[rohc] ROHCv2 Implementation msn, sn and ipid in co_common/co_repair.

[Carl Knutsson <carl.knutsson at effnet.com>]

Hi Kristofer and Ghyslain,

I have some suggestions for the compressed formats co_common and co_repair for all
profiles. I would like to have fields that differ between profiles last in the
packet formats as much as possible, because of implementation reasons. I would
also like to see depending field located after the field they depend on the
compressed formats as much as possible.

I suggest that ipid field is moved after the 'msn' field in co_common and
co_repair for all profiles. ('s/msn/sequencenumber/' for esp profile). ipid is in
many cases dependent of the msn.

I would also like to see a two changes in the encoding method of the sequence
number in the ESP profile:

First, I would like the 'sequencenumber' field to be encoded using the
REORDER_RATIO in co_common. The reson for this is to make packet selection easier
and make the calculation of p the same for all k.

My second suggestion concerning the ESP sequence number is to remove the last 32
bit option in the sdvl-encoding(co_common). 32 bit can be sent co_repair (with
fewer bytes) and there is no need for redundancy here:)

This would mean a new encoding function.

To help for you guys out.. here is the FN (that is if the working group agrees to
my changes).

svdl_esp_sn {
  UNCOMPRESSED {
    sn [ field_width ];
  }

  COMPRESSED lsb7 {
    discriminator =:= '0'               [ 1 ];
    sn =:= msn_lsb(7)                   [ 7 ];
  }

  COMPRESSED lsb14 {
    discriminator =:= '10'              [  2 ];
    sn =:= msn_lsb(14)                  [ 14 ];
  }

  COMPRESSED lsb21 {
    discriminator =:= '110'             [  3 ];
    sn=:= msn_lsb(21)                   [ 21 ];
  }

  COMPRESSED lsb29 {
    discriminator =:= '111'             [  3 ];
    sn =:= msn_lsb(29)                  [ 29 ];
  }
}

/Calle

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