[AG-TECH] Question about Codec

[JongWon Kim]

Dear All,

This is JongWon Kim from K-JIST (Kwang-Ju Institute of Science &
Technology),
who is leading team of students on AG.

Interestingly we were working on the integration of DV into AG, starting
from this
spring. Since we was new to the source level code of AGTK, it took us
some time
to figure out how we handle the integration of DV into AG. Anyway, let
me introduce
the way of integration we are currently considering so that we can get
your valuable
comment.

To integrate DV into AGTk 2.0, we are taking a simplified integration
approach as shown
in the attached figure. Note that at this first phase of integration, we
want to focus on
the integration without having the burdon of seamless tie with existing
video services. 
Once completed, we want to go through another refinement...  
 
1) In order to touch the VV server to a minimum level, we are
considering of launching 
    separate DV producer/consumer services in DV-capable AG nodes only.

   Question: Can someone compare with modifying video producer/consumer
service?

2) DVTS-agent will be added and will handle the announcement of DV flows
(i.e., stream
    description) to all DV-capable nodes with, say, a DV-video multicast
announcement address/port.
    That is, each DV flow (i.e., stream) will use an unique multicast
address/port combination.
    This will allow each node to selectively tune-in the stream
interested.  The rate adjustability
    of DVTS, what Shudo mentioned, is another feature we are considering
in the regard.

    Question: An unique port for each DV flow with shared multicast
address is acceptable with this?

3) We are also confining the integration of DVTS to linux system only
due to source-code 
     availability of DVTS.

Will be very helpful if we can hear your valuable comments on the above
and others.

Yours!

jongwon
--------------------------------------------------------------------
-----    Kwang-Ju Institute of Science & Technology (K-JIST)   -----

   JongWon Kim
   Associate Professor
   Depart. of Information & Commun.          /   /
   E-mail: jongwon at kjist.ac.kr              /  /
   Tel: +82-62-970-2219                  / /    O
   Fax: +82-62-970-2204               <//__ --
                                         /     -/--/ / /
   http://netmedia.kjist.ac.kr            /     \/ \/\/
-------------------------------------------------------------------- 


-----Original Message-----
From: owner-ag-tech at mcs.anl.gov [mailto:owner-ag-tech at mcs.anl.gov]On
Behalf Of shudo at ni.aist.go.jp
Sent: Monday, August 11, 2003 1:48 AM
To: ag-tech at mcs.anl.gov
Subject: Re: [AG-TECH] Question about Codec



Hi all,

From: Gurcharan.S.Khanna at Dartmouth.EDU (Gurcharan S. Khanna)

> i'd like to second what bob would like to do with DVTS. using DV is
really
> great because there is no codec involved, no compression needs to take
> place. the DV stream is simply packetized and sent out and the
receiving
> system just has to render the DV stream which doesn't seem to take
> much cpu power at all.

I second.

We have been having a plan to integrate the DVTS into AG for 2 or 3
years.  Unfortunately, the project could not have enough priority to be
worked on.

In case of AG 1.X, maybe naive integration can be achieved by
modifying slightly the VRM, the VV server, and probably the Event
Listener.  It is only required to invoke the DVTS instead of VIC.
We have not yet considered the integration to AG 2.X.

> now all we need is the multicast user interface and we have DV quality
> video and audio in an AG setting. of course, some will say that 30Mbps
is
> too big a penalty to pay for this simplicity.

DVTS has capability to thin down video streams by sending one video
frame every several frames.  We can specify the number of frames in
which a frame is sent.  Bandwidth can be reduced much if a lower frame
rate is acceptable.

  Kazuyuki Shudo
  Grid Technology Research Center
  National Institute of Advanced Industrial Science and Technology
(AIST)

-------------- next part --------------
A non-text attachment was scrubbed...
Name: KJIST-AGDV.jpg
Type: image/jpeg
Size: 502072 bytes
Desc: KJIST-AGDV.jpg
URL: <http://lists.mcs.anl.gov/pipermail/ag-tech/attachments/20030812/37f9f899/attachment.jpg>