Re: [rohc] ROHC ver 09 question

[Mikael Degermark <micke@CS.Arizona.EDU>]

Hi Ana, 

The time from context damage happens until an update is received consists
of two parts: detection time plus link roundtrip time. 

The detection time is determined by the packet interval time and the values for
k and n used in the k-out-of-n rule used to detect bad context.

The rtt depends on the link used.

For a cellular link and telephony, I would expect that the detection time is 
two to three packets, and the link roundtrip time would be 4-10 packets, 
depending on the link. 3G links will probably have a link rtt of 4-7 packets.

So what will happen to the compressed packets arriving during this time?
A decompressor MAY use the reverse decompression idea, in which case
they will be delayed, but this is not required. For interactive voice, it would 
probably be better if all or at least most were discarded. 

Ok?

Mikael Degermark

At 11:06 AM 3/2/01 +0100, Ana Carolina Minaburo wrote:
>Hi!
>
>I have a technical question for an implemetation issue:
>When the compressor stars sending packets to the decompressor, and the
>decompressor can sends a feedback to it. For example in the
>Bidirectional Optimistic mode.  How many packets are lost before the
>compressor can know there is a problem in the decompressor? Or the
>decompressor keeps the packets? If this is the case, how many packets
>aproximately are kept in the buffer before they can be decompressed?

>Thanks
>Ana
>
>
>---
>Mailing list for Robust Header Compression WG
>Archive: http://www.cdt.luth.se/rohc/
>