JT9 is a mode designed for amateur QSOs and beacon-like transmissions
at MF and LF.  The mode uses the same 72-bit user messages as JT65.
Convolutional error-control coding (ECC) uses constraint length K=32,
rate r=1/2, and a zero tail, which leads to an encoded message length
of (72+31)*2 = 206 bits.  Modulation is 9-FSK: 8 tones for data, one
for synchronization.  Sixteen symbol intervals are used for
synchronization, so a transmission requires a total of 207/3 + 16 = 85
channel symbols.  Symbol durations tsym are set to approximately
(TRperiod-10)/85, where TRperiod is the T/R sequence length in
seconds.  The exact symbol lengths are chosen so as to make nsps, the
number of samples at 12000 samples per second, a number rich in
factors less than 7.  This choice makes for efficient FFTs.  Tone
spacing of the 9-FSK modulation is df=1/tsym, the same as the keying
rate.  The total occupied bandwidth is 9*df.

Parameters of the five JT9 sub-modes are summarized in the following
table, along with S/N thresholds measured by simulation on an AWGN
(additive white Gaussian noise) channel.  

-------------------------------------------------------------------------
Mode     nsps nsps2   df    tsym  BW    S/N*   Tdec  Tfree   Factors 
        12000   750  (Hz)   (s)  (Hz)   (dB)   (s)    (s)    of nsps
-------------------------------------------------------------------------
JT9-1    6912   432  1.736  0.58 15.6  -26.9   52.5   7.5   2^8 3^3
JT9-2   15360   960  0.781  1.28  7.0  -30.2  112.3   7.7   2^10 3 5
JT9-5   40960  2560  0.293  3.41  2.6  -34.4  293.6   6.4   2^13 5
JT9-10  82944  5184  0.145  6.91  1.3  -37.5  591.0   9.0   2^10 3^4
JT9-30 250880 15750  0.048 20.91  0.4  -42.3 1788.5  11.5   2^5 3^2 5^3 7
-------------------------------------------------------------------------
* Noise power measured in a 2500 Hz bandwidth.


Transmitting
------------
1. Source encode the structured message to 72 bits
2. Convolutionally encode (K=32, r=1/2) to (72+31)*2 = 206 bits
3. Interleave to scramble the bit order
4. Assemble 3-bit groups to make (206+1)/3 = 69 symbols
5. Gray-code the symbol values
6. Insert 16 sync symbols ==> 69+15=81 channel symbols, values 0-8


Receiving
---------
1.  Apply noise blanking with the timf2 method
2.  Filter to 500 Hz bandwidth, downsample (1/16) to 750 Hz complex
    (FIR filter with 61 taps).  Complex data to array c0 (max 1.35M)
3.  Compute symbol-length spectra at half-symbol steps.  Use for
    waterfalls s(15750) and detecting sync vectors; save in ss(184,15750)
    and savg(15750)
4.  At time Tdec, look for sync vectors in ss(); get estimates of DF, DT
5.  Do full-length FFTs, NFFT1=96*nsps2
6.  For each candidate signal, do inverse FFT of length 1536.  This 
    will provide 16 complex samples per symbol, and sync tone should be 
    close to f=0.
7.  Use afc65b method to get improved values of DF, DT.
8.  Tweak freq and time offset to 0.
9.  Compute 8-bin spectra of 69 data symbols: s2(8,69).  Re-order the
    bins by removing Gray code.
10. Compute soft symbols for 206 bits
11. Re-order the bits by removing interleaving.
12. Pack bits into bytes, send to Fano decoder
13. If Fano succeeds, remove source encoding and display user message.