RE: [AVT] questions about jitter

["Alan Clark" <alan.d.clark at telchemy.com>]

>So for many uses, while amusing, the RTCP data on loss (and probably
>jitter) is close to worthless, and certainly has to be taken as at best
>fuzzy data.  Post-jitter data (handling late packets, duplicates, etc) is
>more useful in most cases.  I'm sure part of the reason it's defined this
>way is so an RTP stack can calculate the loss & jitter simply without
>having to communicate in any way with the jitter buffer.

This is one reason that the RFC3611 VoIP metrics:-

(a) tightens up the definition of "lost"
(b) reports discarded packets, i.e. a packet is either usable, lost or
discarded
(c) reports discards rather than jitter, as it is a more concrete measure of
what happened as a result of jitter

The whole problem becomes a little more complex when you assume that jitter
is a time variant process, which it generally is.  What is the meaning of a
short term average jitter measure when the jitter level is not constant?

With regard to the relationship between jitter level and jitter buffer
size - this also is a more complex problem in practice as jitter is
typically heavily time varying and jitter buffers are often dynamic.

AS mentioned in an earlier posting to AVT - there is an interesting (free)
tool on the www.voiptroubleshooter.com web site - on the "online diagnosis"
page.  This uses a time series model as a basis for generating a sequence of
packet delays and exports these to a log file for analysis if needed.

Alan



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