[Commotion-dev] Throughput advantage to 802.11s support
Ben West
ben at gowasabi.net
Tue Feb 5 21:38:15 UTC 2013
Hi All,
I'm happy to read that the Hackday just past went well! The roadmap wiki
page<https://code.commotionwireless.net/projects/commotion/wiki/Hackday-Roadmap-Notes>looks
pretty dense, although unfortunately my suggestion would be to
increase its density by adding *802.11s support*. My recent work to
extract Robin/Cloudtrax dashboard stuff into a standalone package was part
of a larger effort to get a modern replacement for a Robin nodes. And,
this is what is now prompting my recommendation for 802.11s support.
Since I have the expectation of remotely re-flashing all the Robin nodes in
the field sometime soon, I did throughput tests for old vs new to make sure
I wouldn't be deploying a regression. And, looks like 802.11g adhoc
meshing, even under Attitude Adjustment with the latest atheros drivers, is
*15% to 20% slower* than the ahdemo meshing done by Kamikaze+madwifi based
Robin firmware! This performance penalty appears to come from the fact
adhoc mode require beacons, while ah-demo does not, hence it not being
supported by the 802.11 standard. Furthermore, the mac80211 /
compat-wireless radio drivers currently packaged into OpenWRT appear to
enforce a maximum 1000ms interval for such beacons.
For 5.8GHz adhoc, a potential work-around is to use short slot-times as
implemented in this patch:
https://patchwork.kernel.org/patch/1835531/
2.8GHz adhoc, however, must still use long slot times to remain
standards-compliant. The work-around is that 802.11s meshing can use short
slot times, which this patch implements:
https://patchwork.kernel.org/patch/2050741/
While I do admit that doing speed tests on near-obsolete platforms like
ath5k would normally yield diminishing returns, these slot-time
enhancements should apply just as well to ath9k, too.
--
Ben West
http://gowasabi.net
ben at gowasabi.net
314-246-9434
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.chambana.net/pipermail/commotion-dev/attachments/20130205/ca899c8c/attachment.html>
More information about the Commotion-dev
mailing list