Tuesday, March 26, 2019

8-ary constellation bursts at 12800bps data rate

For background it might be helpful to read the relevant entries in i56578's blog.

Two consecutive bursts centered on 2670 kHz are shown below in terms of 287-symbol long frames. The preamble and mini-probes have been descrambled (gr-digitalhf) and therefore have phases close to 0 rad. There are 13 frames with payload data per burst, each one having 256 symbols/frame:

Symbols after adaptive filtering and descrambling of
MIL-STD-110C known symbols for consecutive bursts

These signals are perfectly compatible standard MIL-STD-110C for 12,800 bps data rate; note that for 12,800 bps there are no requirements on the interleaver and on the error coding specified. As was observed before, only 8 points out of the QAM-64 constellation are used.

Looking closer at the payload, a periodicity of 160 symbols was found. In terms of 160-symbol long frames the payload is shown below:

Payload in terms of 160 symbol frames.
The fact that (symbol2) above is the same for all frames indicates that the payload consists of BPSK-modulated data which is scrambled to the 8PSK-like constellation. So what is left is to determine the scrambling.

It turns out that the used scrambling sequence is based on one of the scrambling sequences specified in STANAG 4538, i.e., it can be obtained as an extrapolation of one of the S4538 scrambling sequences: writing the 3-bit scrambling sequence in binary notation as PN(i) = bA(i)B(i)C(i), one can verify that A(i), B(i), and C(i) are subsequences of the same pseudo-random binary sequence generated by a certain LFSR. Determining the LFSR and the offsets of the subsequences is left as an exercise for the interested reader.

Once the payload symbols are descrambled, it becomes apparent that they consist of 104 Walsh-modulated di-bits (104×32 = 13×256) where each 4-symbol long Walsh symbol is repeated 8 times:

Descrambled payload revealing 104 Walsh-coded di-bits. 

2 comments:

Antonio Anselmi said...

Hi Christoph, interesting post! You are probably right about the BPSK used for payloads, although 188-110B/C/D App.C (or S4539) does not provides such modulation (the lower is QPSK), but I tend to think that they use indeed the Walsh orthogonal modulation which is applied on each pair of data bits (dibits, not BPSK). Then, the walsh sequences are scrambled to appear on air as the 8PSK-like constellation (the 8 points of the QAM64 outer ring). Email will follow :)
73's
Antonio, i56578

Christoph Mayer said...

> although 188-110B/C/D App.C (or S4539) does not provides such modulation (the lower is QPSK),

The bursts are perfectly compatible with the 110C/S4359 standard.

> but I tend to think that they use indeed the Walsh orthogonal modulation which is applied on each pair of data bits (dibits, not BPSK).

This is what I described in this blog post