[Olsr-dev] mac80211 / link-bandwith from userland forcostcalculation / plugin

[Rogge Henning]

> > > /tmp/olsrd/$REMOTEIP/lqmult
> > > (which is a simple textfile, e.g. "0.1")
> 
> > I don't think this would work well, because we can only 
> encode a range 
> > of 0-255 for LQ/NLQ. If you combine the link loss with the link 
> > bandwidth, you need a different encoding with more range.
> 
> you are right 8-( but 8 bits should be enough for everyone 8-)
We will still have 8 bits in OLSRv2, but it will be an exponential encoding (similar to the validity time in OLSR).
 
> > You could write a metric plugin that reads the bandwith 
> from any kind 
> > of interface and combine it with the link loss to generate 
> LQ/NLQ. No 
> > need for some "lq-params" hack. I would be glad if this 
> parameter will 
> > be unnecessary in future versions because of better metric plugins.
> 
> ok, then simple read e.g.
> 
> /tmp/olsrd/$REMOTEIP/bandwith
> (which is a textfile, units in [kbit], e.g. 48000 or 5500)
> 
> this should default to 1000.
> then we have to find a way to find a calculation for "cost".
> maybe an simple switch()-table:
> 
> [rate] = [cost]
> 1000 = 600.000
> 54000 = 11.111
> 300000 = 2.000
> wire = 1.000
> 
> somebody has to find a good table, maybe not linear...
> the textfile is ofcourse only a quick and dirty solution, but 
> works OS-independent...
> 
> maybe there is no need for link-loss (and therefore 
> combining), because minstrel already finds the "best" rate? 
> (but needs some data-transfer in regular intervals)
This would only work if you send some IP packets over every discovered unicast link every few seconds. Not difficult to do, but without traffic on the link minstrel does not work.

Henning Rogge

--
Diplom-Informatiker Henning Rogge , Fraunhofer-Institut für 
Kommunikation, Informationsverarbeitung und Ergonomie FKIE 
Kommunikationssysteme (KOM) Neuenahrer Straße 20, 53343 Wachtberg, 
Germany Telefon +49 228 9435-961,   Fax +49 228 9435 685 
mailto:(spam-protected) http://www.fkie.fraunhofer.de
GPG: E1C6 0914 490B 3909 D944 F80D 4487 C67C 55EC CFE0