[olsr-dev] adding received signal info

[Federico]

On Tue, 24 May 2005 13:43:10 +0200
Andreas Tønnesen <(spam-protected)> wrote:

 
> spoggle wrote:
> > If you get results from this, please post them here.  How much of a 
> > burden is collecting the signal strengths?
>
> - The results are not consistent with different hardware. I have not
>    tested much with this but I was told that different HW would yield
>    different results. I was considering adding some threshold for
>    neighbor registration - but I ditched that idea after hearing this.


IMHO collecting SNR values could be useful while you are deploying a wireless network,
i.e. while pointing directional antennas. And, yes, you have to keep in mind that SNR measures can be different when using different NICs.

But in an up-and-running network SNR value is useless. You are no more interested in knowing this: you would consider each link's uptime versus downtime ratio.
This value depends on SNR, burst noise from the environment [i.e. microwave ovens] and the NIC's itself quality.

In fact, LQ extension is based on this. IMHO I would consider a long-term mean, like on a day, or a week. Common vaules could run from 10^-6 (very good), 10^-3 (usable) to 1/10 (bad!)

Just my .02 :)

Federico Ceratto