opensand team mailing list archive

[fabrice.arnal@xxxxxxxxxxxxxxxxxxxxx]

Hi all,

For simple introduction of traffic corruption losses in the system (random
errors/losses for instances) that may be interesting to have in some
studies using Opensand, we initially thought that it should be implemented
through the physical layer model (the model will be available in the
upcoming release).

Now,  I feel that a good alternative would be to use rather the tc / netem
interface that also performs this task, e.g:
"tc qdisc add dev eth0 root netem loss 1%" -> drops randomly 1% of all
packets received to the eth0 interface.

If the method was adopted, the physical layer model would be used for
advanced channel model or models based on time-series, whereas the previous
one could be applicable for simplest cases.
There is also the possibility to implement packet duplication, corruption,
re-ordering, and also a more advanced burst error model with 2 parameters.


The questions are the following :

-Has anyone had some experience / feedback on this tool (accuracy, etc.) ?
-According to you, is there any limitation on the place to implement the tc
rule (ST, SAT, GW), with respect to the future channel model ?

Best regards

Fabrice