From: John D. Hay@hays.org
Upon return from my vacation, I read with interest the thread about "Digital ATV".
At first I was somewhat apalled about the notion of putting video on
"unproto" packets - something that would be an unfortunate
misapplication of the technology, but the notion of just using unencrypted MPEG streams would seem to be more reasonable.
About the maximum baud rates: 19600 on 2 meters (even though the
"standard is 19200" but who cares) with 56 "k"baud on 220
and 70cm. Keep in mind that the rules specify symbol rates, not bit rates. Therefore, if one were to run something like 1024-ary
modulation scheme (10 bits-per-baud) and somehow confine its bandwidth appropriately, one could actually run just under 200
kbits-per-second. Of course, the link budget for this sort of circuit would be unrealistic, not to mention the likely cost of the
On the bands above 450, there are no restrictions on the allowable data/baud
rates per se: Anything is allowed as long as the other
rules are followed (i.e. noninterference, allowable modulation schemes and codes, etc.)
As for the suggestion that we as amateurs use the proposed "ATV" (Advanced
TV) schemes that the terrestrial broadcasters are
using, I feel that this would be unrealistic. This system was designed with bandwidth constraints in mind. As far as performance
goes, it is rather poor, actually. A few db of ripple across the passband, and a few extra degrees of incidental phase modulation,
and you have blown your Eb/No away: Considering how most amateur television AM signals look as far as amplitude and phase
response, I don't hold too much hope that it would work well at all - even the broadcast industry is dubious about the ultimate
workability of the terrestrial schemes...
I would suggest that the modulation scheme used for space-based delivery
systems (i.e. direct-to-home satellite) be used, namely
QPSK. This is reasonably robust and spectrum-efficient and really fairly difficult to screw up in an amplifier chain. One just
needs to take reasonable care in short-term stability of the converter chains in the transmitter and receiver (very easy to do,
fortunately) and implement decent tracking demodulators (also quite easy to do...) and you are all set.
As far as how the video should be represented, I think that anything
is fair game for experimental use. But from a truly practical
standpoint, our options are somewhat limited at this point. In my opinion, having worked with compressed digital video
broadcasting for the past several years, the absolute minimum bitrate for "reasonable" quality video and audio, using an MPEG
scheme, would be at T1 rates (about 1.544 megabits-per-second.) This still results in many visual artifacts that many people might
find objectionable, but it would certainly be tolerable for amateur use. (Notice that I haven't mentioned frame rate until now - For
the most part, frame rate is irrelevant when discussing an MPEG encoder that is worth anything... Data rate has to do with
updates of picture elements and how fast how much detail can be resolved more than a "frame rate.")
While they are "neato," software-based videoconferenceing systems shouldn't
really be strongly considered as the basis of the
"ultimate" system... at least for now... Until affordable processing horsepower increases by another order of magnitude (which
should be pretty soon...) the CODECs based on dedicated hardware have the edge.
I would propose looking at something like CCIR-601 for a start.
This is an MPEG-like video compression "standard" that works
pretty well and is well implemented on some pretty inexpensive chips (such as the Analog Devices ADV601 and related series
(see http://products.analog.com/products/info.asp?product=ADV601 for a bit more information...) I would be interested if anyone
has evaulated the $200 "video pipe" evaulation board based on this chip mentioned on this page. It would be interesting to see how
practical it would be to transcode CCIR-601 to a DVB-compliant bitstream (with audio) to allow the use of commercially-available
DVB-compliant satellite recievers (such as the Echostar receivers) which, like most modern satellite receivers, already can tune
the 23cm amateur band...)
I welcome comments, of course...