[R-C] The rate control behavior strange during test

[Michael Welzl]

Hi,

I think it's probably because it was assumed that one cannot send so  
much feedback (because RTP/RTCP would be used). But I think that this  
may be a mistake - see my next email.

Cheers,
Michael


On Apr 1, 2012, at 6:01 AM, Fu Jiantao wrote:

> I'm wondering why the algorithm of get queuing delay in utp ledbat( http://tools.ietf.org/html/draft-ietf-ledbat-congestion-00 
>  ) is not taken here, seems that's much simpler and effective, and  
> it can remove the clock skew.
>
> 2012/3/31 Fu Jiantao <fuji246 at gmail.com>
> Hi,
>
>    I've give the rate control implemented in webrtc a test, seems  
> its behavior is abnormal.
>
>    I'm using the peerconnection_client to do the test, and use  
> netlimit to set one-way bandwidth to 50KB/s, the rate of other  
> direction fluctuate dramatically, from 10KB/s to 200KB/s, I tried to  
> open the relating logs, but after that, the connection won't setup,  
> there seems some bug in the connection setup, I've updated the code,  
> and it failed to compile. Seems the jsep is not ready now. And I  
> can't test on it now.
>
>    So, it's there any problem in my test? Is there any existing test  
> report on this?
>
>
> Thanks,
> Jeromy
>
> _______________________________________________
> Rtp-congestion mailing list
> Rtp-congestion at alvestrand.no
> http://www.alvestrand.no/mailman/listinfo/rtp-congestion