Re: [Ekiga-devel-list] strange functioning jitter buffer

[Eugen Dedu <Eugen Dedu pu-pm univ-fcomte fr>]

alex wrote:
> Hi all,
> I'm run a set of measurements about playing delay of the jitter buffer
> in Opal, using a testbed. For taking out time values and seq number
> packets, I made light modifications in the file jitter.cxx (in
> particular in the function: PBoolean
> OpalJitterBuffer::ReadData(RTP_DataFrame & frame)) and rtp.cxx. These
> values were inserted in files .csv  (sending.csv contains time and seq
> num of packets transmitted, playout.csv those ones of packets played)
> 
> Plotting these  values  extracted,  I noticed an unusual behaviour of
> the playing delay:
> 1- presence of _discrete levels_ in both directions (caller-->callee,
> caller<--callee)
> 2- presence of a _drop_ only in uplink direction (caller-->callee)
> 
> Studying the debug file referring "no traffic" simulation, I observed a
> "periodic" resizing buffer.
> 
> - Could you help me to understand the adaptive algorithm used for
> resizing the buffer?
> 
> - Is it possible that there are two or more buffers implemented in Opal
> or that Ekiga has a particular equipment to equalize the playing delay??
> Would you please  give me  any  advice/help/useful link??

You have done an interesting analysis.  Please send the e-mail to opal
upstream developers at
<opalvoip-user lists sourceforge net>

-- 
Eugen