ROHC-over-802 (was: Re: IPDVB Provisional Agenda (IETF-75))

[Carsten Bormann <[e-mail]>]

> 	   R=1 (what I can decompress)	
> sender <------------------------------ receiver
> 	   R=2 (I will send using these params)
> sender -------------------------------------> receiver
> 	   R=4 (I too will send ROHC using these params)	
> sender <------------------------------------- receiver

Correct, that is what I'm proposing for the bidirectional case (e.g.,  
two devices connected by some form of bridged Ethernet/WLAN  
combination).  (The sender might then occasionally transmit R=4  
packets based on a NUD-like algorithm.)

> I'm not sure how ROHC will work properly with multicast.

That is the other case with R=6 at the end of section 6 (penultimate  
paragraph).
If the sender does not have direct information about the ROHC  
capabilities from its peer (which it would get based on negotiation  
packets with R=1 and R=4), the R=6 case allows setting up the  
compression based on some out-of-band information (such as  
configuration or some other negotiation protocol).  See the caveats in  
that paragraph.

Gruesse, Carsten