[OLSR-users] willingness vs. LQ routing

[Thomas Lopatic]

[...]

> I would suggest the following: instead of a "expected transmission 
> COUNT" metric one could use a "expected transmission TIME" metric.

[...]

Hmmm. Interesting. This should actually be pretty easy to implement, as 
it still remains an additive metric like ETX. Moreover, your idea A is 
pretty close to the per-interface "Multiplier" value from my last e-mail 
to this list. Your design would make our neighbour detect and announce a 
higher packet loss, whereas the "Multiplier" design would make ourselves 
announce a higher packet loss. The results, however, should be equivalent.

So, shall we add the "Multiplier" per-interface configuration parameter 
as a first step for 0.4.9?

I think along the lines of

LinkQualityMult 2.0 192.168.19.72

and

LinkQualityMult 2.0 *

where the first form affects only the packet loss that we report for the 
link between ourselves and 192.168.19.72 via this interface and the 
second form affects the packet loss that we report for any link via this 
interface.

What do you think?

-Thomas