On Oct 15, 2006, at 8:45 AM, Bill Turner wrote:
> You can, of course, digitize the signal just as with any
> analog signal and process it, but the original signal was and is
> analog.
Bill, the RTTY and CW carriers are used to match the digital
information to the channel (in the Shannon sense, "channel
encoding"). The information is otherwise quite discrete.
Claude Shannon's original 1948 Bell Labs paper is reprinted as "A
Mathematical Theory of Communication" C. E. Shannon, W. Weaver. My
copy is a cheap worn out paperback from Univ. of Illinois Press, ISBN
0-252-72548-4. More humanly readable description of modern
communications theory can be seen here http://en.wikipedia.org/wiki/
Information_theory .
Chapter 1 of Shannon is titled "Discrete Noiseless Systems" where he
introduced the central theme of all modern communications - the
entropy of a signal and the channel capacity. The second chapter is
"The Discrete Channel With Noise" and the third chapter is
"Continuous Information" (what I think you have been calling
"analog") followed by "Continuous Channel".
Radio engineers look for the best channel encoder (in the case of ham
digital modes, this would be a choice of the best modulation method
for the HF channel) and also for the best source encoding scheme for
the information (Morse, Baudot, varicode, FEC, etc).
For me, one of the biggest contribution by Shannon leads to the
understanding that the channel encoder (what we call the modem) can
be optimized independently of the source encoder (Baudot, varicode,
etc) and the result of the cascade of the individually optimized
encoders is still globally optimal.
I.e., the optimal FSK modem is independent of whether you pass Baudot
or ASCII characters through it. Whether you choose Baudot, ASCII or
varicode depends on a completely different realm of source encoding;
nothing to do with the modem. This might be "obvious" but Shannon
was the first to establish the mathematics to show it.
If you can design the best modem for a given non-fading white
Gaussian noise channel, PSK beats FSK, and FSK beats OOK (CW). Add
in multipath, selective fading and QSB and things change.
By the way, your example of a house switch is digital from the user
interface viewpoint (on-off keyed), the carrier of the signal (AC
line) is analog to most engineers, but ultimately discrete again to a
quantum Physicist. So just because you think it is analog does not
mean that everyone else does :-). It depends on the hat they are
wearing.
As someone who designs RTTY modems as a hobby, I consider FSK to be
channel coding problem for a discrete set of information (Baudot).
But that is just my hat. Your hat could be the RF amplifiers and
antennas, which of course are continuous devices (unless you are a
quantum mechanic).
As long as the term "analog" is used to describe continuous signals,
I think it is quite appropriate to call discrete information "digital."
The only peace will come when we start calling it "continuous" vs
"discrete" rather than "analog" vs "digital." At that point we can
fall back to Shannon's description of what is discrete and what is
continuous.
Yes, Morse and Baudot are both discrete information sources by
Shannon's mathematics. You can find references to the entropy of
Morse code here for example http://en.wikipedia.org/wiki/
Huffman_coding .
And yes, unless there are quantum Physicists who wants to split
hairs, the HF channel is a continuous channel.
Through that continuous channel, we as hams pass both continuous
information (voice, SSTV) and discrete information (Morse, Baudot,
varicode)
Bill, your other example of digitizing a voltage is just a process of
channel encoding a continuous signal for a discrete channel. So,
channel encoding does not need to work from a discrete information
source to a continuous channel (as in RTTY), it can go between a
discrete source to a discrete channel (your computer bus), a
continuous source to a discrete channel (a digitizer) and a
continuous source to a continuous channel (analog phone lines, SSB,
etc).
"CW" or OOK (on-off keying, i.e., your key or paddle) is just the
channel encoding scheme to carry the discrete information into the HF
channel. The use of short dot and dash sequences to represent more
often used characters is a form of source encoding. The use of
abbreviations ("LL," "FB" "73" etc and Q codes) is also another form
of source encoding.
By the same token, even though they are today displayed on discrete
LCD computer monitors, non-digital SSTV and HF FAX are continuous.
HF FAX does not even use a horizontal sync but depends on a "IOC"
number which represent the ratio of the FAX drum to the raster scan
pitch. "Analog" SSTV modes do have horizontal sync, but there are
no individual pixels, rather they are represented by a continuous
signal that maps to dark and light regions of the picture.
73
Chen, W7AY
_______________________________________________
RTTY mailing list
RTTY@contesting.com
http://lists.contesting.com/mailman/listinfo/rtty
|