[Diffserv-interest] RE: streaming/QoS

>>1.Of all the existing applications,video and audio streaming demands more 
>>QoS mechanisms.
>>Is their is any study relating these streaming applications with diffserv.
>well google seems to have some pointers to various papers. Is that what 
>you're asking?

  well google did point me to this literature .
http://pender.ee.upenn.edu/~guerin/publications/sigcomm_camera_174_final.pdf
	Here,Dr.Guerin studies the mapping by passing video streams through policers
from which application level performance is evaluated.But,this study is basically
focused on how policing actions by EF is translated into application level performance.
But,I think a combination of AF and EF as pointed out in section 2.4 in
http://www.ietf.org/internet-drafts/draft-baker-ieprep-requirements-00.txt will work fine.
But,examining the drop precedence within those seven classes of traffic will be more 
useful.Do you accept my contention?

>>2.This question is more or less related to the first.Please refer to some
>>research work which discusses relation between Audio& video qulaity and 
>>network
>>performance.Some thoughts about this topic is also appreciated.
>
>I didn't see a question mark. What was the question?

  Is there are any existing study discussing the relation between audio&video
quality with network performance ?..This was my question.
Google failed to help me at this point :-)

>>3.Is expedited forwarding only way to give guaranteed treatment to 
>>audio/video
>>streaming applications
>no, it's just the only one that works :^)
>A little more seriously, there are basically three queuing policies that 
>diffserv generally applies to a class. It can give it priority, which it 
>publicly only does using EF. It can give it a rate or a percentage of the 
>line, and use the rate of the queue, augmented in the worst case active 
>marking or dropping, to slow down traffic when the aggregate exceeds the 
>available bandwidth. It can do the second and additionally use specialized 
>marking to ensure that the subset of the aggregate most affected is the 
>part that is over-using some contract. This third, of course, is AF.

If my understading is not wrong,this correctly reflects your idea of seven 
classes as in 
http://www.ietf.org/internet-drafts/draft-baker-ieprep-requirements-00.txt

>>4.E-model of ITU-T estimates MOS by mapping voice quality with impairments 
>>like delay, signaling and user expection.Can we related these impairments 
>>with diffserv drop precedence for mapping streaming application with QoS.

   This reference will be more apt to my question.
http://www.acm.org/sigcomm/ccr/archive/2001/apr01/ccr-200104-cole.pdf
   This work basically describes a method for monitoring the quality of internet
 paths to support voice.I asked this question thinking in the same line as the above 
pointed work.

>Could you forward a pointer to that which doesn't require membership in a 
>closed user group to review it?
   You  mentioned that to be a closed group..Thats why i am  senting  questions to a
open forums like IETF  to get help :-). 

Thanks in advance ,
-senthil

_______________________________________________
Diffserv-interest mailing list
Diffserv-interest@ietf.org
https://www1.ietf.org/mailman/listinfo/diffserv-interest