[Click] Using click to build an 802.11a MAC layer on USRP2
Ruben Merz
ruben at net.t-labs.tu-berlin.de
Wed Jan 27 08:47:47 EST 2010
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
This nice NSDI 09 paper has an interesting related discussion
http://www.usenix.org/events/nsdi09/tech/full_papers/nychis/nychis.pdf
Still, I'm afraid you can't avoid putting some processing on the FPGA.
Ruben
On 1/27/10 12:35 , Harald Schioeberg wrote:
>
>
> Juan Ramon Gutierrez Agullo wrote:
>> Hello everyone
>
>> This is my first message here and, of course, I'm a newbie click user.
>
>> I have an 802.11a transmitter
>> (https://www.cgran.org/wiki/ftw80211ofdmtx) running in the USRP2, but it
>> only generates a standard-compliant IEEE802.11a/g/p OFDM frame - PHY
>> layer.
>> Looking at other projects I found the hydra project, and they used click
>> to implement the MAC layer (CSMA/CA and DCF). I have some questions and
>> I hope you could help me.
>
> USRP uses either usb (USRP1) or ethernet(USRP2) to communicate with the
> host, which probably means delays in the magnitude of 10s of us for a
> frame-transmission over the interface (Take Linux' interrupt latency for
> ethernet-frame-reception into account!)
>
> 802.11 slot time is 4 us, which is the time you have available to
>
> a) receive the last sample at the frontend
> b) decode it at the CPU
> c) deceide whether to ack it
> d) encode the ack at the CPU
> e) have the first samples available for transmission at frontend
>
> Click runs on the host as a scheduled kernel process, which brings you
> to probably 100s of us worst case reaction on incoming samples,
> especially on a Linux machine busy decoding samples.
>
> So I don't see any way to implement 802.11 MAC on USRP on the host side,
> no matter whether you use click or any other host side tool, IMHO: the
> FPGA is your only choice (maybe some fancy real-time kernels might just
> make it...)
>
> Please, please, prove me wrong on that, 802.11 on USRPs would be soooooo
> cool.
>
>> - How can click communicate with GNU Radio?
>
> My approach would be:
> a) implement the MAC on the FPGA,
> b) implement TX-queues on the FPGA that use this MAC,
> c) write a driver that accepts frames and meta-data for this MAC,
> d) write a click element, that takes Ethernet packets and transforms it
> into one of these meta-annotated-frames.
> e) implement things like rate control or scheduling algorithms as click
> elements and have them set proper packet meta-data.
>
>> - Is there any MAC implementation -by default- with click installation?
>
> Nope, as all Wifi cards implement the MAC in hardware, for above
> mentioned reasons.
>
> Now, can sombody please tell this poor guy that I'm a moron and there is
> a smooth and simple way to do all this ....
>
> Harald
_______________________________________________
click mailing list
click at amsterdam.lcs.mit.edu
https://amsterdam.lcs.mit.edu/mailman/listinfo/click
- --
Ruben Merz Deutsche Telekom Laboratories
http://www.net.t-labs.tu-berlin.de/people/ruben_merz.shtml
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
iQEcBAEBAgAGBQJLYEQDAAoJEIem9YkUyRjt+YEH+wRP2+mDh+Ls9bbegWksj+lC
sI2Slql/GYE8rHQiIPu9/XhyrkSfDNsJdpiN09vcjsEI3T8U6Ee3X1eTeUeSoNOd
jVQxXTDuxMYIkuR2sx8vK1FUzPJJkc79TfLdXJKi4o3Fyv197d+t4zEd1a4TxbhV
7zFzUyELhbo7sk+BWfVm1oj3jVQApO28aJwiaWKr5uCeOIZjxZqR8RDcpYJCw2V1
/A06s6geBb3a74B6gK192D7tLSERON8FPrfSY8DRpR66PRHeVMVJneAY3RU9I3cr
rTpPxGw2BZKJNKeicHv4r7L3HY2heSESNzxt+uM3OZSfBzQLiPw26rKxdWQK8Ak=
=IeD4
-----END PGP SIGNATURE-----
More information about the click
mailing list