30 September 2015

CISwitching...


September, 30 at 0711z on 15812.0 KHz on USB: on-the-fly switch from CIS-45 to CIS-60, both belonging to HDR (High Data Rate) family modem. Maybe tests or fast (auto) adatpion to the channel conditions ?

26 September 2015

a CIS-20 variant (37.5Bd, 60Hz)



the so-called CIS-20, in its standard format, is an OFDM 20-tone waveform characterised by π/4-DQPSK modulation at 75 symbols per second and 120Hz channel separation, as described here. CIS-20 is supposed to be the successor of CIS-12.
This variant exhibits the same modulation format but speed and channel separation are exactly the half of the ones of the standard version, i.e. 37.5Bd and 60Hz channel step. The signal bandwidth is unchanged (~2800Hz).

CIS-20 variant - OFDM parameters
CIS-20 single channel analysis

Note, once again, the differences between PSK-8 and π/4-DQPSK constellations (in absolute and relative mode, Diff=0 and Diff=1):

in a real PSK-8 the paths cross the origin of axis

23 September 2015

VEZHA-S (ВЕЖА-С), Russian Telegraph


This is a very old Russian telegraph system carrying information possibly for radio check of lines or for other (uknown) purposes. I could not find news about this signal and Russian monitors too do not have detailed informations about it; asking Karapuz and digging radioscanner forum I found the same signal than mine, and then its name "VEZHA-C" (ВЕЖА-С)[1].
The transmission was heard (and monitored for tens of minutes) at 0725z on 12165.0 KHz cf on September 22, and it did not change its format, sending the same "train" of bits and sometimes something like a "sync" signal . My measurements match the ones obtained by the Russian friends: baudrate 100Bd and 500 Hz frequency shift, frame is 162 bits and the the length of the codeword could be 6 bits. It's interesting to notice that - looking at the spectrum as at a Morse code -  the points are exactly 3 times shorter than the dashes.

22 September 2015

CIS-48 probe/marker signal


heard today on 17289.0 KHz on USB (on May 21 also, but on 17234.0 KHz there) from 1140z, this is the so-called 'CIS-48 probe/marker of channel' signal, according as defined here in radioscanner forum. 
The probes are sent every ~16 Sec - at least in this sample - in quite long sessions w/out any other signal during the monitored time (more than 1 hour) and consist of 4 x DPSK 50 baud modulated tones, located at 500 Hz, 1250 Hz, 1750 Hz, 2500 Hz and spaced by 750, 500 and 750 Hz, bandwidth is 2000 Hz. A short recording is available.




19 September 2015

about π/4 DQPSK modulation (here in CIS-60 30Bd)

looking deeper at CIS-60 signal reported in the previous post, the relative constellation changes in a symbol by symbol way but always exhibiting a 4-ary constellation although absolute constellation in each-1 mode looks like a PSK-8 modulation. This condition is called "π/4 DQPSK modulation" (Pic. 1).
Pic. 1
The π/4 DQPSK modulation format uses two QPSK constellations offset by 45 degrees (π/4 radians) and transitions occur from one constellation to the other, making the illusion of a PSK-8 modulation (Pic. 2). This guarantees that there is always a change in phase at each symbol, making clock recovery easier. The data is encoded in the magnitude and direction of the phase shift, not in the absolute position on the constellation. One advantage of π/4 DQPSK is that the signal trajectory does not pass through the origin, thus simplifying transmitter design.


Pic. 2 - we see PSK-8 in absolute constellation since the π/4 DQPSK modulation format uses two QPSK constellations offset by 45 degrees and transitions occur from one constellation to the other.
I want mean  that we see two 45 deg shifted QPSK (in relative) constellations:
- the one each 1, 3, 5,... symbols
- the other each 2,4,6,.. symbols
and this also influences the pattern that we see in absolute constellation.

In order to be safe about the main modulation (4 or 8-ary) used in this waveform, I highlighted a single channel from the CIS-60 signal (Pic. 3): if it was a clean QPSK or PSK-8 modulation then we should see three harmonics  in the fourth power while there are only two and this outstands a behavior of a π/4 DQPSK modulation (and OQPSK too). 
Now, looking at the phase plane of this channel, notice that the absolute constellation (Diff-0) exhibits 8-ary, BUT, as said above, there is no transition paths through the center (as in case of a clear PSK-8!) and that the relative constellation (Diff-1) looks like QPSK but 90 degrees rotated: this suggest the π/4 DQPSK modulation.
 
Pic. 3

By the way, the π/4 DQPSK modulation format is also used in CIS-45 v2 HDR modem (Pic. 4).

Pic. 4

The channel-by-channel analysis with the SA phase-plane method should be performed on all the channels, but it requires very good signals/recordings and a lot of time; anyway all the channels have been examintaed using SA OFDM module always getting the same  constellations shown in Pic. 2.

14 September 2015

MIL 188-110A/B/C 150bps, 66.6mS ACF

(a discussion about the MIL 188-110A/B/C ACF values)


This is a MIL 188-110A waveform: decoder k500 (and others) recognizes this signal just as a 188-110A and 150bps/S data-rate. The curious is that this signal doesn't exhibit the 200mS ACF value but a strong 66,67mS frame (160 symbols) and an equally strong  "superframe" of 200 mS length that meets the expected value (200mS). Me and Angazu tried to clrafify the reasons of this.

ACF of this signal (150 bps data-rate)

 Let's have a look at the 188-110 protocol and frame formations:

 

The reason is that the standard 188-110A/B/C has always a 200 mS ACF, but in case of low data-rate speeds (from 150 up to 1200 bps) four groups of the pairs probe+data count 160 symbols (4 x 40) and they are just in sync with the scrambler length (160 symbols) causing the strong 66.67mS ACF. 


66.6mS frame (note the 20+20 groups for low data-rate)
MIL 188-110A/B/C 150 bps

In case of 2400 bps (data and/or voice) we'll have groups of 48 symbols (32 symbols for data +16 symbols for probe) and this value is not correlated to the the scrambler length, so the "visible" ACF in this mode (and in 4800 uncoded) is the known 200 mS.
In case of the lowest speed (75bps) there are no channel probe so the 66.6mS ACF is just due to the scrambler length (160 symbols).

MIL 188-110A/B/C 600 bps
MIL 188-110A/B/C 2400 bps

 "So...we can state that ACF is 200 mS, but scrambler and group formation size generate the 66,6 mS ACF in user speeds of 1200Bd and lower" Angazu says.

A signal with these same features (modulation, speed and ACF) is also reported in radioscanner.ru site but named as USAF MIL 188-110A Compatible and nothing more precise.

11 September 2015

NVIS (and DX) links predictions

During these times of poor propagation, browsing the web you can find pretty useful and interactive tools to try to predict the probability and the reliability of NVIS (Near Vertical Incidence Skywave) single hop paths, representative of common tactical military HF communications scenarios, as well as of thousands of kilometers (DX) paths, more suitable for strategic back haul links scenarios. 

VOACAP Online can be used for NVIS and DX paths. The diagrams show for each hour the propability and reliability of links required by users:
http://www.voacap.com/prediction.html
 
VOACAP diagram related to Italy - Far East links

analysis for the requested link Italy-Swaziland

For who loves to listen exploiting the NVIS propagation, these URLs provide near real-time high-resolution maps of F2-layer critical frequencies (foF2) which correspond to the maximum radio frequency that can be reflected by the F2-region of the ionosphere at vertical incidence, that is when the signal is transmitted straight up into the ionosphere (NVIS): just choose the location that you are interested from a quite extensive list.


today foF2 statistics form Roquetes (Spain)
foF2 real-time plot from Brisbane (Australia)

If you are more interested in NVIS, well, this pdf document is a good starting point.

9 September 2015

a new Russian OFDM modem: 4-ary start-tones


this is a modded version of the CIS-OFDM3100, a (new) Russian modem/waveform reported here.This is the most recent recording of this signal that we have in our hands and it dates back to September 10, 2014 (just one year ago). The updates made to the signal are related to pre-tones, and more precisely their modulation: in the original version (recording of April 2014) the 5 pre-tones had a 2-ary modulation (DBPSK) and 50 Baud speed while this version exhibits a 4-ary modulation. Although the signal name, for what we know, be "Preliminary DQPSK", it's not easy to say which of 4-ary modulations we are facing (DQPSK vs OQPSK) because there are some unclear nuances about the manipulation speed in these tones .

The baudrate auto-measurement method of SA returns the plain standard value of 50Bd (Pic. 1) but we know that this feature does not always determines the correct baudrate so we looked at the absolute phase changes in order to receive a confirm or a new value for Br.
In Pic. 2 all changes are periodic and the 4 mSec changes (associated with 250 Baud speed) take place near the standard 20 mSec changes (associated with 50 Baud speed). Forcing these two values in the phase-module of SA I get two different 4-ary constellations:
- OQPSK @ 50 Baud (no diagonal transitions), as in Pic. 3
- DQPSK @ 250 Baud, as in PIc. 4

Assuming that "Preliminary DQPSK" be something like an official item, we should conclude for a DQPSK 250 Baud but, as said, not all shadows are gone.

Pic. 1 
Pic. 2
Pic. 3



Pic. 4
The above pictures are related to the pre-tone #4 but the findings also apply to the other pre-tones. Below (Pic. 5) another SA method to receive the baudrate of a signal by the full-wave detector: this method also confirms the 250 Baud.

Pic. 5

2 September 2015

a new Russian OFDM modem (say CIS-62/CIS-OFDM3100)

 CIS-62/CIS-OFDM3100
The wave file was sent to me by Karapuz and was recorded during the tests of this (new-entry ?) modem. Although the waveform does not fit the models that are built into SA software, I tried to determine the number of channels and their baudrate and the results are the same as those obtained by other friends. 
The official name for this waveform/modem, as well as its nickname, is unknown and CIS-62, or better "CIS-OFDM3100" since the no-CP assumpion (see later), are only my names just chosen in the wake of other similar identifications used by community.
As said, the recording has been sent by Karapuz and I never met this signal in the air neither logs or comments in the web about it. According to his experience this signal is quite rare in the air, he heard it this year too, but with a lower level and with modified preamble and trailer. Apparently, the tests are more distant from Karapuz QTH: perhaps performed in the far east part of Russia or beamed to North and/or Pacific area, but these are only speculative guess.


Assuming that the signal is an OFDM without CP, where channel separation is equal to the frequency manipulation (Br), the number of channels is given by the relationship: N = (BW/Br)-1. Then as first step, I began to calculate the baudrate (pictures 1 and 2) using two methods in order to receive a confirm:

Pic. 1
Pic. 2
Then the spread bandwidth (Pic. 3):

Pic. 3
Substituting the results, the number of channels is given by (3100/50) -1 = 62. 

Karapuz has got the same results in his analysis, but unfortunately he was not able to determine the used modulation. However the edge channels, 1 and 62, exhibit PSK-2 sequentially with the transfer of ones and zeroes, possibly for synchronization (Pic. 4):
 
Pic. 4

In his analysis Angazu used the tool CCF (Pic. 5), noting that at least in this recording it looks like an ARQ system with a 5.8 Sec ACK
(record starting is ACK) and, given the test purposes of the transmissions, it's very likely.

Pic. 5

All first pre-tones are BPSK modulated at 50Bd, speed manipulation has been calculated non only by the SA automatic mode (Pic. 6a and 6b), apparently to last for 78 symbols, as you can see by using SA (Pic. 7), they should carry some information useful for the tuning of the demodulator.

Pic. 6a

 Pic. 6b
Pic. 7
Taking the absolute phase changes of the pre-tones modulation, the result is very interesting, since they have the same group structure. Probably informations are different inside the chunks, but the absolute phase changes are constant in every tone. We do not know the reason.