opensand team mailing list archive

[Julien BERNARD <jbernard@xxxxxxxxxxxxxxxxxxx>]

Hi,

> we have carried out a test with enabling the physical layer. In the GW 
> configurations, we have set up an On/Off uplink attenuation model. 
> What does On/Off stand for ? Does it mean that the model sometimes 
> completely cut the upcoming signal ? You can see inclosed the probes 
> collected for the following ping : -s1000 -i0.15 -c100, which 
> theoretically corresponds to a 53kBits/s throughtput.
> The signal sent to SAT is switched off between 14 and 17s and we were 
> wondering if it was normal, and if it is, when does the attenuation 
> model choose to cut it off ?
I think you may find some answer on the wiki.
For details about how physical layer is designed you can check this page:
http://opensand.org/support/wiki/doku.php?id=design:physical_layer

If you still have questions after this I will be glad to answer.

> We have very few knowledge about satellite communications so here is 
> another question : what are the common throughputs in satellite 
> communications and how large are the packets sent ? At first, we 
> thought that a throughtput between 1 and 10Mbits/s was not absurd, 
> with packets size from 100 to 1000 bytes, but it involves ping 
> intervalls of less than 1ms and we cannot simulate such time intervals.
The throughputs and packets size may vary depending on the satellite 
applications (video broadcasting, internet applications, voip, ...).
Moreover, the traffic profiles depend on your link (forward (GW->ST) or 
return (ST->GW)), you will typically have higher throughputs on forward 
links than you will on return links.

Your problem is that ping is not adapted for traffic generation. You 
should try iperf for example.

Regards,

--
Julien BERNARD