[AVTCORE] comment on draft-westerlund-avtext-rtcp-sdes-srcname

["Pascal Buhler (pabuhler)" <[e-mail]>]

One of the problems mentioned in the draft, that is the case when receiving the RTP streams before the SEDS has arrived is something that needs to be addressed. The option of including the mapping in the SDP is not always appropriate due to devices like an RTP Translator in the middle. Therefore I was wondering if you have considered including a definition of a RTP header extension (http://tools.ietf.org/html/rfc5285) to include the SRCNAME in the RTP stream. Yes it is potentially a lot of bits wasted but it is up to the sender to decide the size of the SRCNAME so that can mitigate it somewhat. Also it may not be required in every packet just packets that make sense for example only in the beginning of a stream, at key points in the stream or else periodically. When using the SRCNAME to correlate two or more separate encodings of the same source it is important as a receiver that when first receiving one encoding and then switching to receive a second encoding to make this correlation as soon as possible.


There is also mention of using this as a way correlating FEC with media. How would this work in the simulcast case where there are multiple encodings of the same source. How to use only SRCNAME to map the FEC stream to the individual media streams when all media streams have the same SRCNAME.

Thanks

Pascal Buhler