2016-05-01 17:51:56 +00:00
|
|
|
/* AMPS audio processing
|
|
|
|
*
|
|
|
|
* (C) 2016 by Andreas Eversberg <jolly@eversberg.eu>
|
|
|
|
* All Rights Reserved
|
|
|
|
*
|
|
|
|
* This program is free software: you can redistribute it and/or modify
|
|
|
|
* it under the terms of the GNU General Public License as published by
|
|
|
|
* the Free Software Foundation, either version 3 of the License, or
|
|
|
|
* (at your option) any later version.
|
|
|
|
*
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
|
|
|
* GNU General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
* along with this program. If not, see <http://www.gnu.org/licenses/>.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/* How does FSK decoding work:
|
|
|
|
* ---------------------------
|
|
|
|
*
|
|
|
|
* AMPS modulates the carrier frequency. If it is 8 kHz above, it is high level,
|
|
|
|
* if it is 8 kHz below, it is low level. The bits are coded using Manchester
|
|
|
|
* code. A 1 is coded by low level, followed by a hight level. A 0 is coded by
|
|
|
|
* a high level, followed by a low level. This will cause at least one level
|
|
|
|
* change within each bit. Also the level changes between equal bits, see
|
|
|
|
* Manchester coding. The bit rate is 10 KHz.
|
|
|
|
*
|
|
|
|
* In order to detect and demodulate a frame, the dotting sequnce is searched.
|
|
|
|
* The dotting sequnece are alternate bits: 101010101... The duration of a
|
|
|
|
* level change within the dotting sequnene ist 100uS. If all offsets of 8
|
|
|
|
* level changes lay within +-50% of the expected time, the dotting sequence is
|
|
|
|
* valid. Now the next 12 bits will be searched for sync sequnece. If better
|
|
|
|
* dotting-offsets are found, the counter for searching the sync sequence is
|
|
|
|
* reset, so the next 12 bits will be searched for sync too. If no sync was
|
|
|
|
* detected, the state changes to search for next dotting sequence.
|
|
|
|
*
|
|
|
|
* The average level change offsets of the dotting sequence is used to set the
|
|
|
|
* window for the first bit. When all samples for the window are received, a
|
|
|
|
* raise in level is detected as 1, fall in level is detected as 0. This is done
|
|
|
|
* by substracting the average sample value of the left side of the window by
|
|
|
|
* the average sample value of the right side. After the bit has been detected,
|
|
|
|
* the samples for the next window will be received and detected.
|
|
|
|
*
|
|
|
|
* As soon as a sync pattern is detected, the polarity of the pattern is used
|
|
|
|
* to decode the following frame bits with correct polarity. During reception
|
|
|
|
* of the frame bits, no sync and no dotting sequnece is searched or detected.
|
|
|
|
*
|
|
|
|
* After reception of the bit, the bits are re-assembled, parity checked and
|
|
|
|
* decoded. Then the process hunts for next dotting sequence.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <stdio.h>
|
|
|
|
#include <stdint.h>
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <string.h>
|
|
|
|
#include <errno.h>
|
|
|
|
#include <math.h>
|
|
|
|
#include "../common/debug.h"
|
|
|
|
#include "../common/timer.h"
|
|
|
|
#include "../common/call.h"
|
|
|
|
#include "../common/goertzel.h"
|
|
|
|
#include "amps.h"
|
|
|
|
#include "frame.h"
|
|
|
|
#include "dsp.h"
|
|
|
|
|
|
|
|
/* uncomment this to debug the encoding process */
|
|
|
|
//#define DEBUG_ENCODER
|
|
|
|
|
|
|
|
/* uncomment this to debug the decoding process */
|
|
|
|
//#define DEBUG_DECODER
|
|
|
|
|
|
|
|
#define PI M_PI
|
|
|
|
|
|
|
|
#define FSK_DEVIATION 32767.0 /* +-8 KHz */
|
|
|
|
#define SAT_DEVIATION 8192.0 /* +-2 KHz */
|
|
|
|
#define TX_AUDIO_0dBm0 45000 /* works quite well */
|
|
|
|
#define BITRATE 10000
|
|
|
|
#define SIG_TONE_CROSSINGS 2000 /* 2000 crossings are 100ms @ 10 KHz */
|
|
|
|
#define SIG_TONE_MINBITS 950 /* minimum bit durations to detect signalling tone (1000 is perfect for 100 ms) */
|
|
|
|
#define SIG_TONE_MAXBITS 1050 /* as above, maximum bits */
|
|
|
|
#define SAT_DURATION 0.100 /* duration of SAT signal measurement */
|
|
|
|
#define SAT_QUALITY 0.85 /* quality needed to detect sat */
|
2016-06-19 09:03:59 +00:00
|
|
|
#define SAT_DETECT_COUNT 3 /* number of measures to detect SAT signal (specs say 250ms) */
|
|
|
|
#define SAT_LOST_COUNT 3 /* number of measures to loose SAT signal (specs say 250ms) */
|
2016-05-01 17:51:56 +00:00
|
|
|
#define SIG_DETECT_COUNT 3 /* number of measures to detect Signalling Tone */
|
|
|
|
#define SIG_LOST_COUNT 2 /* number of measures to loose Signalling Tone */
|
|
|
|
#define CUT_OFF_HIGHPASS 300.0 /* cut off frequency for high pass filter to remove dc level from sound card / sample */
|
|
|
|
#define BEST_QUALITY 0.68 /* Best possible RX quality */
|
|
|
|
|
|
|
|
static int16_t ramp_up[256], ramp_down[256];
|
|
|
|
|
|
|
|
static double sat_freq[5] = {
|
|
|
|
5970.0,
|
|
|
|
6000.0,
|
|
|
|
6030.0,
|
|
|
|
5800.0, /* noise level to check against */
|
|
|
|
10000.0, /* signalling tone */
|
|
|
|
};
|
|
|
|
|
|
|
|
static int dsp_sine_sat[256];
|
|
|
|
|
|
|
|
/* global init for FSK */
|
|
|
|
void dsp_init(void)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
double s;
|
|
|
|
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "Generating sine table for SAT signal.\n");
|
|
|
|
for (i = 0; i < 256; i++) {
|
|
|
|
s = sin((double)i / 256.0 * 2.0 * PI);
|
|
|
|
dsp_sine_sat[i] = (int)(s * SAT_DEVIATION);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void dsp_init_ramp(amps_t *amps)
|
|
|
|
{
|
|
|
|
double c;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "Generating smooth ramp table.\n");
|
|
|
|
for (i = 0; i < 256; i++) {
|
|
|
|
c = cos((double)i / 256.0 * PI);
|
|
|
|
#if 0
|
|
|
|
if (c < 0)
|
|
|
|
c = -sqrt(-c);
|
|
|
|
else
|
|
|
|
c = sqrt(c);
|
|
|
|
#endif
|
|
|
|
ramp_down[i] = (int)(c * (double)amps->fsk_deviation);
|
|
|
|
ramp_up[i] = -ramp_down[i];
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void sat_reset(amps_t *amps, const char *reason);
|
|
|
|
|
|
|
|
/* Init FSK of transceiver */
|
|
|
|
int dsp_init_sender(amps_t *amps, int high_pass)
|
|
|
|
{
|
|
|
|
double coeff;
|
|
|
|
int16_t *spl;
|
|
|
|
int i;
|
|
|
|
int rc;
|
|
|
|
double RC, dt;
|
|
|
|
|
|
|
|
/* attack (3ms) and recovery time (13.5ms) according to amps specs */
|
|
|
|
init_compander(&s->cstate, 8000, 3.0, 13.5, TX_AUDIO_0dBm0);
|
|
|
|
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "Init DSP for transceiver.\n");
|
|
|
|
|
|
|
|
if (amps->sender.samplerate < 96000) {
|
|
|
|
PDEBUG(DDSP, DEBUG_ERROR, "Sample rate must be at least 96000 Hz to process FSK and SAT signals.\n");
|
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
amps->fsk_bitduration = (double)amps->sender.samplerate / (double)BITRATE;
|
|
|
|
amps->fsk_bitstep = 1.0 / amps->fsk_bitduration;
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "Use %.4f samples for full bit duration @ %d.\n", amps->fsk_bitduration, amps->sender.samplerate);
|
|
|
|
|
2016-06-20 15:16:46 +00:00
|
|
|
amps->fsk_tx_buffer_size = amps->fsk_bitduration + 10; /* 10 extra to avoid overflow due to rounding */
|
2016-05-01 17:51:56 +00:00
|
|
|
amps->fsk_tx_buffer = calloc(sizeof(int16_t), amps->fsk_tx_buffer_size);
|
|
|
|
if (!amps->fsk_tx_buffer) {
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "No memory!\n");
|
|
|
|
rc = -ENOMEM;
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
amps->fsk_rx_buffer_length = ceil(amps->fsk_bitduration); /* buffer holds one bit (rounded up) */
|
|
|
|
amps->fsk_rx_buffer = calloc(sizeof(int16_t), amps->fsk_rx_buffer_length);
|
|
|
|
if (!amps->fsk_rx_buffer) {
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "No memory!\n");
|
|
|
|
rc = -ENOMEM;
|
|
|
|
goto error;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* create devation and ramp */
|
|
|
|
amps->fsk_deviation = FSK_DEVIATION; /* be sure not to overflow 32767 */
|
|
|
|
dsp_init_ramp(amps);
|
|
|
|
|
|
|
|
/* allocate ring buffer for SAT signal detection */
|
|
|
|
amps->sat_samples = (int)((double)amps->sender.samplerate * SAT_DURATION + 0.5);
|
|
|
|
spl = calloc(1, amps->sat_samples * sizeof(*spl));
|
|
|
|
if (!spl) {
|
|
|
|
PDEBUG(DDSP, DEBUG_ERROR, "No memory!\n");
|
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
amps->sat_filter_spl = spl;
|
|
|
|
|
|
|
|
/* count SAT tones */
|
|
|
|
for (i = 0; i < 5; i++) {
|
|
|
|
coeff = 2.0 * cos(2.0 * PI * sat_freq[i] / (double)amps->sender.samplerate);
|
|
|
|
amps->sat_coeff[i] = coeff * 32768.0;
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "sat_coeff[%d] = %d\n", i, (int)amps->sat_coeff[i]);
|
|
|
|
|
|
|
|
if (i < 3) {
|
|
|
|
amps->sat_phaseshift256[i] = 256.0 / ((double)amps->sender.samplerate / sat_freq[i]);
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "sat_phaseshift256[%d] = %.4f\n", i, amps->sat_phaseshift256[i]);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
sat_reset(amps, "Initial state");
|
|
|
|
|
|
|
|
/* use this filter to remove dc level for 0-crossing detection
|
|
|
|
* if we have de-emphasis, we don't need it, so high_pass is not set. */
|
|
|
|
if (high_pass) {
|
|
|
|
RC = 1.0 / (CUT_OFF_HIGHPASS * 2.0 *3.14);
|
|
|
|
dt = 1.0 / amps->sender.samplerate;
|
|
|
|
amps->highpass_factor = RC / (RC + dt);
|
|
|
|
}
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
error:
|
|
|
|
dsp_cleanup_sender(amps);
|
|
|
|
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Cleanup transceiver instance. */
|
|
|
|
void dsp_cleanup_sender(amps_t *amps)
|
|
|
|
{
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "Cleanup DSP for treansceiver.\n");
|
|
|
|
|
|
|
|
if (amps->fsk_tx_buffer)
|
|
|
|
free(amps->fsk_tx_buffer);
|
|
|
|
if (amps->fsk_rx_buffer)
|
|
|
|
free(amps->fsk_rx_buffer);
|
|
|
|
if (amps->sat_filter_spl) {
|
|
|
|
free(amps->sat_filter_spl);
|
|
|
|
amps->sat_filter_spl = NULL;
|
|
|
|
}
|
|
|
|
#if 0
|
|
|
|
if (amps->frame_spl) {
|
|
|
|
free(amps->frame_spl);
|
|
|
|
amps->frame_spl = NULL;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
|
2016-06-20 15:16:46 +00:00
|
|
|
static int fsk_encode(amps_t *amps, char bit)
|
2016-05-01 17:51:56 +00:00
|
|
|
{
|
|
|
|
int16_t *spl;
|
|
|
|
double phase, bitstep, deviation;
|
|
|
|
int count;
|
|
|
|
char last;
|
|
|
|
|
|
|
|
deviation = amps->fsk_deviation;
|
|
|
|
spl = amps->fsk_tx_buffer;
|
|
|
|
phase = amps->fsk_tx_phase;
|
|
|
|
last = amps->fsk_tx_last_bit;
|
|
|
|
bitstep = amps->fsk_bitstep * 256.0 * 2.0; /* half bit ramp */
|
|
|
|
|
2016-06-20 15:16:46 +00:00
|
|
|
//printf("%d %d\n", (bit) & 1, last & 1);
|
|
|
|
if ((bit & 1)) {
|
|
|
|
if ((last & 1)) {
|
|
|
|
/* last bit was 1, this bit is 1, so we ramp down first */
|
2016-05-01 17:51:56 +00:00
|
|
|
do {
|
2016-06-20 15:16:46 +00:00
|
|
|
*spl++ = ramp_down[(int)phase];
|
2016-05-01 17:51:56 +00:00
|
|
|
phase += bitstep;
|
|
|
|
} while (phase < 256.0);
|
|
|
|
phase -= 256.0;
|
|
|
|
} else {
|
2016-06-20 15:16:46 +00:00
|
|
|
/* last bit was 0, this bit is 1, so we stay down first */
|
2016-05-01 17:51:56 +00:00
|
|
|
do {
|
2016-06-20 15:16:46 +00:00
|
|
|
*spl++ = -deviation;
|
2016-05-01 17:51:56 +00:00
|
|
|
phase += bitstep;
|
|
|
|
} while (phase < 256.0);
|
|
|
|
phase -= 256.0;
|
|
|
|
}
|
2016-06-20 15:16:46 +00:00
|
|
|
/* ramp up */
|
|
|
|
do {
|
|
|
|
*spl++ = ramp_up[(int)phase];
|
|
|
|
phase += bitstep;
|
|
|
|
} while (phase < 256.0);
|
|
|
|
phase -= 256.0;
|
|
|
|
} else {
|
|
|
|
if ((last & 1)) {
|
|
|
|
/* last bit was 1, this bit is 0, so we stay up first */
|
|
|
|
do {
|
|
|
|
*spl++ = deviation;
|
|
|
|
phase += bitstep;
|
|
|
|
} while (phase < 256.0);
|
|
|
|
phase -= 256.0;
|
|
|
|
} else {
|
|
|
|
/* last bit was 0, this bit is 0, so we ramp up first */
|
|
|
|
do {
|
|
|
|
*spl++ = ramp_up[(int)phase];
|
|
|
|
phase += bitstep;
|
|
|
|
} while (phase < 256.0);
|
|
|
|
phase -= 256.0;
|
|
|
|
}
|
|
|
|
/* ramp up */
|
|
|
|
do {
|
|
|
|
*spl++ = ramp_down[(int)phase];
|
|
|
|
phase += bitstep;
|
|
|
|
} while (phase < 256.0);
|
|
|
|
phase -= 256.0;
|
|
|
|
}
|
|
|
|
last = bit;
|
2016-05-01 17:51:56 +00:00
|
|
|
/* depending on the number of samples, return the number */
|
|
|
|
count = ((uintptr_t)spl - (uintptr_t)amps->fsk_tx_buffer) / sizeof(*spl);
|
|
|
|
|
|
|
|
amps->fsk_tx_last_bit = last;
|
|
|
|
amps->fsk_tx_phase = phase;
|
|
|
|
amps->fsk_tx_buffer_length = count;
|
|
|
|
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
|
|
|
int fsk_frame(amps_t *amps, int16_t *samples, int length)
|
|
|
|
{
|
2016-06-20 15:16:46 +00:00
|
|
|
int count = 0, len, pos, copy, i;
|
2016-05-01 17:51:56 +00:00
|
|
|
int16_t *spl;
|
2016-06-20 15:16:46 +00:00
|
|
|
int rc;
|
|
|
|
char c;
|
|
|
|
|
|
|
|
len = amps->fsk_tx_buffer_length;
|
|
|
|
pos = amps->fsk_tx_buffer_pos;
|
|
|
|
spl = amps->fsk_tx_buffer;
|
2016-05-01 17:51:56 +00:00
|
|
|
|
|
|
|
again:
|
|
|
|
/* there must be length, otherwise we would skip blocks */
|
|
|
|
if (count == length)
|
2016-06-20 15:16:46 +00:00
|
|
|
goto done;
|
2016-05-01 17:51:56 +00:00
|
|
|
|
2016-06-20 15:16:46 +00:00
|
|
|
/* start of new bit, so generate buffer for one bit */
|
2016-05-01 17:51:56 +00:00
|
|
|
if (pos == 0) {
|
2016-06-20 15:16:46 +00:00
|
|
|
c = amps->fsk_tx_frame[amps->fsk_tx_frame_pos];
|
|
|
|
/* start new frame, so we generate one */
|
|
|
|
if (c == '\0') {
|
|
|
|
if (amps->dsp_mode == DSP_MODE_AUDIO_RX_FRAME_TX)
|
|
|
|
rc = amps_encode_frame_fvc(amps, amps->fsk_tx_frame);
|
|
|
|
else
|
|
|
|
rc = amps_encode_frame_focc(amps, amps->fsk_tx_frame);
|
|
|
|
/* check if we have not bit string (change to tx audio)
|
|
|
|
* we may not store fsk_tx_buffer_pos, because is was reset on a mode achange */
|
|
|
|
if (rc)
|
|
|
|
return count;
|
|
|
|
amps->fsk_tx_frame_pos = 0;
|
|
|
|
c = amps->fsk_tx_frame[0];
|
|
|
|
}
|
|
|
|
if (c == 'i')
|
|
|
|
c = (amps->channel_busy) ? '0' : '1';
|
|
|
|
len = fsk_encode(amps, c);
|
|
|
|
amps->fsk_tx_frame_pos++;
|
2016-05-01 17:51:56 +00:00
|
|
|
}
|
|
|
|
|
2016-06-20 15:16:46 +00:00
|
|
|
copy = len - pos;
|
2016-05-01 17:51:56 +00:00
|
|
|
if (length - count < copy)
|
|
|
|
copy = length - count;
|
|
|
|
//printf("pos=%d length=%d copy=%d\n", pos, length, copy);
|
|
|
|
for (i = 0; i < copy; i++) {
|
|
|
|
#ifdef DEBUG_ENCODER
|
2016-06-20 15:16:46 +00:00
|
|
|
puts(debug_amplitude((double)spl[pos] / 32767.0));
|
2016-05-01 17:51:56 +00:00
|
|
|
#endif
|
2016-06-20 15:16:46 +00:00
|
|
|
*samples++ = spl[pos++];
|
2016-05-01 17:51:56 +00:00
|
|
|
}
|
|
|
|
count += copy;
|
2016-06-20 15:16:46 +00:00
|
|
|
if (pos == len) {
|
|
|
|
pos = 0;
|
2016-05-01 17:51:56 +00:00
|
|
|
goto again;
|
|
|
|
}
|
|
|
|
|
2016-06-20 15:16:46 +00:00
|
|
|
done:
|
|
|
|
amps->fsk_tx_buffer_length = len;
|
2016-05-01 17:51:56 +00:00
|
|
|
amps->fsk_tx_buffer_pos = pos;
|
|
|
|
|
|
|
|
return count;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Generate audio stream with SAT signal. Keep phase for next call of function. */
|
|
|
|
static void sat_encode(amps_t *amps, int16_t *samples, int length)
|
|
|
|
{
|
|
|
|
double phaseshift, phase;
|
|
|
|
int32_t sample;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
phaseshift = amps->sat_phaseshift256[amps->sat];
|
|
|
|
phase = amps->sat_phase256;
|
|
|
|
|
|
|
|
for (i = 0; i < length; i++) {
|
|
|
|
sample = *samples;
|
|
|
|
sample += dsp_sine_sat[((uint8_t)phase) & 0xff];
|
|
|
|
if (sample > 32767)
|
|
|
|
sample = 32767;
|
|
|
|
else if (sample < -32767)
|
|
|
|
sample = -32767;
|
|
|
|
*samples++ = sample;
|
|
|
|
phase += phaseshift;
|
|
|
|
if (phase >= 256)
|
|
|
|
phase -= 256;
|
|
|
|
}
|
|
|
|
|
|
|
|
amps->sat_phase256 = phase;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Provide stream of audio toward radio unit */
|
|
|
|
void sender_send(sender_t *sender, int16_t *samples, int length)
|
|
|
|
{
|
|
|
|
amps_t *amps = (amps_t *) sender;
|
|
|
|
int count;
|
|
|
|
|
|
|
|
again:
|
|
|
|
switch (amps->dsp_mode) {
|
|
|
|
case DSP_MODE_OFF:
|
|
|
|
/* silence, if transmitter is off */
|
|
|
|
memset(samples, 0, length * sizeof(*samples));
|
|
|
|
break;
|
|
|
|
case DSP_MODE_AUDIO_RX_AUDIO_TX:
|
|
|
|
jitter_load(&s->sender.audio, samples, length);
|
|
|
|
/* pre-emphasis */
|
|
|
|
if (amps->pre_emphasis)
|
|
|
|
pre_emphasis(&s->estate, samples, length);
|
|
|
|
/* encode sat */
|
|
|
|
sat_encode(amps, samples, length);
|
|
|
|
break;
|
|
|
|
case DSP_MODE_AUDIO_RX_FRAME_TX:
|
|
|
|
case DSP_MODE_FRAME_RX_FRAME_TX:
|
|
|
|
/* Encode frame into audio stream. If frames have
|
|
|
|
* stopped, process again for rest of stream. */
|
|
|
|
count = fsk_frame(amps, samples, length);
|
|
|
|
samples += count;
|
|
|
|
length -= count;
|
2016-06-20 15:16:46 +00:00
|
|
|
if (length)
|
|
|
|
goto again;
|
2016-05-01 17:51:56 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void fsk_rx_bit(amps_t *amps, int16_t *spl, int len, int pos)
|
|
|
|
{
|
|
|
|
int i, ii;
|
|
|
|
int32_t first, second;
|
|
|
|
int bit;
|
|
|
|
int32_t max = -32768, min = 32767;
|
|
|
|
|
|
|
|
/* decode one bit. substact the first half from the second half.
|
|
|
|
* the result shows the direction of the bit change: 1 == positive.
|
|
|
|
*/
|
|
|
|
ii = len >> 1;
|
|
|
|
second = first = 0;
|
|
|
|
for (i = 0; i < ii; i++) {
|
|
|
|
if (--pos < 0)
|
|
|
|
pos = len - 1;
|
|
|
|
//printf("second %d: %d\n", pos, spl[pos]);
|
|
|
|
second += spl[pos];
|
|
|
|
if (spl[pos] > max)
|
|
|
|
max = spl[pos];
|
|
|
|
if (spl[pos] < min)
|
|
|
|
min = spl[pos];
|
|
|
|
}
|
|
|
|
second /= ii;
|
|
|
|
for (i = 0; i < ii; i++) {
|
|
|
|
if (--pos < 0)
|
|
|
|
pos = len - 1;
|
|
|
|
//printf("first %d: %d\n", pos, spl[pos]);
|
|
|
|
first += spl[pos];
|
|
|
|
if (spl[pos] > max)
|
|
|
|
max = spl[pos];
|
|
|
|
if (spl[pos] < min)
|
|
|
|
min = spl[pos];
|
|
|
|
}
|
|
|
|
first /= ii;
|
|
|
|
//printf("first = %d second = %d\n", first, second);
|
|
|
|
/* get bit */
|
|
|
|
if (second > first)
|
|
|
|
bit = 1;
|
|
|
|
else
|
|
|
|
bit = 0;
|
|
|
|
#ifdef DEBUG_DECODER
|
|
|
|
if (amps->fsk_rx_sync != FSK_SYNC_POSITIVE && amps->fsk_rx_sync != FSK_SYNC_NEGATIVE)
|
|
|
|
printf("Decoded bit as %d (dotting life = %d)\n", bit, amps->fsk_rx_dotting_life);
|
|
|
|
else
|
|
|
|
printf("Decoded bit as %d\n", bit);
|
|
|
|
#endif
|
|
|
|
|
|
|
|
if (amps->fsk_rx_sync != FSK_SYNC_POSITIVE && amps->fsk_rx_sync != FSK_SYNC_NEGATIVE) {
|
|
|
|
amps->fsk_rx_sync_register = (amps->fsk_rx_sync_register << 1) | bit;
|
|
|
|
/* check if we received a sync */
|
|
|
|
if ((amps->fsk_rx_sync_register & 0x7ff) == 0x712) {
|
|
|
|
#ifdef DEBUG_DECODER
|
|
|
|
printf("Sync word detected (positive)\n");
|
|
|
|
#endif
|
|
|
|
amps->fsk_rx_sync = FSK_SYNC_POSITIVE;
|
|
|
|
prepare_frame:
|
|
|
|
amps->fsk_rx_frame_count = 0;
|
|
|
|
amps->fsk_rx_frame_quality = 0.0;
|
|
|
|
amps->fsk_rx_frame_level = 0.0;
|
|
|
|
amps->fsk_rx_sync_register = 0x555;
|
2016-06-20 15:15:42 +00:00
|
|
|
amps->when_received = get_time() - (21.0 / (double)BITRATE);
|
2016-05-01 17:51:56 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
if ((amps->fsk_rx_sync_register & 0x7ff) == 0x0ed) {
|
|
|
|
#ifdef DEBUG_DECODER
|
|
|
|
printf("Sync word detected (negative)\n");
|
|
|
|
#endif
|
|
|
|
amps->fsk_rx_sync = FSK_SYNC_NEGATIVE;
|
|
|
|
goto prepare_frame;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
/* if no sync, count down the dotting life counter */
|
|
|
|
if (--amps->fsk_rx_dotting_life == 0) {
|
|
|
|
#ifdef DEBUG_DECODER
|
|
|
|
printf("No Sync detected after dotting\n");
|
|
|
|
#endif
|
|
|
|
amps->fsk_rx_sync = FSK_SYNC_NONE;
|
2016-06-20 15:16:46 +00:00
|
|
|
amps->channel_busy = 0;
|
2016-05-01 17:51:56 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* count level and quality */
|
|
|
|
amps->fsk_rx_frame_level += (double)(max - min) / (double)FSK_DEVIATION / 2.0;
|
|
|
|
if (bit)
|
|
|
|
amps->fsk_rx_frame_quality += (double)(second - first) / (double)FSK_DEVIATION / 2.0 / BEST_QUALITY;
|
|
|
|
else
|
|
|
|
amps->fsk_rx_frame_quality += (double)(first - second) / (double)FSK_DEVIATION / 2.0 / BEST_QUALITY;
|
|
|
|
|
|
|
|
/* invert bit if negative sync was detected */
|
|
|
|
if (amps->fsk_rx_sync == FSK_SYNC_NEGATIVE)
|
|
|
|
bit = 1 - bit;
|
|
|
|
|
|
|
|
/* read next bit. after all bits, we reset to FSK_SYNC_NONE */
|
|
|
|
amps->fsk_rx_frame[amps->fsk_rx_frame_count++] = bit + '0';
|
|
|
|
if (amps->fsk_rx_frame_count > FSK_MAX_BITS) {
|
|
|
|
fprintf(stderr, "our fsk_tx_count (%d) is larger than our max bits we can handle, please fix!\n", amps->fsk_rx_frame_count);
|
|
|
|
abort();
|
|
|
|
}
|
|
|
|
if (amps->fsk_rx_frame_count == amps->fsk_rx_frame_length) {
|
|
|
|
int more;
|
|
|
|
|
|
|
|
/* a complete frame was received, so we process it */
|
|
|
|
amps->fsk_rx_frame[amps->fsk_rx_frame_count] = '\0';
|
|
|
|
more = amps_decode_frame(amps, amps->fsk_rx_frame, amps->fsk_rx_frame_count, amps->fsk_rx_frame_level / (double)amps->fsk_rx_frame_count, amps->fsk_rx_frame_quality / amps->fsk_rx_frame_level, (amps->fsk_rx_sync == FSK_SYNC_NEGATIVE));
|
|
|
|
if (more) {
|
|
|
|
/* switch to next worda length without DCC included */
|
|
|
|
amps->fsk_rx_frame_length = 240;
|
|
|
|
goto prepare_frame;
|
|
|
|
} else {
|
|
|
|
/* switch back to first word length with DCC included */
|
|
|
|
if (amps->fsk_rx_frame_length == 240)
|
|
|
|
amps->fsk_rx_frame_length = 247;
|
|
|
|
amps->fsk_rx_sync = FSK_SYNC_NONE;
|
2016-06-20 15:16:46 +00:00
|
|
|
amps->channel_busy = 0;
|
2016-05-01 17:51:56 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void fsk_rx_dotting(amps_t *amps, double _elapsed, int dir)
|
|
|
|
{
|
|
|
|
uint8_t pos = amps->fsk_rx_dotting_pos++;
|
|
|
|
double average, elapsed, offset;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
#ifdef DEBUG_DECODER
|
|
|
|
printf("Level change detected\n");
|
|
|
|
#endif
|
|
|
|
/* store into dotting list */
|
|
|
|
amps->fsk_rx_dotting_elapsed[pos++] = _elapsed;
|
|
|
|
|
|
|
|
/* check quality of dotting sequence.
|
|
|
|
* in case this is not a dotting sequence, noise or speech, the quality
|
|
|
|
* should be bad.
|
|
|
|
* count (only) 7 'elapsed' values between 8 zero-crossings.
|
|
|
|
* calculate the average relative to the current position.
|
|
|
|
*/
|
|
|
|
average = 0.0;
|
|
|
|
elapsed = 0.0;
|
|
|
|
for (i = 1; i < 8; i++) {
|
|
|
|
elapsed += amps->fsk_rx_dotting_elapsed[--pos];
|
|
|
|
offset = elapsed - (double)i;
|
|
|
|
if (offset >= 0.5 || offset <= -0.5) {
|
|
|
|
#ifdef DEBUG_DECODER
|
|
|
|
// printf("offset %.3f (last but %d) not within -0.5 .. 0.5 bit position, detecting no dotting.\n", offset, i - 1);
|
|
|
|
#endif
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
average += offset;
|
|
|
|
}
|
|
|
|
average /= (double)i;
|
|
|
|
|
|
|
|
amps->fsk_rx_dotting_life = 12;
|
|
|
|
|
|
|
|
/* if we are already found dotting, we detect better dotting.
|
|
|
|
* this happens, if dotting was falsely detected due to noise.
|
|
|
|
* then the real dotting causes a reastart of hunting for sync sequence.
|
|
|
|
*/
|
|
|
|
if (amps->fsk_rx_sync == FSK_SYNC_NONE || fabs(average) < amps->fsk_rx_dotting_average) {
|
|
|
|
#ifdef DEBUG_DECODER
|
|
|
|
printf("Found (better) dotting sequence (average = %.3f)\n", average);
|
|
|
|
#endif
|
|
|
|
amps->fsk_rx_sync = FSK_SYNC_DOTTING;
|
|
|
|
amps->fsk_rx_dotting_average = fabs(average);
|
|
|
|
amps->fsk_rx_bitcount = 0.5 + average;
|
2016-06-20 15:16:46 +00:00
|
|
|
if (amps->si.acc_type.bis)
|
|
|
|
amps->channel_busy = 1;
|
2016-05-01 17:51:56 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* decode frame */
|
|
|
|
void sender_receive_frame(amps_t *amps, int16_t *samples, int length)
|
|
|
|
{
|
|
|
|
int16_t *spl, last_sample;
|
|
|
|
int len, pos;
|
|
|
|
double bitstep, elapsed;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
bitstep = amps->fsk_bitstep;
|
|
|
|
spl = amps->fsk_rx_buffer;
|
|
|
|
pos = amps->fsk_rx_buffer_pos;
|
|
|
|
len = amps->fsk_rx_buffer_length;
|
|
|
|
last_sample = amps->fsk_rx_last_sample;
|
|
|
|
elapsed = amps->fsk_rx_elapsed;
|
|
|
|
|
|
|
|
for (i = 0; i < length; i++) {
|
|
|
|
#ifdef DEBUG_DECODER
|
|
|
|
puts(debug_amplitude((double)samples[i] / (double)FSK_DEVIATION));
|
|
|
|
#endif
|
|
|
|
/* push sample to detection window and shift */
|
|
|
|
spl[pos++] = samples[i];
|
|
|
|
if (pos == len)
|
|
|
|
pos = 0;
|
|
|
|
if (amps->fsk_rx_sync != FSK_SYNC_POSITIVE && amps->fsk_rx_sync != FSK_SYNC_NEGATIVE) {
|
|
|
|
/* check for change in polarity */
|
|
|
|
if (last_sample <= 0) {
|
|
|
|
if (samples[i] > 0) {
|
|
|
|
fsk_rx_dotting(amps, elapsed, 1);
|
|
|
|
elapsed = 0.0;
|
|
|
|
}
|
|
|
|
} else {
|
|
|
|
if (samples[i] <= 0) {
|
|
|
|
fsk_rx_dotting(amps, elapsed, 0);
|
|
|
|
elapsed = 0.0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
last_sample = samples[i];
|
|
|
|
elapsed += bitstep;
|
|
|
|
// printf("%.4f\n", bitcount);
|
|
|
|
if (amps->fsk_rx_sync != FSK_SYNC_NONE) {
|
|
|
|
amps->fsk_rx_bitcount += bitstep;
|
|
|
|
if (amps->fsk_rx_bitcount >= 1.0) {
|
|
|
|
amps->fsk_rx_bitcount -= 1.0;
|
|
|
|
fsk_rx_bit(amps, spl, len, pos);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
amps->fsk_rx_last_sample = last_sample;
|
|
|
|
amps->fsk_rx_elapsed = elapsed;
|
|
|
|
amps->fsk_rx_buffer_pos = pos;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
/* decode signalling tone */
|
|
|
|
/* compare supervisory signal against noise floor on 5800 Hz */
|
|
|
|
static void sat_decode(amps_t *amps, int16_t *samples, int length)
|
|
|
|
{
|
|
|
|
int coeff[3];
|
|
|
|
double result[3], quality[2];
|
|
|
|
|
|
|
|
coeff[0] = amps->sat_coeff[amps->sat];
|
|
|
|
coeff[1] = amps->sat_coeff[3]; /* noise floor detection */
|
|
|
|
coeff[2] = amps->sat_coeff[4]; /* signalling tone */
|
|
|
|
audio_goertzel(samples, length, 0, coeff, result, 3);
|
|
|
|
|
|
|
|
quality[0] = (result[0] - result[1]) / result[0];
|
|
|
|
if (quality[0] < 0)
|
|
|
|
quality[0] = 0;
|
|
|
|
quality[1] = (result[2] - result[1]) / result[2];
|
|
|
|
if (quality[1] < 0)
|
|
|
|
quality[1] = 0;
|
|
|
|
|
|
|
|
PDEBUG(DDSP, DEBUG_NOTICE, "SAT level %.2f%% quality %.0f%%\n", result[0] * 32767.0 / SAT_DEVIATION / 0.63662 * 100.0, quality[0] * 100.0);
|
|
|
|
if (amps->sender.loopback || debuglevel == DEBUG_DEBUG) {
|
|
|
|
PDEBUG(DDSP, debuglevel, "Signalling Tone level %.2f%% quality %.0f%%\n", result[2] * 32767.0 / FSK_DEVIATION / 0.63662 * 100.0, quality[1] * 100.0);
|
|
|
|
}
|
|
|
|
if (quality[0] > SAT_QUALITY) {
|
|
|
|
if (amps->sat_detected == 0) {
|
|
|
|
amps->sat_detect_count++;
|
|
|
|
if (amps->sat_detect_count == SAT_DETECT_COUNT) {
|
|
|
|
amps->sat_detected = 1;
|
|
|
|
amps->sat_detect_count = 0;
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "SAT signal detected with level=%.0f%%, quality=%.0f%%.\n", result[0] / 0.63662 * 100.0, quality[0] * 100.0);
|
|
|
|
amps_rx_sat(amps, 1, quality[0]);
|
|
|
|
}
|
|
|
|
} else
|
|
|
|
amps->sat_detect_count = 0;
|
|
|
|
} else {
|
|
|
|
if (amps->sat_detected == 1) {
|
|
|
|
amps->sat_detect_count++;
|
|
|
|
if (amps->sat_detect_count == SAT_LOST_COUNT) {
|
|
|
|
amps->sat_detected = 0;
|
|
|
|
amps->sat_detect_count = 0;
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "SAT signal lost.\n");
|
|
|
|
amps_rx_sat(amps, 0, 0.0);
|
|
|
|
}
|
|
|
|
} else
|
|
|
|
amps->sat_detect_count = 0;
|
|
|
|
}
|
|
|
|
if (quality[1] > 0.8) {
|
|
|
|
if (amps->sig_detected == 0) {
|
|
|
|
amps->sig_detect_count++;
|
|
|
|
if (amps->sig_detect_count == SIG_DETECT_COUNT) {
|
|
|
|
amps->sig_detected = 1;
|
|
|
|
amps->sig_detect_count = 0;
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "Signalling Tone detected with level=%.0f%%, quality=%.0f%%.\n", result[2] / 0.63662 * 100.0, quality[1] * 100.0);
|
|
|
|
amps_rx_signalling_tone(amps, 1, quality[1]);
|
|
|
|
}
|
|
|
|
} else
|
|
|
|
amps->sig_detect_count = 0;
|
|
|
|
} else {
|
|
|
|
if (amps->sig_detected == 1) {
|
|
|
|
amps->sig_detect_count++;
|
|
|
|
if (amps->sig_detect_count == SIG_LOST_COUNT) {
|
|
|
|
amps->sig_detected = 0;
|
|
|
|
amps->sig_detect_count = 0;
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "Signalling Tone lost.\n");
|
|
|
|
amps_rx_signalling_tone(amps, 0, 0.0);
|
|
|
|
}
|
|
|
|
} else
|
|
|
|
amps->sig_detect_count = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* decode signalling/audio */
|
|
|
|
/* Count SIG_TONE_CROSSINGS of zero crossings, then check if the elapsed bit
|
|
|
|
* time is between SIG_TONE_MINBITS and SIG_TONE_MAXBITS. If it is, the
|
|
|
|
* frequency is close to the singalling tone, so it is detected
|
|
|
|
*/
|
|
|
|
void sender_receive_audio(amps_t *amps, int16_t *samples, int length)
|
|
|
|
{
|
|
|
|
transaction_t *trans = amps->trans_list;
|
|
|
|
int16_t *spl;
|
|
|
|
int max, pos;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* SAT detection */
|
|
|
|
|
|
|
|
max = amps->sat_samples;
|
|
|
|
spl = amps->sat_filter_spl;
|
|
|
|
pos = amps->sat_filter_pos;
|
|
|
|
for (i = 0; i < length; i++) {
|
|
|
|
spl[pos++] = samples[i];
|
|
|
|
if (pos == max) {
|
|
|
|
pos = 0;
|
|
|
|
sat_decode(amps, spl, max);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
amps->sat_filter_pos = pos;
|
|
|
|
|
|
|
|
/* receive audio, but only if call established and SAT detected */
|
|
|
|
|
|
|
|
if ((amps->dsp_mode == DSP_MODE_AUDIO_RX_AUDIO_TX || amps->dsp_mode == DSP_MODE_AUDIO_RX_FRAME_TX)
|
|
|
|
&& amps->sender.callref && trans && trans->sat_detected) {
|
|
|
|
int16_t down[length]; /* more than enough */
|
|
|
|
int pos, count;
|
|
|
|
int16_t *spl;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* de-emphasis */
|
|
|
|
if (amps->de_emphasis)
|
|
|
|
de_emphasis(&s->estate, samples, length);
|
|
|
|
/* downsample */
|
|
|
|
count = samplerate_downsample(&s->sender.srstate, samples, length, down);
|
|
|
|
expand_audio(&s->cstate, down, count);
|
|
|
|
spl = amps->sender.rxbuf;
|
|
|
|
pos = amps->sender.rxbuf_pos;
|
|
|
|
for (i = 0; i < count; i++) {
|
|
|
|
spl[pos++] = down[i];
|
|
|
|
if (pos == 160) {
|
|
|
|
call_tx_audio(amps->sender.callref, spl, 160);
|
|
|
|
pos = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
amps->sender.rxbuf_pos = pos;
|
|
|
|
} else
|
|
|
|
amps->sender.rxbuf_pos = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Process received audio stream from radio unit. */
|
|
|
|
void sender_receive(sender_t *sender, int16_t *samples, int length)
|
|
|
|
{
|
|
|
|
amps_t *amps = (amps_t *) sender;
|
|
|
|
double x, y, x_last, y_last, factor;
|
|
|
|
int32_t value;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* high pass filter to remove 0-level
|
|
|
|
* if factor is not set, we should already have 0-level. */
|
|
|
|
factor = amps->highpass_factor;
|
|
|
|
if (factor) {
|
|
|
|
x_last = amps->highpass_x_last;
|
|
|
|
y_last = amps->highpass_y_last;
|
|
|
|
for (i = 0; i < length; i++) {
|
|
|
|
x = (double)samples[i];
|
|
|
|
y = factor * (y_last + x - x_last);
|
|
|
|
x_last = x;
|
|
|
|
y_last = y;
|
|
|
|
value = (int32_t)(y + 0.5);
|
|
|
|
if (value < -32768.0)
|
|
|
|
value = -32768.0;
|
|
|
|
else if (value > 32767)
|
|
|
|
value = 32767;
|
|
|
|
samples[i] = value;
|
|
|
|
}
|
|
|
|
amps->highpass_x_last = x_last;
|
|
|
|
amps->highpass_y_last = y_last;
|
|
|
|
}
|
|
|
|
|
|
|
|
switch (amps->dsp_mode) {
|
|
|
|
case DSP_MODE_OFF:
|
|
|
|
break;
|
|
|
|
case DSP_MODE_FRAME_RX_FRAME_TX:
|
|
|
|
sender_receive_frame(amps, samples, length);
|
|
|
|
break;
|
|
|
|
case DSP_MODE_AUDIO_RX_AUDIO_TX:
|
|
|
|
case DSP_MODE_AUDIO_RX_FRAME_TX:
|
|
|
|
sender_receive_audio(amps, samples, length);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Reset SAT detection states, so ongoing tone will be detected again. */
|
|
|
|
static void sat_reset(amps_t *amps, const char *reason)
|
|
|
|
{
|
|
|
|
PDEBUG(DDSP, DEBUG_DEBUG, "SAT detector reset: %s.\n", reason);
|
|
|
|
amps->sat_detected = 0;
|
|
|
|
amps->sat_detect_count = 0;
|
|
|
|
amps->sig_detected = 0;
|
|
|
|
amps->sig_detect_count = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
void amps_set_dsp_mode(amps_t *amps, enum dsp_mode mode, int frame_length)
|
|
|
|
{
|
|
|
|
#if 0
|
|
|
|
/* reset telegramm */
|
|
|
|
if (mode == DSP_MODE_FRAME && amps->dsp_mode != mode)
|
|
|
|
amps->frame = 0;
|
|
|
|
#endif
|
|
|
|
if (mode == DSP_MODE_FRAME_RX_FRAME_TX) {
|
|
|
|
/* reset SAT detection */
|
|
|
|
sat_reset(amps, "Change to FOCC");
|
|
|
|
}
|
|
|
|
if (amps->dsp_mode == DSP_MODE_FRAME_RX_FRAME_TX
|
|
|
|
&& (mode == DSP_MODE_AUDIO_RX_AUDIO_TX || mode == DSP_MODE_AUDIO_RX_FRAME_TX)) {
|
|
|
|
/* reset SAT detection */
|
|
|
|
sat_reset(amps, "Change from FOCC to FVC");
|
|
|
|
}
|
|
|
|
|
|
|
|
amps->dsp_mode = mode;
|
|
|
|
if (frame_length)
|
|
|
|
amps->fsk_rx_frame_length = frame_length;
|
|
|
|
|
|
|
|
/* reset detection process */
|
|
|
|
amps->fsk_rx_sync = FSK_SYNC_NONE;
|
2016-06-20 15:16:46 +00:00
|
|
|
amps->channel_busy = 0;
|
2016-05-01 17:51:56 +00:00
|
|
|
amps->fsk_rx_sync_register = 0x555;
|
|
|
|
|
|
|
|
/* reset transmitter */
|
|
|
|
amps->fsk_tx_buffer_pos = 0;
|
2016-06-20 15:16:46 +00:00
|
|
|
amps->fsk_tx_frame[0] = '\0';
|
|
|
|
amps->fsk_tx_frame_pos = 0;
|
2016-05-01 17:51:56 +00:00
|
|
|
}
|
|
|
|
|