rohc team mailing list archive

[Didier Barvaux <didier@xxxxxxxxxxx>]

> a/ Compressor & d-ecompressors are directly connected through wire.
> Netem queue add losses for compressed traffic which send out fromthe
> ethernet interface at compressor end.Furthermore, UDP/IPv4 traffic
> (periodic pattern) is generated from compressor towards de-compressor
> through MGEN. 

So, it's probably very regular. The packet loss shall not cause damage
on packets after the loss.


> b/ The mode which i have noticed through packet traces is O-mode.
> Because there is no periodic updates.

The ROHC library performs always periodic refreshes, even in O-mode.


> e/ The de-compressor fails for a consecutive number of packet drop
> e.g. 30 or more than 30 packets which is understood thing. But
> the surprising thing is that some time compressor goes back to a
> lower compression state (IR or FO) to send more information to the
> decompressor without error recovery request. It is like periodic
> refresh or time out period at compressor for a queue packet
> loss/delay. I cannot say it as periodic update because it is not
> unidirectional mode.

There are periodic refreshes in bidirectional mode. You can check the
hypothesis by looking at the library traces. There are dedicated traces
printed when periodic refreshes are performed.


Regards,
Didier

Attachment: signature.asc
Description: PGP signature