Hi, Does anyone know why the jitter buffer behave that way? I have encounter the scenario describe in my previous mail multiple times while running some test with pjsua and I would very much like to know if this is normal. Any help would be appreciated, Alexandre Devos On Mon, Jul 8, 2013 at 5:40 PM, Alexandre Devos <alexandre.devos at genaker.net > wrote: > Hi, > > I have been looking at the jitter buffer for a few days now and I notice > something strange about the way the burst level is handle. > > Lets take an example with a few frames coming in and out of the buffer: > > ... > PUT -> framelist size = 7 , jb_level = 2 > PUT -> framelist size = 8 , jb_level = 3 > GET -> framelist size = 7 , jb_level = 4 > > jbuf_calculate_jitter(...) is called > > GET -> framelist size = 6 , jb_level = 1 > GET -> framelist size = 5 , jb_level = 2 > GET -> framelist size = 4 , jb_level = 3 > PUT -> framelist size = 5 , jb_level = 4 > > PUT -> framelist size = 6 , jb_level = 1 > PUT -> framelist size = 7 , jb_level = 2 > PUT -> framelist size = 8 , jb_level = 3 > GET -> framelist size = 7 , jb_level = 4 > > jbuf_calculate_jitter(...) is called > > ... > > As I see it, the first GET of a series is counted in the PUT burst level > and the first PUT of a series is counted in the GET burst level. > But what if the level suddenly rise on the first PUT? > We get something like this: > > ... > > jbuf_calculate_jitter(...) is called > > GET -> framelist size = 6 , jb_level = 1 > GET -> framelist size = 5 , jb_level = 2 > GET -> framelist size = 4 , jb_level = 3 > PUT -> framelist size = 22 , jb_level = 21 > > PUT -> framelist size = 22 , jb_level = 1 (frame arriving out of order) > PUT -> framelist size = 22 , jb_level = 2 (frame arriving out of order) > PUT -> framelist size = 22 , jb_level = 3 (frame arriving out of order) > GET -> framelist size = 21 , jb_level = 4 > > jbuf_calculate_jitter(...) is called > > ... > > Here, the rise in level is not taken into account when > jbuf_calculate_jitter(...) is called. > > So my question is: > Do we not take the first PUT of a series into account on purpose? (and > why?) > Or is this a bug? > > Best regards, > > Alexandre Devos > > > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.pjsip.org/pipermail/pjsip_lists.pjsip.org/attachments/20130717/8fa1ca51/attachment-0001.html>
