Re: [Roll] ETX metric

[Henning Rogge <hrogge at googlemail.com>]

Am Dienstag 13 Oktober 2009 14:23:17 schrieb Richard Kelsey:
> > > ... On the other side ETX might be easy to implement on any kind
> > > of layer-2 hardware, but is only a rough estimation of the "link
> > > quality" which does not consider long time link statistics
> > > (stability, variance, ...)  or transmission speed.
> 
> > I have seen other people on this thread indicating that RSSI
> > measurements tell a rather polluted truth.  This is also our
> > experience.
> 
> I agree with you about raw RSSI.  On the other hand,
> combining raw RSSI with sampled noise measurement to get a
> measured SNR (or SINR) does much better.  It gives some idea
> of the margin available on the link.
> 
> > Translated to the real world, ETX information based on
> > the number of received Acks from lower layers is a cheap
> > and often sufficient metric.
> 
> I and at least some others have not found this to work well
> for 802.15.4, in that ETX isn't stable enough to be a good
> predictor except in the very short term.  I am aware that
> there disagreement on this :-).
My experience comes from 802.11, so we have more range compared to 802.15.4.
If all 802.15.4 implementations have a good RSSI and noise/interference level 
measurement, this might be the better default metric for ROLL.

802.11 implementations sometimes have a poor, ugly or non-existant RSSI 
measurement, and noise/interference level measurement is worse. That's why I 
would not suggest signal-strength/SNIR based metrics for WLAN. I'm not sure 
how well 802.15.4 do the necessary measurements.

Henning

Attachment: signature.asc
Description: This is a digitally signed message part.