strength readings.
As for accuracy, the wireless cards all generate an RSSI (receive
signal strength indicator) reading on a scale of 0 to 254. This is
converted in either the firmware (for SNMP) or in software to -dBm
signal levels. The granularity of the -dBm output largely depends on
how much effort the programmists put into the conversion. In most
cases, it's just a lookup table with no interpolation. They may offer
good resolution for weak signals, where you need it, and use big jumps
at high signals, where nobody really cares. Some manufacturers
calibrate each table individually while others just use a best guess.
There are at least 5 other ways to generate the numbers including "bit
distance" calcs.
http://archives.neohapsis.com/archiv...1-q4/0305.html
http://archives.neohapsis.com/archiv...1-q4/0295.html
--
Jeff Liebermann
(E-Mail Removed)
150 Felker St #D
http://www.LearnByDestroying.com
Santa Cruz CA 95060
http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558