>73,
>Daniel Fox
>KF9ET
Dan,
I subscribe to the tapr HFSIG. And about a year or 2 ago, for about a year,
discussed HF modem
techniques. The idea is to spead out the information so fading and multipath
can be supressed
and still recover the data. There were/are PHD^2 kind of guys and real
engineers who really work making
HF modems for NATO and that mil standard 188. So the discussion was really
state of the art and
then some. I have built some modems (in a previous life) so I understand the
problems, but it
was everthing I could do just to hang on to the discussion. Anyhow this is
all archived and it does
make some interesting reading. It will probably take about 6 months to
absorb it.
Anyhow there are modulation schemes, that can be received by different
bandwidth receivers.
ie a wide band receiver receives the data faster than a narrow band
receiver, but they both
receive the data, but at different rates!!! There's a name for the signal
but I don't remember it.
Something like orthogonal but I'm not sure. There was a paper in IEEE
Proceedings on this,
I'm sure I've saved it somewhere.
Something like this might be wavelets. (Wavelets are like Fourier
Transforms, but instead of
integer harmonics, you use octave or double the frequencies, more or less).
The modulation
would send the data at the fundimental, then send it twice at twice the data
rate, then send it
4 times at 4 times the data rate, etc etc. All of this at the same time........
I was thinking about this for my (imaginary) active seti transmission. eg
send prime number
sequence of pulses.... The fundimental would be 1 pulse per day. Leave the
transmitter on
for a day and let the earth spray out the signal. Then off for a day; then
on for 2 days and off
for 2 days; then on for 3 off for 3; then on for 5 off for 5; etc up to
maybe 31 or a years worth.
This would produce a pulse about 10 minutes long per day at the receive end.
Now on this 10 minute pulse would be the pattern. 5min on 5min off; the 5
min pulse would have
the pattern on it; etc etc. or if more of a true wavelet, the pattern of
the first 10 or so primes
would sent in the pulse, and in these faster pulses would also be made up of
the sequence pattern
etc etc
The idea here is that however you integrated, eg over a month, every day
then averaged 30 days,
over 10 minutes and average a months worth of 10 minutes, etc etc ...the
pattern would
bubble up out of the noise.
over....Don wa8qzz