[Lowfer] Opera over night

Garry Hess k3siw at sbcglobal.net
Sat Dec 21 08:19:54 EST 2013


Ran the deep-search opera32 code all night and it paid off since after 
WD2XES went QRT VO1NA showed up, and hung in there until near his 
sunrise. No TA but it was a noisy night with storms to the South and 
DK7FC was up on 474.2 kHz working WSPR. Tried to run regular opera to 
compare to deep search but CPU overload wouldn't allow it (hence the 
restart). Have the same Dell XP SP2 PC for a second setup and it runs 
both program together fine. Go figure.

sdr-iq1, eprobe2
2013-12-21 10:38:37 VO1NA   2836km 137554.999Hz   5mHz -41.9dBOp  93% 15.7dB
2013-12-21 10:03:38 VO1NA   2836km 137554.999Hz   5mHz -39.5dBOp  99% 18.4dB
2013-12-21 09:28:37 VO1NA   2836km 137555.000Hz   3mHz -40.5dBOp  93% 18.2dB
2013-12-21 08:18:37 VO1NA   2836km 137554.998Hz   3mHz -46.3dBOp  93% 17.2dB
2013-12-21 07:43:36 VO1NA   2836km 137554.999Hz   4mHz -45.6dBOp  99% 18.7dB
2013-12-21 07:08:35 VO1NA   2836km 137554.999Hz   3mHz -38.6dBOp  93% 19.1dB
2013-12-21 06:33:36 VO1NA   2836km 137554.999Hz   2mHz -45.4dBOp  99% 18.1dB
2013-12-21 05:58:35 VO1NA   2836km 137554.999Hz   2mHz -39.6dBOp  93% 19.5dB
2013-12-21 05:23:35 VO1NA   2836km 137554.998Hz   7mHz -44.3dBOp  99% 16.9dB
2013-12-21 04:48:34 VO1NA   2836km 137554.999Hz   4mHz -43.6dBOp  93% 17.7dB
2013-12-21 04:13:34 VO1NA   2836km 137554.999Hz   5mHz -48.1dBOp  98% 17.0dB
2013-12-21 03:38:34 VO1NA   2836km 137555.001Hz  18mHz -49.9dBOp  93% 15.2dB
2013-12-21 02:58:29 WD2XES  1365km 137514.999Hz   3mHz -28.9dBOp  63% 19.4dB
2013-12-21 02:25:33 WD2XES  1365km 137514.999Hz   8mHz -32.8dBOp 100% 20.6dB
2013-12-21 01:52:39 WD2XES  1365km 137514.998Hz   4mHz -34.7dBOp  65% 16.4dB
restart
2013-12-21 01:52:39 WD2XES  1365km 137514.999Hz   2mHz -36.0dBOp  96% 20.3dB
2013-12-21 01:19:43 WD2XES  1365km 137514.999Hz   3mHz -36.3dBOp  89% 19.6dB
2013-12-21 00:46:47 WD2XES  1365km 137514.998Hz   3mHz -30.3dBOp  98% 20.8dB
2013-12-21 00:13:51 WD2XES  1365km 137514.999Hz   2mHz -28.9dBOp  99% 21.1dB
2013-12-20 23:40:55 WD2XES  1365km 137514.998Hz   3mHz -30.2dBOp  90% 20.5dB
2013-12-20 23:07:57 WD2XES  1365km 137514.999Hz   5mHz -29.4dBOp  95% 21.3dB
2013-12-20 22:35:01 WD2XES  1365km 137514.999Hz   3mHz -29.6dBOp 100% 20.9dB

On the subject of how opds works, here are some comments that may be of 
interest by Markus, DF6NM, the developer:

Markus Vester
23 Feb
View Source
Andy,

yes I am using the second method. The predefined list of callsigns is 
converted to prestored Opera 01 sequences, based on the protocol 
description published by Guido PE1NNZ. For any 35 minute time slot, the 
received signal is correlated to each of the candidates, and the winning 
correlation peak is displayed if it exceeds an SNR threshold (14 dB). 
There is no attempt to decode anything, and the program will never come 
up with a callsign which was not in the list. In that sense it is really 
only a detection tool, but it does provide more than one bit due to 
unique identification (eg. 5 bits worth for 32 candidates), along with 
precise frequency and time information.

Experiments with simulated incoherent signals (ie. random phase for each 
dash) indicate that the correlation process in itself is about 6 dB more 
sensitive than the standard 28 bit decoding. This is more or less paid 
for by the reduced information content. For coherent transmissions, 
another 5 or 6 db are gained by being able to sum up the real complex 
amplitude rather than the power of all dashes. Loosely speaking, this 
can reduce the equivalent noise bandwidth from the inverse length of a 
single symbol (0.125 Hz), down to something proportional to the inverse 
of the whole sequence (0.5 mHz).

Best 73,
Markus (DF6NM)

Re: [rsgb_lf_group] Re: 0p-32 correlation results online

What I am not sure about, Markus...

In your software do you actually decode the data in some way, therefore 
using knowledge of the Opera signal coding OR....

Do you precode a callsign (your "precalculated templates from a limited 
list of callsigns") then correlate the received data against the pattern 
generated for that callsign.    If this latter is the case, then really 
you are decoding just one bit of infomration.  The presence or absence 
of a known call.

I suspect many readers hoping for an independent Opera are thinking the 
first option; but your description makes me think its the latter.  In 
which case, its certainly not a generalised data communication protocol, 
but a weak signal error corrected path checker.

Andy
G4JNT

-- 
73, Garry, K3SIW, EN52ta, Elgin, IL


More information about the Lowfer mailing list