NetNews Usenet Archive 1992 #16/NN_1992_16.iso/spool/comp/dcom/cellrel/228

Text File | 1992-07-21 | 3.4 KB | 81 lines

Newsgroups: comp.dcom.cell-relay
Path: sparky!uunet!gatech!cc.gatech.edu!news
From: [e-mail] (Ioanis Nikolaidis)
Subject: Re: Issues, mostly unresolved
Message-ID: <[e-mail]>
Followup-To: comp.dcom.cell-relay
Sender: [e-mail]
Organization: Georgia Institute of Technology
References: <[e-mail]>
Date: Tue, 21 Jul 1992 20:10:52 GMT
Lines: 68

Related to article <[e-mail]>
from [e-mail]:
>3) 10 Mbps cell relay will not support HDTV?
>
>  True, but it supports compressed video quite nicely, and uncompressed
>low resolution video, as well has high quality sound, voice mail, and
>works great for image distribution and distributed data bases.  It allows
>for deskttop controlled conferencing.  These things are important, probably
>more important than HDTV in the private network business.

In the document COM XVIII-R 34-E of CCITT Study Group XVIII -
(Report R 34) of June 1990, on page 46, a specification of the
service integration for video is given.
They recognize the enormous differences of the
video terminal equipment capabilities and link speeds and
propose the following "solution":

Video coding is performed in a "layered" fashion.
The sender of a video connection encodes the
signal with a layered approach. Each layer "adds up" some more
detail on the visual result of the previous layers.
If all layers are decoded and presented at the video
terminal, then the images are presented
with the best fidelity.

However, the terminal equipment may be capable of receiving
only some of these layers (and not necessarily all). The number
of layers received by a terminal is related of course to the
quality of the equipment. A terminal is free to decode only part
of the transmitted layers to obtain just enough information to
reconstruct the image at its own level of image fidelity.
This alone "solves" the diversity problem of video terminals.

The same approach can be used to cope with limited bandwidth
BUT we need an active network entity that selectively discards
the cells related to the "redundant" layers and in a way "filters
out" the portion of the bandwidth that is used for improvement
of the image. Hence, the connection can be squeezed in a slower
link but with a penalty on image quality.

The subject was left for "further study" in report 34.

My question is the following:

If an approach is implemented as described above, what is the
network component that will discard the "upper" "enhancement"
layers of the video connection to reduce the required bandwidth
(to be conveyed on a (say) 10Mbps link)?
Will this component work on the ATM layer or on the AAL?
(If you reply "the ATM layer" you must show me a place in
the header where the video layer "level" is placed. If you
reply "the AAL" the question is if we will need reassembly
of video packets.)

Has there been any followup to the idea?
OR is my copy of the standard terribly outdated?

Final note: The direction towards a "layered" approach
does not encourage diversity of video coding
algorithms. In fact, all applicable coding methods must be
in reference (at least) to the overall layered structure -- i.e. if we
decide that layer 1 is luminosity and layer 2 is hue (a B+W receiver only
uses layer 1) then a coding algorithm that "mixes"
luminosity and hue information is not appropriate.

---
Yannis Nikolaidis
([e-mail])