Re: Re-send: Simple estimate for Nr. of retransmissions (WAS:RE: [AVT] IESG comments on draft-ietf-avr-rtp-retransmission-11.txt - Issue 4: more parameters needed for RTX?)

Jose Rey wrote:
> 
> I have reviewed the details again and made the assumptions a bit
> clearer.  I still have one question to Magnus: you said the
> avg_rtcp_size would be around 125 bytes.  Is this based on your
> experience? 
> 

125 bytes is a figure I use for correctly formed RTCP with CNAME strings 
but still almost minimal assuming no extra blocks than NACKs. If one 
actually uses more information then this figure is of course larger. I 
think it is a reasonable lower size of the RTCP packet even if smaller 
is possible.

Example

  40 IPv6
   8 UDP
  32 Receiver Report with 1 report block
  29 SDES: CNAME (23 chars CNAME string)
  16 FB: NACK
====
125

 From the above it becomes obvious that shorter packets are definitely 
smaller, but I don't see that any packet below 100 bytes are really 
useful in the discussed scenario and I think there are a couple of other 
configurations that also end up in around 125 bytes. For example the 
streaming case with retransmission with SSRC multiplexing which becomes:

  20 IPv4
   8 UDP
  56 Receiver Report with 2 report block (Primary and RTX SSRC)
  29 SDES: CNAME (23 chars CNAME string)
  16 FB: NACK
====
129

A Sender packet is only 4 bytes smaller than the above due to the SR + 1 
report block.

Thus I feel it is a reasonable number, but definitely not the only one 
one can use.

Cheers

Magnus Westerlund

Multimedia Technologies, Ericsson Research EAB/TVA/A
----------------------------------------------------------------------
Ericsson AB                | Phone +46 8 4048287
Torshamsgatan 23           | Fax   +46 8 7575550
S-164 80 Stockholm, Sweden | mailto: magnus.westerlund@ericsson.com

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