Wireshark-users: [Wireshark-users] A question about calculating "Mean Jitter" KPI for RTP flow

From: Nan Xiao <xiaonan830818@xxxxxxxxx>
Date: Wed, 22 Sep 2021 16:33:57 +0800
Hi Community,

Greetings from me! Very sorry for interrupting again!

I find the calculation of "Mean Jitter" KPI for RTP flow is like this (https://gitlab.com/wireshark/wireshark/-/blob/master/ui/tap-rtp-analysis.c#L497):

statinfo->mean_jitter = (statinfo->mean_jitter*(statinfo->total_nr-1) + current_diff) / statinfo->total_nr;

But in my humble opinion, maybe we should use current_jitter instead of current_diff? Like following one:

statinfo->mean_jitter = (statinfo->mean_jitter*(statinfo->total_nr-1) + current_jitter) / statinfo->total_nr;

Actually, I submitted an issue in gitlab before (https://gitlab.com/wireshark/wireshark/-/issues/17600), but unfortunately, it seems no one noticed it. So very sorry for reposting here, and hope someone can help to clarify it.

Thanks very much!

Best Regards
Nan Xiao