[Olsr-dev] Different time and interval configurations

Joe Gio (spam-protected)
Fri Jun 20 13:46:44 CEST 2008

We have tried using weighted windows where newer
hello samples have more significance towards the
total LQ than older samples.


Mitar wrote:
> Hi!
> What are consequences of having nodes with different *interval (like
> HelloInterval) and *time (like HelloValidityTime) values in a mesh? It
> is not maybe even good that they would be a little different and so
> make recovery even more "organic" and non-deterministic? Maybe having
> those values being a random value would also be interesting? Or some
> adaptation/learning algorithm where nodes would gradually come to the
> optimal values by themselves?
> (This thinking comes from not really knowing what values are "the
> best" and playing with different values and seeing what are the
> differences in a mesh behavior - and why would not nodes do this by
> themselves - what I am doing now is almost the same as random.)
> Mitar

More information about the Olsr-dev mailing list