Re: help

[Dr G Fairhurst <gorry@erg.abdn.ac.uk>]

The current ID was published some months ago, and the authors would
like to update it soon, so these thoughts come at a good time....

ghommam mohsen wrote:
> 
> Hello and thank you for your help, so I want to clarify with you the
> following points:
> 
> These questions are dealing with a new optimised method for
> encapsulation of packets: Data piping to minimise overhead of MPE ( 16
> bytes)
> 
> 1.the LL_pid ( 2 bytes) and packet lengh (2 bytes) are used in the IP
> format packet ( level 3 of IP V4) proposed by Mr Clausen ( working on
> this solution, so this is why also we can avoid them, at level 2, what
> do you think?
> 


Can you explain more about how you propose to find the length of a
frame (SNDU) at the receiver?

> 2.to distinguish between traffics ( vidéo, IP, etc) we can use the PID
> ( 13 bits) of TS MPEG, so no need of LL_pid and lengh in the
> encapsulation?

So, the encapauslation would normally be used with IP traffic, and
MPEG video would be sent with a different PID - unless it was using
RTP (etc) to send it as IP packets. In fact, I'd expect many applications
to carry several PIDs on the same MPEG-2 TS, each using the encapsulation,
to provide more flexibility.

> 3.May be we need only CRC ( 4 bytes) please why, it is 4 bytes?
 
Because that seems to be safe (see the work in the IETF PILC WG)

> Regading the points above, I think the overhead in datapiping will be
> only 4 bytes (CRC) or 0 bytes, if we can avoid CRC ?

Avoiding a CRC - or using a weak CRC - means that (a) you have to 
provide a reliable way to detect MPEG-2 TS Packet loss, multiplexing
errors, etc. (b) You also need to provide guarentees that the MUX, Gateway
and Receiver do not corrupt packets. This may be harder. Since IP
relies on this information for proper operation, it's safest to 
provide this at this layer - an assumption of IPv6.

> Please I need your comments about what I have said.
> 
> Another thing is what other parameters than overhead, we can measure,
> for example to install quality of service and measure it ( can you
> give an idea how to simulate or to implement the quality of service
> for datapiping and measure it, please give some WWW links useful to
> find these information or send some of them by mail if you have this
> information.

Perhaps someone else on this list has information to hand?

> and thank you very much
> 
> ----------------------------------------------------------------------
> Do You Yahoo!? -- Une adresse @yahoo.fr gratuite et en français !
> Testez le nouveau Yahoo! Mail

Gorry Fairhurst.