some questions

Asked by regulararmy on 2013-06-04

Hello Didier,
         I am not sure some logics for you codes, so ask you.
         Firstly,as for the function d_operation_mode_feedback(...) in rohc_decomp.c, there is a switch(mode) in it,the first case is U_MODE, which does not need feedback, however, the statement “break;” is removed by annotation //,is it a bug? what is more ,the feedback for R_MODE is not implemented yet, can I use your library in my production realization, Thanks.
         Secondly,I wonder if the compression or decompression logic of your library codes is as follows: the comp and decomp entity is initialized to U_mode ,then, the comp and decomp entity transiton to O_MODE or R_MODE or back to U_MODE automatically and adaptively ,I do not need to think them over?

Regards
chenxisheng

Question information

Language:
English Edit question
Status:
Solved
For:
rohc Edit question
Assignee:
No assignee Edit question
Solved by:
Didier Barvaux
Solved:
2013-06-08
Last query:
2013-06-08
Last reply:
2013-06-06

This question was reopened

Hello,

> Firstly,as for the function d_operation_mode_feedback(...) in rohc_decomp.c,
> there is a switch(mode) in it,the first case is U_MODE, which does not need
> feedback, however, the statement “break;” is removed by annotation //,is it a
bug?

The decompressor starts in U-mode. When it decompresses the first packets, if it is associated with a compressor, it will always try to send a positive ROHC feedback with mode O-mode to the compressor. The U-mode -> O-mode transition is achieved this way. If you don't want the decompressor to ask compressor for O-mode, don't associate it with a local compressor.

> what is more ,the feedback for R_MODE is not implemented yet, can I use
> your library in my production realization

This question was already asked several times on the mailing list. Please check list archives before asking questions. R-mode is not implemented at all in the ROHC library. The library can be used in production (in U- and O-modes). In fact, I got reports that the library is currently used in production without any problem: no stability problem & great performances :)

> Secondly,I wonder if the compression or decompression logic of your
> library codes is as follows: the comp and decomp entity is initialized to
> U_mode ,then, the comp and decomp entity transiton to O_MODE or
> R_MODE or back to U_MODE automatically and adaptively ,I do not need
> to think them over?

Compressor/decompressor start in U-mode. If a feedback channel is defined (= decompressor associated with another local compressor), it will go to O-mode. R-mode is not implemented. There is no O-mode -> U-mode transition. All of those transitions are automated, you don't need to do anything. Just configure or do not configure a feedback channel for your decompressors.

Regards,
Didier

regulararmy (regulararmy) said : #2

I am sorry,,,Thank you for answering me patiently.do you prepare to realize the R_MODE in the near future? how long is it,my teacher? In additon, when will the TCP/IP profile be realized and appear in front of us readers?

regulararmy (regulararmy) said : #3

 As for the TCP profile. It is available in a separate dev branch: https://code.launchpad.net/~didier-barvaux/+junk/rohc-tcp . Use with some caution, it is still beta quality, can I use it now?

regulararmy (regulararmy) said : #4

As for what you answered above,I still have a question .Compressor/decompressor start in U-mode, If a feedback channel is not defined (= decompressor is not associated with another local compressor), just did not configure a feedback channel for your decompressors.if so,there is no feedback datas between two communicaiton entity(Compressor or decompressor) in the interactive process at all? is it right,please correct what I said,Thanks.

Hello,

> do you prepare to realize the R_MODE in the near future?
> how long is it, my teacher?

No plan yet. If you're interested in it quickly, either contribute to the library, or support someone to do it if you cannot do it yourself.

> In additon, when will the TCP/IP profile be realized and appear in front
> of us readers?
> As for the TCP profile. It is available in a separate dev branch: [...]
> Use with some caution, it is still beta quality, can I use it now?

The TCP profile lies in a separate branch for the moment. It is beta quality now, so you can use it for testing purposes but you should not use it for anything serious for the moment. I hope to integrate the TCP profile in the 1.7.0 release (no release date scheduled exactly, but don't expect it before 6 months at least since 1.6.0 is about to be released).

> As for what you answered above,I still have a question .
> Compressor/decompressor start in U-mode, If a feedback channel is
> not defined (= decompressor is not associated with another local
> compressor), just did not configure a feedback channel for your
> decompressors.if so,there is no feedback datas between two
> communicaiton entity(Compressor or decompressor) in the
> interactive process at all? is it right,please correct what I said

It is correct. If you don't configure a feedback channel, you won't have any feedback mechanism.

Regards,
Didier

regulararmy (regulararmy) said : #6

Thanks Didier Barvaux, that solved my question.

regulararmy (regulararmy) said : #7

Hello,
         If there are two communication entity(for example, femto cell and UE) ,each of them has ROHC function(compress and decmpress), in femto, decompressor is not associated with another local
compressor,just did not configure a feedback channel ,however, in UE,it has configured a feedback channel, in this case ,can they communicate with each other normally?

regulararmy (regulararmy) said : #8

what is more,the UE (user equipment) has implemented R_MODE of ROHC, but the femto production has not realized the R_MODE because of using your library codes, so,in this case, if they both have configured a feedback channel ,can they communicate with each other normally or compressing effectively?

Hello,

> If there are two communication entity(for example, femto cell and UE) ,
> each of them has ROHC function(compress and decmpress), in femto,
> decompressor is not associated with another local compressor,just did
> not configure a feedback channel ,however, in UE,it has configured a
> feedback channel, in this case ,can they communicate with each other
> normally?

They will communicate in U-mode. But some feedbacks will be transmitted from UE to femto for nothing.

> what is more,the UE (user equipment) has implemented R_MODE of ROHC,
> but the femto production has not realized the R_MODE because of using
> your library codes, so,in this case, if they both have configured a feedback
> channel ,can they communicate with each other normally or compressing
> effectively?

They will communicate in O-mode at best. Change to R-mode by the UE will be ignored by the femto.

Regards,
Didier

regulararmy (regulararmy) said : #10

Thanks Didier Barvaux, that solved my question.