This signal, heard on 8542.0 KHz/USB looks like a normal MS 188-110 Serial consisting of a 1800Hz tone, modulated with PSK-8 at rate of 2400 symbols/second... but since the presence of several serial-tone waveforms on the air and the Appendix C and D of MS 188-110B/C, I preferred to give it a closer look just to see which animal it could be.
Looking for its speed I got the expected value of 2400 Baud, but looking for the frequency of the carrier noticed different harmonics patterns that suggested the presence of different modulation techiniques. Then I searched the PSK costellations in all the messages of the transmission getting PSK-8, QAM-16 and QAM-64 (pictures 1,2,3).
Well, these waveforms are described by the'Appendix C' of MIL 188-110B (and the more recent C) High-rate serial-tone HF waveformsas well as by the'Appendix D' of MIL 188-110C WBHF Block 1 capability, since 188-110A doesn't provide those modulations:later we'll try to determine which of the two we are facing. Anyway, the data-rates provided by the signal are 4800bps (in PSK-8), 6400bps (in QAM-16) and 9600bps (in QAM-64).
Well, these waveforms are described by the'Appendix C' of MIL 188-110B (and the more recent C) High-rate serial-tone HF waveformsas well as by the'Appendix D' of MIL 188-110C WBHF Block 1 capability, since 188-110A doesn't provide those modulations:later we'll try to determine which of the two we are facing. Anyway, the data-rates provided by the signal are 4800bps (in PSK-8), 6400bps (in QAM-16) and 9600bps (in QAM-64).
It is worth noting the differences bewteen the harmonics generated by PSK-8 and QAM-64 modes (pic. 4). It's also visible that the typical 188-110 QAM-64 constellation in 'circular rings' is modified with respectto the standard one (pic. 5).It's not so frequent to see QAM-64 in HF, so my thanks to KarapuZ for giving this signal for my analysis.
 |
| pic. 1 |
 |
| pic . 2 |
 |
| pic. 3 |
 |
| pic.4 |
 |
| pic.5 |
Talking with AngazU in order to determine whichsignalis, we passed to study the ACF/frame and the synchronization preamble.
frame length and ACF
188-110B/C App.C has a fixed 288 symbols lenght frame for all its waveforms (pic. 6) while the 188-110C App.D has different frame structure depending on the bandwidth and the waveform number (pic. 7).
frame length and ACF
188-110B/C App.C has a fixed 288 symbols lenght frame for all its waveforms (pic. 6) while the 188-110C App.D has different frame structure depending on the bandwidth and the waveform number (pic. 7).
 |
| pic. 6 - frame structure for 188-110C App. C |
 |
| pic. 7 - frame lenghts and structure for 188-110C App. D |
Since the bandwidth of the heard signals is within 3 KHz, we have that the two frames differ by only 1 symbol: 287 (Appendix C) and 288 (Appendix D for waveform numbers 7-10). The measuredACF (pic. 8) is ~120ms lenght, then almost 288 symbols, and matches the value reported in the standard in picture 7. This result leads to think to the Appendix D.
 |
| pic. 8 - 120ms (288 symbols) frame |
synchronization preamble
To be sure about it, we should look at the synchronization preambles.
It's worth to note that the synchronisation preambles in both the appendices are always modulated with PSK-8. Appendix C states: "The synchronization preamble shall consist of two parts. The first part shall consist of at least N blocks of 184 8-PSK symbols to be used exclusively for radio and modem AGC. The value of N shall be configurable to range from values of 0 to 7 (for N=0 this first section is not sent at all). The second section shall consist of 287 symbols. The first 184 symbols are intended exclusively for synchronization and Doppler offset removal purposes while the final 103 symbols, which are common with the reinserted preamble, also carry information regarding the data rate and interleaver settings." So, the legth of the App.C sync preamble is the sum of the two sections: (N * 184) + 287 symbols and since the symbol rate for all symbols is 2400 symbols-per-second, the lenght expressed in milliseconds will be: (N * 76.6) + 119.6 ms
![]()
The lenght of the App.D sync preamble is a bit more difficult to calculate, but it's interesting the comparison in picture 9
 |
| pic. 9 - sync preamble comparison between a secure App.C and the sygnal under analysis |
The upper signal in pic. 9 has a 426ms length preamble and matches the N=4 in the above table while the bottom signal has a shorter preamble and does not fallwithin the rangeof the allowed values. The preambles have PSK-8 modulation and the final sections of preambles and the mini-probes have QPSK modulation.
re-inserted preamble
The Cross-Correlation Function (CCF) on App.C signal clearly exhibits re-insertions each 72 data-blocks, or 8600ms (pic. 10) while the same function on the signal under analysis doe not exhibit such feature (pic. 11): this is another evidence in favor of App.D
 | |
| pic. 10 |
 |
| pic. 11 |
So, reassuming the clues:
- ACF lenght = 120ms, or 288 bit (is 119.42, or 287 bit for App.C)
- synchronization preamble not in range of App.C possible values
- no preamble re-insertions (App.C needs re-insertion each 72 data-blocks)
most likely the heard signal belongs to 188-110C App.D 3KHz bandwidth. As a confirm I analyzed a QAM-16 decoded bitstream obtained from SA, the bitstream has been previously converted into ASCII binary file using a little tool wrote by me in Lua: in pic. 12 it's easy to see the 32 symbols (128 bits as the modulation is QAM-16) forming the known-data of miniprobes
 |
| pic. 12 |