Jared Boone
a7afc58cf8
More consistent use of pass-by-ref, const in baseband code.
2015-12-10 15:18:02 -08:00
Jared Boone
82f367dfea
Move code into BasebandThread.
2015-12-10 14:41:07 -08:00
Jared Boone
f369885062
Extract processor factory function.
...
A small code size reduction. I tried using std::unique_ptr, but code bloated between 150 and 400 bytes!
2015-12-10 13:21:27 -08:00
Jared Boone
ef315f0d92
Const/pass-by-reference a bunch more baseband stuff.
2015-12-10 12:36:12 -08:00
Jared Boone
3ed77af0c1
BasebandProcessor: Streamline construction of some messages.
2015-12-10 12:25:11 -08:00
Jared Boone
77aceaaf56
More BasebandProcessor pass by reference.
2015-12-10 12:17:40 -08:00
Jared Boone
14f24d8584
Move BasebandProcessor message.push() from dedicated methods to existing lambdas.
2015-12-10 12:04:43 -08:00
Jared Boone
98abc4bba8
Create baseband statistics messages on stack.
...
Code size shrank ~200 bytes with this change.
2015-12-10 11:57:57 -08:00
Jared Boone
3bed016128
BasebandProcessor interface change to pass references, not copies.
2015-12-10 11:56:16 -08:00
Jared Boone
eb1402764e
Move ERT packet type to its rightful place.
...
No longer mixed up with Message types.
2015-12-08 16:19:27 -08:00
Jared Boone
d0d97e92cb
Rename ::Packet to baseband::Packet. Remove IPC packet types.
2015-12-08 15:15:51 -08:00
Jared Boone
bd33e652ea
Abstract packet type/implementation details.
2015-12-07 12:35:05 -08:00
Jared Boone
2d0052fb14
Generalize ERT offset tracking a little.
2015-12-07 09:58:08 -08:00
Jared Boone
b9643dc988
Clean up ERT packet type notation.
...
Still not great, but at least repeated code is reduced.
2015-12-06 15:47:04 -08:00
Jared Boone
4205793b54
Clean up narrowing conversion warning.
2015-12-03 20:58:17 -08:00
Jared Boone
82ff45860e
Add basic ERT (OOK) utility meter reception.
...
Extracted the Manchester formatting function for use in ERT, now also returns an indication of bits in error.
2015-12-01 11:24:48 -08:00
Jared Boone
435824e9b5
Clean up buffer_t constructors.
2015-11-20 11:21:31 -08:00
Jared Boone
600a983d5c
Clean up complex<int16_t> -> complex<float> casting.
2015-11-20 11:01:41 -08:00
Jared Boone
85e984ed3b
Clean up complex<int8_t> -> complex<float> casting.
2015-11-19 12:42:35 -08:00
Jared Boone
060da5d227
Add ChannelDecimator decimate by 2 and no-shift options.
2015-11-19 12:24:20 -08:00
Jared Boone
ccd64f3cd9
Add function for c8->c16 decimate by two without fs/4 shift.
2015-11-19 12:20:20 -08:00
Jared Boone
8967f0eaa1
Move common matchers to PacketBuilder.
2015-11-16 10:33:50 -08:00
Jared Boone
1770dff522
Finish object-izing Baseband and RSSI threads.
2015-11-11 10:54:36 -08:00
Jared Boone
d9106b6f44
Start object-ization of RSSI thread.
2015-11-11 10:11:37 -08:00
Jared Boone
19c89def75
Start to object-ize the baseband thread function.
2015-11-11 10:02:06 -08:00
Jared Boone
2c64454172
More constructor style clean-up.
2015-11-11 09:35:28 -08:00
Jared Boone
d02698a6de
Apparent bug where min/max aren't initialized from value inside the buffer bounds.
...
The fix is less than ideal, it assumes that an incoming buffer length is always >= 1.
2015-11-11 09:30:15 -08:00
Jared Boone
ebf103363c
Constructor style clean-up.
2015-11-11 09:16:20 -08:00
Jared Boone
9f6c495fef
Add TPMS initial demodulator implementation.
...
Right now, 2FSK, 19200 baud, +/-38400Hz deviation. No effort is made to check CRC/checksums or decode packets.
2015-11-10 15:19:56 -08:00
Jared Boone
6cdef7026d
Remove spectrum window-presum due to lack of code RAM.
2015-11-10 15:16:27 -08:00
Jared Boone
99871dcb84
Move a few member variables to stack/constants.
2015-11-10 15:10:52 -08:00
Jared Boone
981c2fbfbd
Complete FSKProcessor's trip to being AISProcessor.
2015-11-10 15:04:38 -08:00
Jared Boone
111a5f10c0
Add some baseband constructors.
...
Some of them I think I took out earlier...
2015-11-10 14:04:50 -08:00
Jared Boone
add6171410
Move BasebandStatistics from heap to stack.
2015-11-10 13:59:17 -08:00
Jared Boone
0789d50fdf
Rework BitHistory, BitPattern, PacketBuilder.
...
Templatize PacketBuilder to optionally bit unstuff, and have flexible packet termination behavior.
2015-11-09 14:38:09 -08:00
Jared Boone
5dcca89172
FSKProcessor: remove separate channel filter.
...
RRC filter alone will do the trick!
2015-11-07 14:10:44 -08:00
Jared Boone
fc037c739a
Comment about correct MatchedFilter usage.
...
Because the filter expects a combined translate/rotate and low-pass filter in the taps, the taps must be a multiple of the translate sinusoid period.
2015-11-07 11:27:31 -08:00
Jared Boone
ae225e8c3b
Further MatchedFilter optimization using algebraic identities and stuff.
...
Can't stop premature optimization!!!
2015-11-06 09:40:05 -08:00
Jared Boone
100761481e
Fix inaccurate comment.
2015-11-06 09:40:05 -08:00
Jared Boone
3d5c8056d9
Premature optimization of MatchedFilter::shift_by_decimation_factor().
2015-11-06 09:40:05 -08:00
Jared Boone
4afcc83873
Mark MatchedFilter::is_new_decimation_cycle() as const.
2015-11-06 09:40:05 -08:00
Jared Boone
3c33e20156
Rename MatchedFilter::decimation_factor.
2015-11-06 09:40:05 -08:00
Jared Boone
7a59823211
Stop computing channel spectrum during FSK demod.
...
Data is unused, as screen is displaying decoded packets.
2015-11-06 09:40:05 -08:00
Jared Boone
ddd34793a6
Default constructors and configure methods for baseband classes.
2015-11-06 09:40:05 -08:00
Jared Boone
51df92a05a
Rename baseband_ais -> ais_baseband.
...
So that ais_baseband and ais_application filenames can be sorted near each other.
2015-11-06 09:40:04 -08:00
Jared Boone
26c23dc8a0
Use new FIR decimation in FSK demodulator.
...
Reduces FIR calculation effort when output samples are skipped at FIR filter, not after.
2015-11-06 09:40:04 -08:00
Jared Boone
86b068be62
Early calculation to reduce pressure on registers.
2015-11-06 09:40:04 -08:00
Jared Boone
a074e7d8d8
Remove commented code.
2015-11-06 09:40:04 -08:00
Jared Boone
424c0eac3f
FIRAndDecimateBy2Complex: expose decimation_factor, bring work function into class.
2015-11-06 09:40:04 -08:00
Jared Boone
cde15e4271
Generalize fir_and_decimate_by_2_complex_fast.
...
Permit decimation_factors other than 2.
Permit tap counts != 64 (but still must be multiple of 8).
Half the amount of tap memory required.
Performance is significantly degraded due to greater flexibility -- most likely due to separate sample buffer shift phase, instead of performing shift during output sample calculation.
2015-11-06 09:40:04 -08:00
