[Prism54-devel] Prism-USB in ad-hoc and master mode

Jean-Baptiste Note jean-baptiste.note at wanadoo.fr
Sat May 14 16:13:39 UTC 2005


Hi again Jan,

> Mmh, to understand to concept behind: Does the 802.11 stack at some
> point decide to switch a link to a different speed (like you switch a
> RS232 line), or is this performed individually for every transmitted
> packet. May we even have asymmetrical links (with respect to bitrate)
> between two stations? 

I answered your question badly, because i didn't read your mail
carefully; sorry about that, and for sounding a little bad.

We had a discussion about this some time ago on the mailing
list. Physically, the transmition on the wireless medium occurs as thus:
preamble @ 1Mbps, which contains the speed of the following data, which
is then transmitted at the given speed. So physically, the data rate is
per-packet, and can be fully asymmetrical without any problems. The only
condition is that you use a rate that the other can understand. This is
ensured by higher levels of the protocol.

The 802.11 stack knows from the AP (or negociates with its peer in case
of IBSS == ad-hoc) which rates it has the right to use to communicate
with the AP and its neighbours, and which rates it /must/ support in
order to join the BSS. More on this in the thread started by Denis :

http://prism54.org/pipermail/prism54-devel/2005-March/002682.html

and the following reply about ratesets where there are links to the
relevant portions of the 802.11 specification :

http://prism54.org/pipermail/prism54-devel/2005-March/002713.html

To sum it up for what concerns us : the 802.11 stack chooses as it
pleases from the set of available rate the one it prefers. This is the
job various rate control algo available in madwifi; even when high rates
are attainable, you don't want to use them if they make you drop many
packets.

So to reiterate, the idea for reverse-engeneering the rate setting of
the prism softmac device is to use the 802.11 specification to force the
device to restrict it's rateset, so that we can observe them one by one.

Cheers,
JB

-- 
Jean-Baptiste Note
+33 (0)6 83 03 42 38
jean-baptiste.note at wanadoo.fr


More information about the Prism54-devel mailing list