[Olsr-dev] Little change to olsr core

Teco Boot (spam-protected)
Thu Jul 26 17:24:30 CEST 2012

Responding somewhat late....

I'm concerned on manipulating ETX. Nodes exchange LQ / NLQ and calculate ETX. If a node has a configured ETX that differs from what can be distilled from packets seen over wire / ether, it could be harmful. Instead, you could use an LQ configuration, still in its current range (0,255). This results in symmetric ETX

MPR is already broken, but this should be repaired. I hope you don't bring up more obstacles.

Using LQ multipliers would have my preference. What is the shortcoming? Can it be improved?


Op 24 jul. 2012, om 10:09 heeft ZioPRoTo (Saverio Proto) het volgende geschreven:

>> So you are working on a fixed alternative to the LQ multiplier?
> Correct !
> Moreover if we have two nodes A and B, we would like to have in the
> OLSR database a cost for A to B and a cost for B to A that could be
> different.
> The cost of A to B is set on the configuration file of node A. The
> user should specify the directly the ETX.
> I would like to comment on this sentence of my student:
>> This will change only on the local router the cost of the specified link, without propagation of the fixed value
> He means that a node with a fixed ETX value in the configuration, will
> continue anyways to speak correctly the HELLO protocol for Link
> Quality Measurements for two reasons:
> 1) to let the node on the other side of the EDGE compute the normal
> ETX, in case in his configuration file there is not a fixed ETX
> 2) to check what the actual ETX would be, because if the real cost is
> greater than LINK_COST_BROKEN in will return
> of course in the TC messages the link will be propagated with the
> Fixed ETX cost, because the routing of the network must converge.
> Because we don't want to waste our time doing useless work that will
> not be merged upstream, tell us if we can start coding in this
> direction.
> ciao :)
> Saverio
> -- 
> Olsr-dev mailing list
> (spam-protected)
> https://lists.olsr.org/mailman/listinfo/olsr-dev

More information about the Olsr-dev mailing list