On Mon, 26 Mar 2007 20:41:32 +0000 (UTC), dold@05.usenet.us.com wrote:

>Jeff Liebermann <jeffl@comix.santa-cruz.ca.us> wrote:

>> 2. I lock the speed to 54Mbits/sec and take a walk while downloading
>> some giant file and measuring either thruput or PER (packet error
>> rate). I draw a curve of the thruput versus distance at 54Mbits/sec.
>> The drop off should be quite abrupt. I repeat the test for other
>> wireless speeds.

>Is the locked-to-54 a go/no-go?

Yep. The transition between working well, and useless is quite sharp.
Try it and see for thyself. However, I wouldn't leave it that way.
The error rate will climb very rapidly if the speed is fixed. The
default mode is to have the speed decrease with increasing error rate.
That gives more range at the expense of speed.

>I'm not following the need for a locked
>speed, as opposed to watching the "current bandwidth" change as an
>indicator. I agree that some data needs to be moving for the reading to be
>of value.

It's not really a requirement. However, locking the speed makes the
transition between sufficient SNR and insuficient (lots of errors)
rather abrupt. It's a really sharp indication of the maximum range at
a given speed.

If I lock the speed, but monitor the "current bandwidth", what happens
is quite different. The speed (thruput) might go from the normal
25Mbits/sec to perhaps 12Mbits/sec (about half). That means that
every packet is getting retransmitted at least once. It also means
that only half the packets transmitted are getting through.

However, if I don't lock the wireless speed, the transition is much
slower and difficult to quantify. I've had several system that worked
just fine locked at 12Mbits/sec. However, I couldn't leave it like
that because there are a substantial number of users with 802.11b only
hardware. Anyways, these speed locked systems were fun to test. I
would walk away with a laptop running streaming video across the
parking lot. At some point, the thruput would abruptly drop to zilch.
The difference between working and dead was about 3-6ft at a range of
about 30ft.

This might help:
<http://www.sss-mag.com/ebn0.html>
Fig 1 is the chart of Eb/No (signal to noise ratio) versus the BER
(bit error rate) for various modulation schemes. I have other such
charts of family of curves for other modulation schemes and speeds.
Most are much sharper than this chart. What happens is that above
about 1E-05 BER (1 error in 10^5 packets), communications and thruput
are quite good. 10dB below this value, and the radios start
retransmitting packets due to errors. Note that a small change in
Eb/No (SNR) results in a huge change in BER (due to demodulator
threshold effects). That's what causes the abrupt transition.

Incidentally, I have some much better charts and curves at home and
can post them when I get around to fixing my broken bed scanner.




--
# Jeff Liebermann 150 Felker St #D Santa Cruz CA 95060
# 831-336-2558 jeffl@comix.santa-cruz.ca.us
# http://802.11junk.com jeffl@cruzio.com
# http://www.LearnByDestroying.com AE6KS