[SIPForum-discussion] RTP packet loss

Fri Feb 20 22:56:13 UTC 2015

I don’t think we have any control over ISP – they are not regulated to offer VoIP QoS.

From direct experience – either use dedicated data network that can provide priority to RTP packets or try to mitigate it a bit with adaptive jitter buffer (and on access side you need a higher value than in the core) with a maximum window toward 800ms-1sec that can deal with packets out of order. If the jitter buffer you are using is too small the RTP packets are going to be discarded. So to normal packet loss over such long haul, the discarded packets also look like packet loss from user audio experience perspective.

I know that none of these is easy to get but at least plan for something. I learned that the SIP applications at two ends can make a difference in the way they are designed. At first I could not believe that such high jitter buffer would work but in our case if a SIP call comes over internet from a client and gets out to the internet to the same type of client, it works pretty well. But client to our core, not really – it is bad experience on core side receiving (that has the jitter buffer very low).

From: Raghul Prasanna [mailto:raghul82 at yahoo.co.uk]
Sent: February 20, 2015 01:24 AM
To: Aniella Juverdeanu; Rajesh Kumar Talluri; discussion at sipforum.org
Subject: Re: [SIPForum-discussion] RTP packet loss

HI All,

Thanks for your responses.

Apart from the AudioCodes, I have already tried most of the things suggested by you guys.
Checking wireshark stats etc yes it does say packet loss\ wrong sequence\ skew etc, but my question is where is it happening.

Certainly not in our network.
Talked with ISP, not very helpful though. They even advise me to run MTR traces,which is NOT very useful in my opinion and when I share the MTR trace results with them unfortunately they were NOT able to pin point anything.

Problem is some of these calls cross continents.

Sometimes I wish we just stayed TDM, until we find enough bandwidth for everyone....

So what do you guys reckon, somehow get the ISP to capture some traces on their equipment to show the media leaves them and take it from there?

Any other thoughts please?

Raghul

On Friday, 20 February 2015, 3:31, Aniella Juverdeanu <Aniella.Juverdeanu at telus.com<mailto:Aniella.Juverdeanu at telus.com>> wrote:

Yes, AudioCodes SBC – we do not use it – just read their white paper

thanks

From: Rajesh Kumar Talluri [mailto:rajeshsources at gmail.com]
Sent: February 19, 2015 06:39 PM
To: Aniella Juverdeanu
Cc: Raghul Prasanna; discussion at sipforum.org<mailto:discussion at sipforum.org>
Subject: Re: [SIPForum-discussion] RTP packet loss

Hi Aniella,
You mean to say that Audiocodes box will check the QoS parameters on the service provider network and provide 1-2 sec. Are you talking about Audiocodes SBC here? Or Is it a different?
Thanks,
Raj
On Feb 19, 2015 12:56 AM, "Aniella Juverdeanu" <Aniella.Juverdeanu at telus.com<mailto:Aniella.Juverdeanu at telus.com>> wrote:
>
> Hi,
>
>
>
> You might need to analyze the Jitter buffer on the segment that is the access into your network. If yours I setup too low, then RTP packets are discarded. Usually this happens if the access is from IP networks where there is no QOS (i.e. ISP providers). AudioCodes has a good approach to this adaptive jitter buffer – can provide up to 1-2 sec JB
>
>
>
> Thanks
>
> Aniella
>
>
>
> From: discussion-bounces at sipforum.org<mailto:discussion-bounces at sipforum.org> [mailto:discussion-bounces at sipforum.org<mailto:discussion-bounces at sipforum.org>] On Behalf Of Raghul Prasanna
> Sent: February 13, 2015 04:07 AM
> To: discussion at sipforum.org<mailto:discussion at sipforum.org>
> Subject: [SIPForum-discussion] RTP packet loss
>
>
>
> HI All,
>
>
>
> If there is rtp packet loss resulting in bad audio and the loss is NOT happening on the network we control, what are the options to identify where it's happening and possibly correct it.
>
>
>
> MTR traces, I am unable to rely on this as core routers generally tend to discard these ICMP packets giving us distorted results.
>
>
>
> Any thoughts?
>
>
>
> Thanks
>
>
> _______________________________________________
> This is the SIP Forum discussion mailing list
> TO UNSUBSCRIBE, or edit your delivery options, please visit http://sipforum.org/mailman/listinfo/discussion
> Post to the list at discussion at sipforum.org<mailto:discussion at sipforum.org>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://sipforum.org/pipermail/discussion/attachments/20150220/2f6fd46f/attachment-0002.html>