[Olsr-dev] OLSRd 0.5.4 and 0.5.5 with ETT available
Erik Tromp
(spam-protected)
Wed Feb 20 14:32:02 CET 2008
See also my previous e-mail.
If you know a way to retrieve the actual per-neighbor auto-rate as chosen by
the radio, I would like to feed that in OSLR-LC . I'm unsure whether current
ad-hoc mode drivers keep that information and can deliver the information at
that detail level (e.g. via /proc ) to external processes.
The current OLSR-LC provisions in taking into account the per-link speed in
its link cost calculation. By default it is set, for each link on a certain
network interface, to the pre-configured speed value for that interface
(parameter "MediumSpeed").
The current OLSR-LC does not provision in asymmetric link costs; that would
be a nice extension for a next releast.
_____
Van: (spam-protected) [mailto:(spam-protected)]
Namens Markus Kittenberger
Verzonden: woensdag 20 februari 2008 14:19
Aan: Erik Tromp
CC: Henning Rogge; (spam-protected)
Onderwerp: Re: [Olsr-dev] OLSRd 0.5.4 and 0.5.5 with ETT available
> Don't you think the subject of your mail is a little bit misleading ?
If you choose a fixed (pre-configured) bit speed value for each network
interface (e.g. 11 or 54 MBit/s) the link cost is effectively the ETT as it
would be with that bit speed.
fixed values will only work with fixed wireless rates,
(using the same value) on both senders and receiver side!!
If you have a mechanism to read the actual speed of each available network
interface that would be a good improvement. In that way we could use the
radio-layer ARF mechanism as indication of link speed.
are you keeping in mind, that in auto rate mode the link speed may/will
differ to different neighbours
so there is no actual speed per interface, on a wireless network device,
just per link.
even if you set a fixed transmission rate, receiving rate is determined by
the neighbours transmission rate
Markus
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.olsr.org/pipermail/olsr-dev/attachments/20080220/623b3602/attachment.html>
More information about the Olsr-dev
mailing list