[olsr-dev] Fwd: Frottle and Meshes

[Tim Schmidt]

That looks interesting, and possibly more efficient than my scheme
(although without an accurate simulation of each I can't tell). 
However, it's also drastically more complex.  My scheme would require
only the use of the ETX table already maintained by the OLSR ETX
plugin, a timeout timer, and some code to perform handshaking and
token passing...  Important features of my scheme are:

- no node is a 'master'
- no node needs topology knowledge beyond who it can hear / who can
hear it (already provided by ETX plugin)
- traffic isn't corralled into 'rings', nodes dynamically determine
if, when, and who it's OK to talk to based on the simple, predictable
behavior of all other nodes it can hear.

I'm not saying my scheme is perfect...  Hell, I don't even know if
it'll work.  Even if it does it's a solution at the wrong layer.  But
I think it would be easy for a talented person to code up quickly, and
I'd like to see it tested.
--tim

On 3/2/06, Jim Thompson <(spam-protected)> wrote:
> You might investigate the "wireless token ring protocol"
> http://www.eecs.berkeley.edu/~ergen/docs/WTRP_SCI02.pdf
> http://www.eecs.berkeley.edu/~ergen/docs/main.pdf
> http://www.eecs.berkeley.edu/~ergen/WTRP/
>
> I've considered coding this as a replacement for net80211 (the 802.11
> MAC portion of the 'madwifi' driver), rather than running over the top
> of the 802.11 protocol.
>
> OLSR could still run on top of WTRP, tying together multiple rings.
>
> Jim