This burst signal, spotted same days ago on 6550.0 KHz/USB, has already been studied in this post. In short, it consists of a 3 x 1048ms bursts transmission, each burst consists of a preamble followed by data blocks structured in 32 symbols frames. The used waveform is the conventional PSK-8 modulation of a 1800Hz carrier and symbol rate of 2400 Bd. The 3-bursts transmission is repeated at regular intervals.
Talking about this unidwaveform, my friend AngazU - intrigued by its weird 8-ary constellation - suggested a different analysis approach, based on the DSSS (Direct Sequence Spread Spectrum) technique.
Being an interesting as much as new point of view, I asked AngazU to produce a very basic introduction to DSSS and then we re-thinked the analysis of the burst waveform bearing in mind the main features of some DSSS signals, such as GLONASS and PacTOR IV (shortly analyzed here): the results of this comparison are very interesting.
Below the story, keep in mind we deal only with free/cheap software from the Internet. No expensive hard /soft tools as the ones available for official organizations. Should the reader have some quality wav recordings of potential DSSS or FHSS signals, we will do our best to analyze them
Direct Sequence Spread Spectrum (DSSS)
DSSS ( Direct Sequence Spread Spectrum) techniques are becoming quite popular. Some well known uses are Sats (GPS,Glonass, etc) 3G mobile comms ( CDMA) , W-LAN, etc.
In DSSS the message signal is used to modulate a bit sequence known as the Pseudo-random Noise (PN) code; this PN code consists of pulses (chips) of a much shorter duration (and then larger bandwidth) than the pulse duration of the message signal, therefore the modulation by the message signal has the effect of chopping up the pulses of the message signal and thereby resulting in a signal which has a bandwidth nearly as large as that of the PN sequence. The resulting signal resembles white noise, just like an audio recording of "static". However, this noise-like signal is used to exactly reconstruct the original data at the receiving end, by multiplying it by the same PN sequence (de-spreading) which is known by the receiver.
Key parameters are: PN sequence length, the chip-rate and the spread factor, that refers to the expansion of signal spectrum. The chip rate of a code is the number of pulses per second (chips per second) at which the code is transmitted (or received). The chip rate is larger than the symbol rate, meaning that one symbol is represented by multiple chips. The ratio is known as the spreading factor (SF) or processing gain:
PacTOR IV
We know that Pactor IV uses M-ary modulation and DSSS techniques in some modes. We run the envelope detector in order to get the modulation speed (pic. 1): the recording in use for analysis is from our friend Karapuz.
 | |
| pic. 1 - modulation speed in the AM envelope |
The AM envelope shows a clear line at 1800 Hz but in the three middle segments there are also other lines (eight) starting starting from 225 Hz and multiples. This is an odd feature in normal PSK signals but indeed a characteritic of DSSS. In this case the symbol-rate is 225 sps and the chip-rate is 1800 cps, that make a spread factor (SF) = 8.
Another important feature can be seen looking at the phase vector of te DSSS segments (pic. 2)
 |
| pic. 2 |
In this picture the phase evolution has a sawtooth structure: according to MIL-STD 188-181A (1), this means that the constellation keeps rotating point by point in only one sense.
Note that in PacTOR IV signals the 8-ary constellation is obtained using the value of the chip-rate (1800 symbols/sec) rather than the (lower) real symbol-rate. The reason is that we only see the over-the-air bitstream and it should be de-spreaded before to get a valid demodulation.
GLONASS
Glonass is the russian GPS, and uses FDMA and DSSS for every carrier. Data is well known, so we use it to prove the concept. We know that Glonass sats DSSS use a chip rate of 511 Kb/s. In this case, chip rate is much bigger than symbol rate. Picture 3 shows speed measurement of a Glonass sat signal. It is the chip rate and we get a quite good value, since the nominal is 511000 MHz.
 |
| pic. 3 |
 |
| pic. 4 |
Using the carrier and the symbol-rate we get the PSK-2 constellation of the signal (pic. 5)
 |
| pic. 5 |
After demodulation you can get bits from GLONASS signals and use a bit editor to prove that chip code is 511 bits (pic. 6)
 |
| pic. 6 |
the 'unid PSK-8 burst waveform'
In light of what was seen aboout signals that use DSSS, mainly the presence of "harmonics" in the envelope detector and the shape of the phase vector, it's worth resuming some aspects of this signal.
For what concerns the modulation speed, we get a clear speed line in 2400 but also many lines starting from 150 Hz and up in multiples (pic. 7) that are a clue of the DSSS technique: in this case using a spread factor SF = 16 (2400/150).
 |
| pic.7 |
Another important clue comes from the analysis of phase vector of the signal (pic. 8)
 |
| pic.8 |
The phase vector has mostly four states and segments of eight states and probably this is the reason of the "nuanced points" in the constellation of picture 9. The insteresting thing is its shape that in some parts is like to the sawtooth seen in the phase vector of PacTOR IV. As said, it means that the constellation rotates in only one sense and according to MIL-STD 188-181A, this will produce an undesired carrier shitf that could be a source of errors in a concentional PSK demodulators.
 |
| pic.9 |
The differences against a conventional PSK-8 waveform (MS188-110 Serial Tone), mostly evident in the phase vector evolution, are shown in picture 10:
Assuming the value of 2400 as the chip rate, the resulting 96 bit period should be the length of the PN sequence (pic. 11)
 |
| pic. 10 - AM envelope and phase vector for the MS188-110 ST waveform |
 |
| pic.11 |
That said, there is a chance that this PSK-8 burst signal use a DSSS technique characterized by a base speed of 150 symbols/sec, a spread factor of 16 that makes a chip rate of 2400 cps and a PN sequence length of 96 bits.
by AngazU & Antonio
---
(1) from MIL-STD 188-181B.4.3 Phase vector rotation. The Dapper/Hill paper recommends that the direction of the phase vector rotation during a phase transition be implemented so that transitions to the 180° state occur by alternately rotating the phase in the clockwise and counter-clockwise directions from the 0° position. Rotation back to 0° is in the opposite direction from that most recently taken. In other words, the direction of rotation reverses upon reaching the 180° state, resulting in a change of the direction of phase vector rotation every other phase transition (see Figure B-2).
According to the paper, there is an offset of the carrier frequency equal to one-fourth the data rate if the phase is rotated in the same direction for each data bit transition.This means there is a 600-Hz offset when the data rate is 2.4 kbps. The reason given is that a nonzero average value disturbs the phase-error measurement of conventional demodulation techniques, which are unable to separate transitional information from phase-error measurement.
 |
| Figure B-2 |