NetNews Usenet Archive 1993 #1/NN_1993_1.iso/spool/sci/electron/22469

Internet Message Format | 1993-01-12 | 3.6 KB

Path: sparky!uunet!munnari.oz.au!spool.mu.edu!agate!stanford.edu!enterpoop.mit.edu!eru.mt.luth.se!lunic!sunic!ugle.unit.no!humpty.edb.tih.no!lumina.edb.tih.no!ketil
From: [e-mail] (Ketil Albertsen,TIH)
Newsgroups: sci.electronics
Subject: Re: HDTV Question
Message-ID: <[e-mail]>
Date: 12 Jan 93 10:36:36 GMT
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Sender: [e-mail] (Ketil Albertsen,TIH)
Organization: T I H / T I S I P
Lines: 53
Posting-Front-End: Winix Conference v 92.05.15  1.20 (running under MS-Windows)

In article <[e-mail]>, [e-mail]
(Scott Dorsey) writes:

> [...] it would be very nice to have a system that could work with
>varying bandwidths and resolutions.  It would mean that the same standard
>could be used for broadcast television as would be used for a wider bandwidth
>cable system.  It could mean that black and white programs could be shown
>with sqrt(3) times the resolution of color programs, without using any
>additional bandwidth, just by reallocating color information into resolution.
>And, existing NTSC programs could be broadcast over the HDTV network without
>using up the full channel bandwidth, since it wouldn't be needed.

There is a tiny little start along those lines in the CCITT G.722 voice/data
standard, essentially ISDN-oriented. If you during a phone call would like
to fax a drawing to the one you are talking to, 8 or 16 kbits/s (of the 64
available) is "stolen" from the sound channel on the fly, and the sound must
manage with 56 or 48 kbits for a while. This will reduce the dynamic range
of the bass range, until the fax transfer is complete, but that will be
almost unnoticable for speech.

Obviously, fax is just one example of data that could be transmitted. And
if you only have a 48 kbit channel available, you could still use the G.722
scheme with no data channel.

Digitial mobile phone (GSM in Europe, the US won't accept any European
standards so they are making another one...:->) uses a lot more advanced
related techniques. Here, noise reduces (dynamically) the *effective*
bandwidth, and the goal of the encoding is to make the best possible
sound quality given the S/N level at the moment. They do it by using a
relatively larger fraction (by adding extensive error correction) of the
effective bandwidth for the most audible sound qualities, while more subtle
nuances with little error correction falls outside the effective bandwidth
when noise levels are high.

Obviously, video requires a lot more processing and channel capacity than
sound, but at least at high levels, the principles are roughly the same.
Eg. for 64 Kbit/s picturephone, the largest changes are transmitted first,
then the details. Watch a person changing position: Immediately, the screen
(or that part of it) breaks into large blocks, which gradually breaks into
smaller ones, granulating to full resolution within a second or two. If you
talk through the phone at the same time, it takes longer time because the
sound dynamically steals up to 14 kbit/s (in the system I have seen),
leaving only 50 to the picture.

A friend of mine works with digital HDTV, compressing the raw bit rate of
roughly 2 Gbit/s to 27 Mbit/s. He claims that the methods are the same,
although at 27 Mbits/s the time to reach full color and geometric resolution
is so short that you normally won't notice it.

(But don't ask me for details about G.722 and GSM coding - I have tried to
understand the standards, but I have given up. Insufficient math background!)

K.A.