[Olsr-dev] Seeking comments: OLSR+ETX v/s DSR+ETX

[Henning Rogge]

Am Mittwoch 19 Dezember 2007 17:49:15 schrieb elektra:
> as I understand the idea is to measure the actual throughput for unicast
> traffic. With broadcasts you can't measure link speed - since broadcast are
> transmitted at the same broadcast rate. You can - however - transmit with
> different broadcast rates and see which neighbor will still respond to your
> packets and measure the packet loss for different broadcast rates.
Yes, with layer 2 support that would be possible.

> still I think that I have to justify what we did years ago: we thought the
> approach should be as generic and as universally  applicable as possible.
> counting lost broadcasts was just that.
I don't think you need. I experimented with RFC OLSR implementations and the 
ETX version of olsr.org during my diploma. RFC OLSR sucks... ETX is not 
perfect, but it's much better than anything without ETX I tested.

Just an idea... would it be possible to extract the calculation of the link 
costs of the routing graph into some kind of plugin system, so other metrics 
could be implemented easily ? With a system like this, olsrd could get an 
optional wrapper for understanding radiotap messages so some of the metrics 
could involve layer 2 informations.

Henning Rogge