Re: [Fwd: Re: incompatible feedback type 2 causes problems in reusing the context for profile change]

Suneeta, Biplab,

Suneeta, thank you for bringing this issue to our attention. The different
modes in RFC3095 have always been a bit of a pain. I am not however
convinced that this issue is a very big problem in ROHC. It will only affect
the compression efficiency. The compressor will stay in the current mode
 (U-mode in this case) until the feedback for IR/IR-DYN/UOR-2-Ext3 is
recieved with a mode field that acknowledge the mode transfer.

The compressor could probably figure out that it received feedback
from another "context". The sequence number may give a clue or
mode-bits in the following feedback. The CRC in the feedback is not
bullet-proof and there is always a possibillity of bit-errors in
feedback.

Biplab, I am not aware of any backward compatibility issues between
RFC3095 and RFC5225 profiles. If you have found any such issues,
please send details to the rohc-list.


Best Regards,
Carl Knutsson, ROHC WG Chair


On Mon, Sep 8, 2008 at 2:40 PM, Carl Knutsson <[email protected]> wrote:
>
>
> -------- Original Message --------
> Subject:        Re: [rohc] incompatible feedback type 2 causes problems in
> reusing the context for profile change
> Date:   Mon, 08 Sep 2008 17:49:32 +0530
> From:   Sarkar Biplab <[email protected]>
> Reply-To:       [email protected]
> To:     Suneeta Rao <[email protected]>
> CC:     [email protected]
> References:     <[email protected]>
>
>
>
> Hi Suneeta,
>
> I believe you can't have direct backward compatibility between ROHCv1
> and ROHCv2. Only one should exists. This problem will not exist between
> feedbacks from different protocols in the same version of the ROHC
> specification.
>
> Thanks & Regards,
> Biplab
>
>
> On Mon, 2008-09-08 at 04:16 -0700, Suneeta Rao wrote:
>>
>> Hello ROHCers,
>>
>> I have a question on changing profile from ROHCv2 to RFC3095 and vice
>> versa. Please note that I only have a basic simulation environment.
>>
>> When I change from ROHCv2 to RFC3095 in compressor, I am getting some
>> stale feedback from the peer decompressor which was for ROHCv2.
>> However, because of incompatible formats between ROHCv2 and RFC3095,
>> compressor is interpreting parts of Ack no as "mode bits" of RFC 3095.
>> This is being treated as a mode transition.
>>
>> I'm pointing this out as this looks like a ROHC specification issue.
>> In general, what is the expected behaviour if profile change is done
>> from ROHCv2 to ROHCv1? How are stale packets to be handled? I haven't
>> looked at other profiles, but I guess the issue might be there for TCP
>> profile also.
>>
>> Regards,
>> Suneeta.
>>
>>
>> _______________________________________________
>> Rohc mailing list
>> [email protected] <mailto:[email protected]>
>> https://www.ietf.org/mailman/listinfo/rohc
>
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.6 (GNU/Linux)
>
> iD8DBQBIxRhU3rQBTW1UPu4RAr2JAKCmdEeeN9uw/EhDUotjLn3hvpyTUwCgtmse
> WvwudYd9SMChxqZepI5Uiwg=
> =3gms
> -----END PGP SIGNATURE-----
>
>
> _______________________________________________
> Rohc mailing list
> [email protected]
> https://www.ietf.org/mailman/listinfo/rohc
>
>
>