[CWN-Summit] Atheros sold to Qualcomm.

Paul Fuxjäger fuxjaeger at ftw.at
Mon Jan 10 14:09:31 CST 2011


Aaron Kaplan wrote on 08.01.11 17:34:

> GUess it is time for our own software defined RX/TX solution.


sorry for the last post, I mistakenly sent it to the list instead of a
direct message... my bad. In case you haven't guessed it already the German
comment read: software-radio nodes are probably not going to be as cheap as
those closed, fixed-silicon solutions in the beginning. But this will change
over the years as it has so many times before.



Anyway, I'd like to use this opportunity to say a few words on what has been
done in the last months in this context:

802.11a/g RX in GNURadio is 99% done, Giammarco Zacheo (colleague of mine
here at FTW) did a really amazing job and implemented very efficient
receiver code. He will hopefully be able to publicly release it in a few
weeks.

So now we basically have both TX and RX PHYs.
In combination with a HW-platform that is supported by GNURadio and is able
to sustain 20MS/s IQ-baseband data-rate (e.g. USRP2) this enables to build a
wireless transceiver that:

1) can talk/listen to all other legacy non-SDR Wi-Fi 11a/g radios
2) is completely open in terms of experimentation/modifications/improvements
on layer 1+2 (community driven innovation!)
3) is not bound to specific parts of wireless spectrum (can be used with all
kinds of up/downconverter daughterboards)

There are a couple of BUTS involved:
1) The price for the HW is still very high nowadays - 2000 USD per node
2) no 802.11b (we prioritized OFDM for the time being)
3) does not yet meet turnaround latency constraints to be FULLY compatible
to IEEE standards (the old problem with SDR - but there are workarounds for
this)

And finally:
Before we can really use this for arbitrary IP communication we need to glue
these PHYs to a basic MAC state machine and make the whole stack work. This
will take some time still... if you already have done something like this or
know someone who has:

Please get in touch with us and share your ideas. Main goal would be come up
with a way that reuses as much as possible from the FOSS world, is easy to
use and modular. We think of stuff like: click modular router, TUN/TAP
methods etc.


paul



More information about the CWN-Summit mailing list