[Commotion-dev] Building a Wireless Network for a High Density of Users

[Ben West]

The suggestion to use a prime number for the beacon interval to avoid
clients from becoming synchronized is pretty clever.  Seems like that would
work as desired, although I wonder if the speaker actually had a chance to
test that.  I think some of the ath5k and ath9k chipsets, as their
compat-wireless drivers are implemented, enforce a maximum 1000ms beacon
interval, so *997* will have to be your friend.

I'm also surprised he was successfully running APs at just 4mW transmit
power for roughly room-wide coverage.  He did mention his preference for
high-gain antennas over increased power, implying that low power could have
been in conjunction with large (e.g. ~10db) dipoles.

There is also this PDF from Bastian Bittorf / Freifunk that was published 2
years ago.  Similar advice, including tips specifically gears towards adhoc
mesh operation.
http://wiki.freifunk.net/images/7/73/Vortrag.pdf

Bastian's presentation also encourages adoption of zram for more efficient
memory footprint on RAM-limited devices.  This is now in OpenWRT trunk, and
I'm curious if such might permit operation of minimal Commotion-OpenWRT
(but with the luci-splash package) on 16MB devices, like those hundreds of
FONeras Sascha still has sitting around. ;)
https://dev.openwrt.org/changeset/35025/

On Tuesday, February 12, 2013, wrote:

> David Lang has some practical experience setting up WiFi access for up
> to 2000 simultaneous users:
>
> https://www.usenix.org/conference/lisa12/building-wireless-network-high-density-users
>
>
> _______________________________________________
> Commotion-dev mailing list
> Commotion-dev at lists.chambana.net
> https://lists.chambana.net/mailman/listinfo/commotion-dev
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.chambana.net/pipermail/commotion-dev/attachments/20130212/f2dc16b9/attachment.html>