Re: [AVT] Lack of clarity in RFC 3550 on packet loss

[Rajesh Kumar <rkumar at cisco.com>]

Colin,
Thanks for the clarification.

RFC 3611 (RTCP-XR) defines a loss rate, which it says is after taking 
redundancy and FEC into account. It also defined a discard rate. It says 
nothing about whether this discard rate is after taking redundancy and 
FEC into account, but should we assume that this is so? I would think so 
- in order for RTCP-XR to be consistent with itself in this regard, even 
if it is not consistent with RTCP.

Thanks,

Rajesh

Colin Perkins wrote:

> Rajesh,
>
> On 3 Jun 2005, at 21:07, Rajesh Kumar wrote:
>
> > RFC 3550 does not explicitly say that the "fraction lost" and the  
> > "cumulative number of packets lost" fields in RTCP sender reports  
> > and receiver reports refer to packet losses after the effects of  RFC 
> > 2198 redundancy and RFC 2733 FEC have been taken into account.  
> > However, this is a reasonable assumption to make, and is in sync  
> > with RFC 3611, which does make such an assertion.
> >
> > Is this what the intent of RFC 3550 was?
>
>
> No, that is exactly the opposite of the intent of RFC 3550.
>
> The "fraction lost" and "cumulative number of packets lost" fields in  
> RTCP SR and RR reports always refer to the number of packets lost on  
> the wire.  The use of RFC 2198 redundancy or RFC 3722 FEC changes the  
> amount of payload data received, but does not affect the pattern of  
> packet loss on the network, and hence doesn't affect the packet loss  
> statistics reported by RTCP (indeed, reporting loss after the effects  
> of RFC 2198 or RFC 2733 would break applications that use RTCP RR  
> packets to control the amount of redundancy they add).
>
> Colin

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