[OLSR-users] LQ Routing and 802.11b Link Speed

[Sven-Ola Tuecke]

Thomas,

consider upvaluing links within multihomed hosts. Such links are usually 
faster because no store-and-forward needed. Example with 3 nodes using 
different channels or modes:

fastest: (1:WL) -> (2:WL+WL) -> (3:WL)
fast: (1:WL) -> (2:WL+ET) -> (3:ET+WL) -> (4:WL)
slow: (1:WL) -> (2:WL) -> (3:WL)

The "fast" example is a double node, built with two WRT devices bound by 
gaffa tape and an ethernet wire. The "slow" example are 3 nodes in ad-hoc.

Regards, Sven-Ola

""Thomas Lopatic"" <(spam-protected)> schrieb im Newsbeitrag 
news:(spam-protected)
> Hi Andrew,
>
> Yes, you're correct. If all three links have zero packet loss, then the
> direct link is better ETX-wise. The idea behind ETX is to calculate the
> necessary retransmissions of a packet. From A directly to C is only one
> transmission (assuming zero packet loss). From A via B to C would be two
> transmissions of the packet (again assuming zero packet loss).
>
> Once the LQ code is stable, we should definitely think through this whole
> ETX business again. Maybe we can come up with something even more
> suitable.
>
> What you can do as of 0.4.9, however, is to use the LinkQualityMult
> parameter to make the link between A and C artificially bad. Set the
> parameter to 0.2 on one of the two nodes (see the LQ read-me file that
> comes with the distribution) and the link quality used by olsrd will only
> be 20% of the actual link quality.
>
> -Thomas
>
> _______________________________________________
> olsr-users mailing list
> (spam-protected)
> https://www.olsr.org/mailman/listinfo/olsr-users