On Tue, Mar 02, 2004 at 01:00:47PM +0100, Robert Lillack wrote:
> | Maximum data length:                    [ 2304 ]
> | RTS/CTS handshake threshold:            [ 2347 ]
> 
> How does 'collision detection/avoidance' work then? Will
> my station just send stuff (and after each packet listen
> for a while) until it get's an ACK?

Right, it listens for an ACK.

> 
> > Most of the notorious 802.11 overhead comes from the
> > 144-bit preamble and 48-bit PLCP header, which are sent at 1Mb/s (unless
> > "short preamble" is enabled).
> 
> I knew there was more.... ;)
> So, how do I enable short preambles on NetBSD? I've only
> seen this on Windows boxes so far.

You don't yet enable short preambles on NetBSD.

> And what does this actually mean?  Is the preamble sent at another
> speed or is it only shorter?

Both.

> 
> [...]
> 
> And while we're at it: What is micro wave oven robustness?
> a joke?

In the United States, microwave oven interference (MOI) is a 60Hz train
of pulses, with a 50/50 duty cycle, driven by the household current.
The frequency is different in parts of the world with 50Hz household
current.

802.11 can (sort of) coexist with strong MOI if the packets fit in the
1/120th second between pulses. Long packets sent slow will not fit between
the pulses, though. For example, a 1524-byte TCP segment sent at 1Mb/s
takes 1524 bytes * 8 bits/bytes * 1s/Mb = 12192 microseconds, so it's not
going to fit into the 8333-microsecond slot between pulses from the oven.

I believe that microwave oven robustness (MOR), as Intersil implements it
and as Lucent used to, keeps the rate adaptation algorithm from choosing
the 1Mb/s rate---maybe 2Mb/s, too.  I think that you can do a lot better
than that: just for instance, I think that Lucent firmware has taken to
dynamically changing the fragmentation threshold.

Dave

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