Hi,
Yes, the Wireshark implementation of a 'jitterbuffer', if you may call it, is
rather poor. But then again, do you really need a high performance jitter
buffer? It's not that you can use Wireshark to do MOS tests for you. That is
really up to the performance of the endpoint, not the 'man in the middle' as
Wireshark is.
Thanks,
Jaap
Krzysztof Cabaj wrote:
Hi,
Due to my research concerning jitter buffer, I tried to use Wireshark
to evaluate a number of packets dropped by buffer. However, there is
no description of jitter buffer implementation in Wireshark.
Therefore, I have verified source codes for version 1.2.4. I analyzed
file gtk/rtp_player.c and function decode_rtp_stream.
And, I have few concerns.
There is no real buffer, but only a number representing hypothetical
buffer size (variable jitter_buffer).
Packets are “dropped” when variable diff is greater than jitter_buffer
(line 758).
Variable diff represents absolute value in ms between real time of
receiving this packet and time when this packet should be
received (line 742).
I would be very grateful if somebody who know this part of code could
confirm these?
Regards,
Krzysztof Cabaj
