RE: [rohc] Default decompression algorithms

["Abbie Barbir"<[e-mail]>]

Richard,

Thanks for the responce.
I am aware of what the compressor need to do. However, your remark states

"the compressor downloads a new algorithm to the decompressor when it 
judges the overall performance to be better than using one of the existing 
algorithms"

>From an implementation point of view, 
1. are you saying that the compressor must/should keep a list that relates
 a device's algorithm with the compression ratio etc that it could achieve?
2. Or the compressor will first attempt to compress using the algorithm that
is avilable at the decompressor and then it realize that i can get better
compression if this xyz algorithm is used, so what will do 
  - try to remember that for the next message
  - decide to go ahead with this message and then download a new algorithm 
    to the UDVM

Definitly other options could be used. The point here, is that the above
statement
can be hard to implement in the field.

abbie


> -----Original Message-----
> From: Price, Richard [mailto:[e-mail]]
> Sent: Thursday, February 21, 2002 12:57 PM
> To: Barbir, Abbie [CAR:1A00:EXCH]; 'Lars-Erik Jonsson (EPL)'; Conroy,
> Lawrence (SMTP)
> Cc: [e-mail]
> Subject: RE: [rohc] Default decompression algorithms
> 
> 
> ------------------- SNIP SNIP 
> > Making one or more of these well-known algorithms mandatory at the 
> > decompressor doesn't contradict the UDVM approach because 
> the decision on 
> > which algorithm to use (or whether to download a new one) 
> is still made by 
> > the compressor. So we have a "win-win" situation; the 
> compressor downloads 
> > a new algorithm to the decompressor when it judges the 
> overall performance 
> > to be better than using one of the existing algorithms. 
> > 
> -------------- 
> > Can you suggest how this would be done (implemented) from 
> the compressor end. 
> > abbie
> 
> Hi,
> 
> There are quite a few issues that the compressor will need to 
> consider when
> judging overall performance, including some or all of the 
> following metrics:
> 
> 1   Compression ratio
> 
> 2   Processing efficiency
> 
> 3   Memory efficiency
> 
> 4   Implementation effort
> 
> 5   IPR considerations
> 
> The IPR considerations are particularly important, since 
> encumbered algorithms
> tend to substantially outperform unencumbered algorithms in 
> terms of metrics
> 1, 2, and 3.
> 
> So if a compressor has licensed the bytecode for an 
> encumbered algorithm, it
> is usually preferable to download this bytecode to a 
> decompressor that does
> not support the encumbered algorithm, rather than picking one 
> of the weaker
> unencumbered algorithms already available at the decompressor.
> 
> Regards,
> 
> Richard Price
>