full-duplex 802.11?

[Bruce Janson]

    ...
    From: "Michael Scholten" <michael at affairs.net.au>
    ...
    -----Original Message-----
    From: wireless-admin at lists.samba.org
    [mailto:wireless-admin at lists.samba.org]On Behalf Of Brett Lymn
    Sent: Tuesday, 16 July 2002 12:21 PM
    ...
    On Tue, Jul 16, 2002 at 12:32:42PM +1000, Bruce Janson wrote:
    >     ...
    >     That was my understanding but I did not (and still do not) know
    > whether the current wireless devices allow selection of two channels
    > sufficiently far apart to avoid that overlap.  Moreover, given that the
    > pairs of colocated cards would be only "bus-slots" apart within the PC
    > there might be other shielding-related problems.
    ...
    I have talked to someone that has put two wireless lan cards into a
    laptop, from what he was saying he had no problems with the set up.  I
    was surprised he could do it physically but he told me he had a
    wavelan card and a cisco card (the cisco does not have the lump at the
    end) so they physically fitted into the pcmcia slots.
    
    Given this, I would not totally discount the idea.  Perhaps instead of
    going full duplex you could just trunk the interfaces together and let
    the trunk driver load balance the packets for you?
    -----End Original Message-----
    ...

    I've successfully put two Lucent cards into a desktop using two ISA->PCMCIA
    adapters and, using Linux drivers setting the cards to different channels in
    Adhock Demo Mode (rather than IBSS or whatever the newer Adhock mode is)
    without a problem.  The machine is configured as a relay point though, with
    both antenna pointing in different directions.  (I tested without external
    antenna first and it worked fine then too.)
    ...

Brett, Michael,
    Thanks.  The tests that you report are interesting.  In a more conclusive
test we would first send one unidirectional stream (say, UDP) across one
half of such a link and record its throughput.  Then we would send two
unidirectional streams in opposite directions simultaneously (one on each
"plex" of the link) and note the combined throughput.

    I tried using the IBSS mode and found that even though I attempted to
    specify a channel the card automatically switched over to the other cards
    frequency.  I also noticed this "feature" by setting a frequency in a
    notebook and bringing the machine close to a strong signal on a different
    frequency - it simply "switched over".

Perhaps this (here, inappropriate) feature can be disabled (via one
of Jean's wireless extensions?).

Common, there must be someone out there with a couple of PCs,
four cards, (maybe some cable and antennas) and a few days(?) to
spare just itching to resolve this burning question? :-)

Regards,
Bruce Janson, School of Information Technologies, Email:  bruce at cs.usyd.edu.au
F09 Madsen Building, Eastern Avenue               Phone:  +61-2-9351-3423/4
University of Sydney, N.S.W., 2006, AUSTRALIA     Fax:    +61-2-9351-3838