Testing this I do see some more decodes. Good job Iztok.
But on my I5 4-core 8-proc 2..8Ghz it takes almost the full 15-second period to
finish decoding and it' s not using 100% CPU.
This needs to be multi-threaded if possible which seems like it should be
possible.
de Mike W9MDB
Here are some of the extra decodes pulled out
150530 -20 0.4 232 ~ CQ IV3GVS JN66 ta150530 -9 0.4
2797 ~ FG8OJ WB8FVB EL28 t150600 -20 0.4 232 ~ CQ
IV3GVS JN66 ta150715 -24 0.9 1156 ~ WZ8DX IZ4TNN -05
t150800 -17 0.4 1601 ~ CQ W9AV EN53
ta150800 -19 0.1 2041 ~ WD5COV S52D -03 t150845 4
0.3 709 ~ DL8SCG W4LOM EL99 t150845 -14 -0.2 1276 ~
LZ1ND K9AEG DM43 t150900 -13 0.4 1795 ~ SV3ALC PZ5RA -04
t150915 -19 0.4 1849 ~ PZ5RA F6DWX R-10
t150915 9 -0.4 2101 ~ SV1LK WD5COV -21 t151000 -13
0.3 2462 ~ LZ1LZ W1WAB -12 t151030 -18 0.4 2313 ~ CQ
OS8A JO20 ta151115 -12 0.6 494 ~ CQ WA1MTI FN41
t151130 -19 0.5 238 ~ EA4HBK IK2WRV JN45
t151145 -12 0.6 494 ~ CQ WA1MTI FN41 t151200 -15
1.1 1659 ~ CQ O82QTA KA28 ta151230 -15 0.4 2313 ~ CQ
OS8A JO20 ta
On Friday, February 1, 2019, 12:06:45 AM CST, Iztok Saje
<iztok.s...@telekom.si> wrote:
Hello!
Following some discussion on this forum, where I was too noisy, make me think.
When joining two uncorrelated FT8 frames (taken on exactly same frequency, but
not at the same time)
main problem is how to synchronize all tones?
My solution is to shift frames 7 times and do decoding after each shift, thus
every tone get
summed up in phase (or close enough).
CPU load is few times higher as with standard decoder, but my I7 CPU can do it
on time.
Testing in 1.9.1 gave me excellent results: I made several QSOs where plain FT8
decoder would fail.
Source code for WSJTX 2.0.0 with more detailed description is now tested a bit
and available at
http://lea.hamradio.si/~s52d/ft8div page.
You have to compile it yourself.
Time diversity (also called incremental redundancy) works by combining two
consecutive
even/odd frames, good to take repeated messages out of the QRM/QRN.
Space diversity (called stereo diversity by W8JI) works when twp copies of
WSJTX are run
with phase synchronuos RX like K3, IC-7610 etc. Bursts from two RX/antennas
are combined.
Data is exchanged using temporary files, so both WSJTX copies need separate
configuration file.
IMHO this approach is worth to be considered to be included in one of the
future releases.
Best 73, mni DX, CU
Iztok, S52D
[http://psn.sdn.si/ts/podpis-neo-4.jpg]
<https://www.neo.io/>
Pravni pogoji / Legal disclaimer
Telekom Slovenije, d.d., Ljubljana <http://www.telekom.si/disclaimer>
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
_______________________________________________
wsjt-devel mailing list
wsjt-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/wsjt-devel
