[Battlemesh] Battlemesh v5 tests

Clauz clauz at ninux.org
Sun Mar 11 12:48:08 CET 2012


On 03/10/2012 01:57 PM, Martin Hundebøll wrote:
> Hi Clauz,
> 
> I would like to introduce another test for throughput. I work with the
> network coding[1], and would like to see some tests that show some
> scenarios where network coding is feasible and some where it is not.
> 
> For example, the chain topology is a nice scenario:
> 
>     A -- B -- C -- D -- E
> 
> because network coding can increase throughput at each hop, if data goes
> in both direction. However, when data goes in one direction only, our
> implementation delays packets at each hop.
> 
> Also, two different paths on the same channel:
> 
>     A -- B -- C
> 
>     D -- E -- F
> 
> because network coding can reduce the air traffic and thus increase the
> total throughput (again, when data flows in both directions.
> 
> What do you think?

Network coding is cool, but has not much to do with routing protocols.
Moreover we would need a new and different firmware (btw did you
implement this in some specific driver?) to be flashed into some of the
devices, increasing too much the level of battlemesh enthropy :)

Clauz



More information about the Battlemesh mailing list