[AVT] Non-compound RTCP and Early AVPF feedback

[Colin Perkins <csp at csperkins.org>]

One of the open issues with the non-compound RTCP draft is whether  
multiple non-compound early packets can be sent between regularly  
scheduled packets, when the RTP/AVPF profile is used. The draft  
notes, in section 5.1, that:

   In order to get as much as possible out of of the non-compound RTCP
   concept, it is necessary to relax the sending rules in RFC 4585
   [RFC4585] in such a way that it is possible to transmit more that  
one
   early feedback RTCP between regular RTCP, presently this is
   controlled by means of the allow_early flag which has the effect  
that
   only one early feedback message is allowed between the regularly
   scheduled RTCP.  The proposed relaxation is reasonable as the  
size of
   the non-compound RTCP is expected to be small compared to the  
regular
   RTCP.  The relaxation however needs to be done in a way that it is
   both bandwidth conservative and robust/scalable for an increased
   number of users.  A possible outline of an algorithm that does this
   overrides the allow_early flag to the value TRUE if the bandwith
   consumed by non-compound RTCP is within some predefined limits, one
   such limit may be that non-compound RTCP should not be allowed to
   consume more than a predefiend fraction of the total RTCP bandwidth.
   Another option may be to use the SDP attribute "trr-int" as limiting
   factor.

While relaxing the sending rules for non-compound early feedback  
packets might be an appropriate thing to do, it is important to  
remember that the AVPF rules were chosen after considerable  
discussion, to address certain requirements. Any decision to relax  
those rules would have to be accompanied by a discussion of how and  
why the requirements have changed since RFC 4585 was under  
discussion, and a justification why the existing mechanisms are not  
sufficient.

--
Colin Perkins
http://csperkins.org/



_______________________________________________
Audio/Video Transport Working Group
avt at ietf.org
https://www1.ietf.org/mailman/listinfo/avt