Re: [rtcweb] RTP Usage: Is RTP Retransmission REQUIRED or RECOMMENDED

["Roy, Radhika R CIV (US)" <[e-mail]>]

Hi, all:

Real-time protocols (Audio/Video payloads for two-way conversations) over UDP/IP were being developed based on the single thing: Retransmission delays are unacceptable (more over congestion problems especially for the high-bandwidth videos).

If this fundamental performance characteristic for the Real-time protocols (Audio/Video payloads for two-way conversations) need to be changed, the experimental results with lots of actual measurements must be produced to the technical communities.

Otherwise, someone may make some subjective statements here and there without any actual measurements will only provide evidences what those folks do not know what they are actually talking about creating technical confusions.

Best regards,
Radhika

-----Original Message-----
From: [e-mail] [mailto:[e-mail]] On Behalf Of Cameron Byrne
Sent: Wednesday, June 27, 2012 12:19 PM
To: Fabio Pietrosanti (naif)
Cc: [e-mail]
Subject: Re: [rtcweb] RTP Usage: Is RTP Retransmission REQUIRED or RECOMMENDED

I think Bernard noted this as well, but it is not so much buffer bloat... but the issue is indeed latency, not loss.  Consequently, re-transmission not productive.

On Wed, Jun 27, 2012 at 12:42 AM, Fabio Pietrosanti (naif) <[e-mail]> wrote:
> If you use RTP over a Mobile Networks subject to strong packet loss 
> (especially 3G or area plenty of many mobile devices), the RTP 
> retransmission became a requirement.
>

Please cite the packet loss you are referring to.

Here is one imperfect data source,
http://pam2012.ftw.at/papers/PAM2012paper6.pdf.

I say it is imperfect because it relies on SACK, which can infer packet loss, but frequently for mobile it just infers delay.  So, the mobile packet loss is overstated.

Mobile (UMTS, ...) networks have extremely robust forward error correct and re-transmission capabilities at layer 1 and layer 2, this induces delay but ensure packet delivery.

See: http://en.wikipedia.org/wiki/Hybrid_automatic_repeat_request

Attempting to optimize for packet delivery at higher levels (RTP
re-transmission...)  leads to various counter productive race conditions, since the RAN (radio access network) is designed for 100% packet delivery.

CB

> If there is RTCP support within RTCWeb, we may define some "quality 
> threshold" after that the RTP Retransmition became REQUIRED.
>
> Fabio
>
> On 6/27/12 9:36 AM, Magnus Westerlund wrote:
>> WG,
>>
>> We had a discussion at the interim if RTP Retransmission is to be 
>> considered REQUIRED or RECOMMENDED to implement. I would like to see 
>> if we can first have some discussion on this topic before moving on 
>> to see if we can get a consensus here on the mailing list.
>>
>> Please provide your views on this topic.
>>
>> Cheers
>>
>> Magnus Westerlund
>> (As Chair and document editor)
>>
>>
>> ---------------------------------------------------------------------
>> - Multimedia Technologies, Ericsson Research EAB/TVM
>> ---------------------------------------------------------------------
>> - Ericsson AB                | Phone  [phone] Färögatan 6                
>> | Mobile [phone]
>> SE-164 80 Stockholm, Sweden| mailto: [e-mail]
>> ---------------------------------------------------------------------
>> -
>>
>>
>> _______________________________________________
>> rtcweb mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/rtcweb
>
> _______________________________________________
> rtcweb mailing list
> [e-mail]
> https://www.ietf.org/mailman/listinfo/rtcweb
_______________________________________________
rtcweb mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/rtcweb