[CUWiN-Dev] G and A and realworld throughput

Stelios Valavanis stel at onShore.com
Wed Jul 13 17:41:49 CDT 2005


yes, got it.

On Wednesday 13 July 2005 04:12 pm, David Young wrote:
> On Wed, Jul 13, 2005 at 01:52:00PM -0500, Stelios Valavanis wrote:
> > ok. that means that even if it's uplink is 20 hops away it's going to
> > connect to the next node over because of fewer packets dropped.
>
> I think we understood each other, but I just want to make sure:
>
> The routers are always sending beacons to their neighbors within radio
> range, called "Hellos."  They count the number of Hellos that pass
> successfully in each direction.  From that, a router estimates, for each
> neighbor, roughly how many times its radio will retransmit a packet before
> the neighbor successfully receives it.  This "expected (re)transmission
> count" was abbreviated ETX by the researchers at MIT who came up with it.
>
> Routers assign "scores" to every path on the network.  The path's score
> is the sum of the ETXs on all of the links on that path.  Paths with
> lower scores are preferred.
>
> Dave

-- 
_______________________________________
stel valavanis  http://www.onshore.com/


More information about the CU-Wireless-Dev mailing list