43 Commits

Author	SHA1	Message	Date
Steve Franke	59278b9b6b	Tweak ndeep determination.	2018-10-10 16:06:53 -05:00
Steve Franke	4b54873c81	Implement AP for all contest modes and also for Hound mode. AP is always active for Hounds. AP is automatically disabled if MyCall or DxCall are extended callsigns.	2018-10-04 17:48:55 -05:00
Steve Franke	0b3b050221	First cut at AP decoding for all values of ncontest.	2018-10-04 17:48:23 -05:00
Steve Franke	b5cc35e430	Cleaned up and simplified AP decoding. Enabled AP decoding for ncontest=0 and ncontest=4 (RU) so far. Missing MyCall or DxCall or nonstandard calls will disable AP passes that wouldn't make sense.	2018-10-04 17:45:15 -05:00
Steve Franke	77bb49e9ac	Beginnings of AP for RU. Not working yet.	2018-10-04 17:44:34 -05:00
Steve Franke	d9cf9bed3c	Better handling of decodes with unhashed callsigns. ... (cherry picked from commit 3b783a083e)	2018-09-28 17:07:28 -05:00
Steve Franke	13b5cc479b	Fix and re-enable polyfit SNR.	2018-09-26 08:29:01 -05:00
Steve Franke	669d54197d	Polynomial-fit SNR is still not right. Disable it again, for now.	2018-09-24 19:36:30 -05:00
Steve Franke	591de66a31	Fix up SNR calculation so that it reverts to SJF SNR when polyfit SNR is not sane.	2018-09-24 19:04:34 -05:00
Steve Franke	9075fef228	Revert to Steve's SNR estimator, for now.	2018-09-24 17:40:14 -05:00
Steve Franke	332a3f1561	Calibrate noise-baseline-fit based SNR estimate for FT8 v2 and print those estimates with decodes.	2018-09-23 13:03:20 -05:00
Steve Franke	642540ff8f	unpack77 now returns logical variable unpk77_success, which will be .false. if there is something obviously wrong with the message. Still to do: set the flag if the unpacked message contains CQ in an implausible place.	2018-09-12 15:06:26 -05:00
Joe Taylor	201b29993c	More fixes related to nonstandard calls.	2018-08-29 14:18:56 -04:00
Steve Franke	f7d423296f	Minor code cleanup.	2018-07-29 17:15:02 -05:00
Steve Franke	b67589d29f	Bring isync=1 message simulator, ft8sim, up to date. Fix ap decoding for isync=1 messages.	2018-07-29 16:29:11 -05:00
Steve Franke	df0b5e2622	Progress on AP decoding for 77bit messages. AP is basically working for regular i3=1 messages.	2018-07-28 17:32:45 -05:00
Steve Franke	bee1ecd109	Detect and reject messages with invalid i3 and or n3.	2018-07-28 08:50:50 -05:00
Joe Taylor	017629ac36	Suppress any "QU1RK" decodes, they are necessarily bad.	2018-07-27 10:43:58 -04:00
Steve Franke	59f713ba76	The change committed to decodedtext.cpp fixes a crash that occurred when processing certain isync=2 decodes. Is this the right fix? Other changes are minor code cleanups.	2018-07-17 17:48:23 -05:00
Joe Taylor	2c5dd053f3	Don't send "mygrid" where it's no longer needed.	2018-07-11 16:32:08 -04:00
Joe Taylor	c1e71a46ef	Remove more "bcontest" stuff from Fortrn code.	2018-07-10 15:32:59 -04:00
Steven Franke	e7b19b29b0	Screen out FT8 decodes with excessively large number of hard errors. These are almost certainly undetected errors.	2018-07-08 13:29:12 +00:00
Steve Franke	42f75bf404	Fix graymap.	2018-07-07 08:18:31 -05:00
Joe Taylor	e5e81e01fb	Make packjt77 into a module; move all its routines into file packjt77.f90.	2018-07-05 14:07:17 -04:00
Steve Franke	ed92d5a86c	SNR estimator is broken for isync=2 messages, due to different signal normalization. For now, go back to SJF's old signal/(noise+interference) estimator.	2018-07-04 16:58:24 -05:00
Steve Franke	a448dc5640	Use 37 character message for de-duping. Comment out some code that needs to be updated to work with new message type.	2018-07-04 16:20:12 -05:00
Steve Franke	df652e383b	More progress on making ft8sim2 and ft8b_2 encode and decode all message types. For now, decoder.f90 is hardwired to always print 37 characters.	2018-07-04 10:15:01 -05:00
Steve Franke	24f4792b23	Fix incorrect reference to extract77 - should have been unpack77	2018-07-03 15:59:24 -05:00
Steve Franke	57ebc30369	Call the new extract77 in ft8b_2. Define msg37 and msgsent37 to replace their 22-bit predecessors.	2018-07-03 15:34:39 -05:00
Steve Franke	c39dd63f79	Send (i3,n3) from mainwindow to genft8. First steps at interpreting (i3,n3) to send itype=1 for classic messages and itype=2 otherwise. Very buggy at this stage.	2018-07-03 09:28:16 -05:00
Steve Franke	5fc4fe6eda	Fix an off-by-one error. Remove attempt to refine DT estimate using polynomial interpolation.	2018-07-02 16:13:27 -05:00
Steve Franke	26de3219d3	Fix an off-by-one problem.	2018-06-30 14:56:19 -05:00
Steve Franke	83d81bdcd2	Streamline the sequence detector code.	2018-06-27 15:11:39 -05:00
Steve Franke	229ee09b99	Minor code cleanup.	2018-06-27 09:55:34 -05:00
Steve Franke	03abada09b	More progress on sequence demodulator. Can now do nsymb=1,2, and/or 3.	2018-06-26 11:11:58 -05:00
Steve Franke	7684bd0e78	Work toward a generalized sequence detector for nsymbols=1, 2, or 3. Works for nsymbol=3, but not yet for nsymbol=1, or 2.	2018-06-25 17:01:34 -05:00
Steve Franke	0f5759270e	Added 2-symbol sequence detection for proof-of-concept tests.	2018-06-22 11:20:28 -05:00
Steve Franke	3f8c218ebe	Make changes to ft8b_2.f90 to accommodate the new 77-bit in/out encoding/decoding routines. With these changes, WSJT-X decodes (174,91) frames.	2018-06-17 12:11:32 -05:00
Steve Franke	a5837bd5c7	AP decoding now works for type 2 frames. Also cleaned up and simplified AP implementation in ft8b_2.	2018-06-11 15:10:44 -05:00
Steve Franke	68beed88a7	Change to Gray bit to symbol mapping for (174,91) messages.	2018-06-08 16:21:21 -05:00
Steve Franke	a0db6a8a08	Disable AP passes for type 2 frames, for now. Write some diagnostics to lun 81.	2018-06-07 11:40:25 -05:00
Steve Franke	a23e2b6f8b	Fix a bug that caused some second-decoding-attempt log-likelihoods to become NaN.	2018-06-06 11:21:00 -05:00
Steve Franke	48b60eab71	Make two sync passes, one for each sync array. Label candidates with their type and decode accordingly.	2018-06-05 16:44:52 -05:00