RE: [rohc] RE: Default decompression algorithms

> -----Original Message-----
> From: ext Dr. Carsten Bormann [mailto:cabo@tzi.org]
> Sent: February 26, 2002 10:28 AM
> To: Miguel A. Garcia; Liu Zhigang.C (NRC/Dallas)
> Cc: rohc@ietf.org
> Subject: RE: [rohc] RE: Default decompression algorithms
> 
> 
> > I also agree with the requirement whereby a decompressor informs
> > the compressor of the algorythms that are already uploaded, and
> > therefore, don't need to be uploaded again. This is just to
> > save bandwidth over the air interface and avoid unneeded uploads.
> > I think this is a very valid requirement.
> 
> How many algorithms are there going to be in the cache of a typical
> decompressor?

That depends on the cache size. For a mobile, it should be a 
small number. For the network-side decompressor, it anyway has
to store decompression codes corresponding to each active compressor
(in mobiles). So, it may has a large number of codes.

> What is the cost of supplying this information?

The announcement message overhead = n * 6, where n is the number 
of decompression codes and 6 is the number of bytes for a hash. 
Let s be the average code size and p(n) be the probability of 
a hit. Then the savings = (s - 6n) * p(n).

Of course, the RTT as you mentioned is also a cost. But if the
announcement can be piggybacked on the reply from the decompressor
during the negotiation of the use of SigComp, it is not an extra
cost anymore. (But I don't know if the message sequence Hans 
outlined sometime ago is still valid.)

BR,
Zhigang

_______________________________________________
Rohc mailing list
Rohc@ietf.org
https://www1.ietf.org/mailman/listinfo/rohc