mirror of
https://github.com/f4exb/sdrangel.git
synced 2024-11-14 12:22:00 -05:00
416 lines
14 KiB
C++
416 lines
14 KiB
C++
/* fir.c
|
|
|
|
This file is part of a program that implements a Software-Defined Radio.
|
|
|
|
Copyright (C) 2013, 2016, 2022 Warren Pratt, NR0V
|
|
Copyright (C) 2024 Edouard Griffiths, F4EXB Adapted to SDRangel
|
|
|
|
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 2
|
|
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, write to the Free Software
|
|
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
|
|
|
|
The author can be reached by email at
|
|
|
|
warren@pratt.one
|
|
*/
|
|
|
|
#define _CRT_SECURE_NO_WARNINGS
|
|
#include "fftw3.h"
|
|
#include "comm.hpp"
|
|
#include "fir.hpp"
|
|
|
|
namespace WDSP {
|
|
|
|
double* FIR::fftcv_mults (int NM, double* c_impulse)
|
|
{
|
|
double* mults = new double[NM * 2]; // (double *) malloc0 (NM * sizeof (dcomplex));
|
|
double* cfft_impulse = new double[NM * 2]; // (double *) malloc0 (NM * sizeof (dcomplex));
|
|
fftw_plan ptmp = fftw_plan_dft_1d(NM, (fftw_complex *) cfft_impulse,
|
|
(fftw_complex *) mults, FFTW_FORWARD, FFTW_PATIENT);
|
|
memset (cfft_impulse, 0, NM * sizeof (dcomplex));
|
|
// store complex coefs right-justified in the buffer
|
|
memcpy (&(cfft_impulse[NM - 2]), c_impulse, (NM / 2 + 1) * sizeof(dcomplex));
|
|
fftw_execute (ptmp);
|
|
fftw_destroy_plan (ptmp);
|
|
delete[] cfft_impulse;
|
|
return mults;
|
|
}
|
|
|
|
double* FIR::get_fsamp_window(int N, int wintype)
|
|
{
|
|
int i;
|
|
double arg0, arg1;
|
|
double* window = new double[N]; // (double *) malloc0 (N * sizeof(double));
|
|
switch (wintype)
|
|
{
|
|
case 0:
|
|
arg0 = 2.0 * PI / ((double)N - 1.0);
|
|
for (i = 0; i < N; i++)
|
|
{
|
|
arg1 = cos(arg0 * (double)i);
|
|
window[i] = +0.21747
|
|
+ arg1 * (-0.45325
|
|
+ arg1 * (+0.28256
|
|
+ arg1 * (-0.04672)));
|
|
}
|
|
break;
|
|
case 1:
|
|
arg0 = 2.0 * PI / ((double)N - 1.0);
|
|
for (i = 0; i < N; ++i)
|
|
{
|
|
arg1 = cos(arg0 * (double)i);
|
|
window[i] = +6.3964424114390378e-02
|
|
+ arg1 * (-2.3993864599352804e-01
|
|
+ arg1 * (+3.5015956323820469e-01
|
|
+ arg1 * (-2.4774111897080783e-01
|
|
+ arg1 * (+8.5438256055858031e-02
|
|
+ arg1 * (-1.2320203369293225e-02
|
|
+ arg1 * (+4.3778825791773474e-04))))));
|
|
}
|
|
break;
|
|
default:
|
|
for (i = 0; i < N; i++)
|
|
window[i] = 1.0;
|
|
}
|
|
return window;
|
|
}
|
|
|
|
double* FIR::fir_fsamp_odd (int N, double* A, int rtype, double scale, int wintype)
|
|
{
|
|
int i, j;
|
|
int mid = (N - 1) / 2;
|
|
double mag, phs;
|
|
double* window;
|
|
double *fcoef = new double[N * 2]; // (double *) malloc0 (N * sizeof (dcomplex));
|
|
double *c_impulse = new double[N * 2]; // (double *) malloc0 (N * sizeof (dcomplex));
|
|
fftw_plan ptmp = fftw_plan_dft_1d(N, (fftw_complex *)fcoef, (fftw_complex *)c_impulse, FFTW_BACKWARD, FFTW_PATIENT);
|
|
double local_scale = 1.0 / (double)N;
|
|
for (i = 0; i <= mid; i++)
|
|
{
|
|
mag = A[i] * local_scale;
|
|
phs = - (double)mid * TWOPI * (double)i / (double)N;
|
|
fcoef[2 * i + 0] = mag * cos (phs);
|
|
fcoef[2 * i + 1] = mag * sin (phs);
|
|
}
|
|
for (i = mid + 1, j = 0; i < N; i++, j++)
|
|
{
|
|
fcoef[2 * i + 0] = + fcoef[2 * (mid - j) + 0];
|
|
fcoef[2 * i + 1] = - fcoef[2 * (mid - j) + 1];
|
|
}
|
|
fftw_execute (ptmp);
|
|
fftw_destroy_plan (ptmp);
|
|
delete[] fcoef;
|
|
window = get_fsamp_window(N, wintype);
|
|
switch (rtype)
|
|
{
|
|
case 0:
|
|
for (i = 0; i < N; i++)
|
|
c_impulse[i] = scale * c_impulse[2 * i] * window[i];
|
|
break;
|
|
case 1:
|
|
for (i = 0; i < N; i++)
|
|
{
|
|
c_impulse[2 * i + 0] *= scale * window[i];
|
|
c_impulse[2 * i + 1] = 0.0;
|
|
}
|
|
break;
|
|
}
|
|
delete[] window;
|
|
return c_impulse;
|
|
}
|
|
|
|
double* FIR::fir_fsamp (int N, double* A, int rtype, double scale, int wintype)
|
|
{
|
|
int n, i, j, k;
|
|
double sum;
|
|
double* window;
|
|
double *c_impulse = new double[N * 2]; // (double *) malloc0 (N * sizeof (complex));
|
|
|
|
if (N & 1)
|
|
{
|
|
int M = (N - 1) / 2;
|
|
for (n = 0; n < M + 1; n++)
|
|
{
|
|
sum = 0.0;
|
|
for (k = 1; k < M + 1; k++)
|
|
sum += 2.0 * A[k] * cos(TWOPI * (n - M) * k / N);
|
|
c_impulse[2 * n + 0] = (1.0 / N) * (A[0] + sum);
|
|
c_impulse[2 * n + 1] = 0.0;
|
|
}
|
|
for (n = M + 1, j = 1; n < N; n++, j++)
|
|
{
|
|
c_impulse[2 * n + 0] = c_impulse[2 * (M - j) + 0];
|
|
c_impulse[2 * n + 1] = 0.0;
|
|
}
|
|
}
|
|
else
|
|
{
|
|
double M = (double)(N - 1) / 2.0;
|
|
for (n = 0; n < N / 2; n++)
|
|
{
|
|
sum = 0.0;
|
|
for (k = 1; k < N / 2; k++)
|
|
sum += 2.0 * A[k] * cos(TWOPI * (n - M) * k / N);
|
|
c_impulse[2 * n + 0] = (1.0 / N) * (A[0] + sum);
|
|
c_impulse[2 * n + 1] = 0.0;
|
|
}
|
|
for (n = N / 2, j = 1; n < N; n++, j++)
|
|
{
|
|
c_impulse[2 * n + 0] = c_impulse[2 * (N / 2 - j) + 0];
|
|
c_impulse[2 * n + 1] = 0.0;
|
|
}
|
|
}
|
|
window = get_fsamp_window (N, wintype);
|
|
switch (rtype)
|
|
{
|
|
case 0:
|
|
for (i = 0; i < N; i++)
|
|
c_impulse[i] = scale * c_impulse[2 * i] * window[i];
|
|
break;
|
|
case 1:
|
|
for (i = 0; i < N; i++)
|
|
{
|
|
c_impulse[2 * i + 0] *= scale * window[i];
|
|
c_impulse[2 * i + 1] = 0.0;
|
|
}
|
|
break;
|
|
}
|
|
delete[] window;
|
|
return c_impulse;
|
|
}
|
|
|
|
double* FIR::fir_bandpass (int N, double f_low, double f_high, double samplerate, int wintype, int rtype, double scale)
|
|
{
|
|
double *c_impulse = new double[N * 2]; // (double *) malloc0 (N * sizeof (complex));
|
|
double ft = (f_high - f_low) / (2.0 * samplerate);
|
|
double ft_rad = TWOPI * ft;
|
|
double w_osc = PI * (f_high + f_low) / samplerate;
|
|
int i, j;
|
|
double m = 0.5 * (double)(N - 1);
|
|
double delta = PI / m;
|
|
double cosphi;
|
|
double posi, posj;
|
|
double sinc, window, coef;
|
|
|
|
if (N & 1)
|
|
{
|
|
switch (rtype)
|
|
{
|
|
case 0:
|
|
c_impulse[N >> 1] = scale * 2.0 * ft;
|
|
break;
|
|
case 1:
|
|
c_impulse[N - 1] = scale * 2.0 * ft;
|
|
c_impulse[ N ] = 0.0;
|
|
break;
|
|
}
|
|
}
|
|
for (i = (N + 1) / 2, j = N / 2 - 1; i < N; i++, j--)
|
|
{
|
|
posi = (double)i - m;
|
|
posj = (double)j - m;
|
|
sinc = sin (ft_rad * posi) / (PI * posi);
|
|
switch (wintype)
|
|
{
|
|
case 0: // Blackman-Harris 4-term
|
|
cosphi = cos (delta * i);
|
|
window = + 0.21747
|
|
+ cosphi * ( - 0.45325
|
|
+ cosphi * ( + 0.28256
|
|
+ cosphi * ( - 0.04672 )));
|
|
break;
|
|
case 1: // Blackman-Harris 7-term
|
|
cosphi = cos (delta * i);
|
|
window = + 6.3964424114390378e-02
|
|
+ cosphi * ( - 2.3993864599352804e-01
|
|
+ cosphi * ( + 3.5015956323820469e-01
|
|
+ cosphi * ( - 2.4774111897080783e-01
|
|
+ cosphi * ( + 8.5438256055858031e-02
|
|
+ cosphi * ( - 1.2320203369293225e-02
|
|
+ cosphi * ( + 4.3778825791773474e-04 ))))));
|
|
break;
|
|
}
|
|
coef = scale * sinc * window;
|
|
switch (rtype)
|
|
{
|
|
case 0:
|
|
c_impulse[i] = + coef * cos (posi * w_osc);
|
|
c_impulse[j] = + coef * cos (posj * w_osc);
|
|
break;
|
|
case 1:
|
|
c_impulse[2 * i + 0] = + coef * cos (posi * w_osc);
|
|
c_impulse[2 * i + 1] = - coef * sin (posi * w_osc);
|
|
c_impulse[2 * j + 0] = + coef * cos (posj * w_osc);
|
|
c_impulse[2 * j + 1] = - coef * sin (posj * w_osc);
|
|
break;
|
|
}
|
|
}
|
|
return c_impulse;
|
|
}
|
|
|
|
double *FIR::fir_read (int N, const char *filename, int rtype, double scale)
|
|
// N = number of real or complex coefficients (see rtype)
|
|
// *filename = filename
|
|
// rtype = 0: real coefficients
|
|
// rtype = 1: complex coefficients
|
|
// scale = a scale factor that will be applied to the returned coefficients;
|
|
// if this is not needed, set it to 1.0
|
|
// NOTE: The number of values in the file must NOT exceed those implied by N and rtype
|
|
{
|
|
FILE *file;
|
|
int i;
|
|
double I, Q;
|
|
double *c_impulse = new double[N * 2]; // (double *) malloc0 (N * sizeof (complex));
|
|
file = fopen (filename, "r");
|
|
for (i = 0; i < N; i++)
|
|
{
|
|
// read in the complex impulse response
|
|
// NOTE: IF the freq response is symmetrical about 0, the imag coeffs will all be zero.
|
|
switch (rtype)
|
|
{
|
|
case 0:
|
|
{
|
|
int r = fscanf (file, "%le", &I);
|
|
fprintf(stderr, "^%d parameters read\n", r);
|
|
c_impulse[i] = + scale * I;
|
|
break;
|
|
}
|
|
case 1:
|
|
{
|
|
int r = fscanf (file, "%le", &I);
|
|
fprintf(stderr, "%d parameters read\n", r);
|
|
r = fscanf (file, "%le", &Q);
|
|
fprintf(stderr, "%d parameters read\n", r);
|
|
c_impulse[2 * i + 0] = + scale * I;
|
|
c_impulse[2 * i + 1] = - scale * Q;
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
fclose (file);
|
|
return c_impulse;
|
|
}
|
|
|
|
void FIR::analytic (int N, double* in, double* out)
|
|
{
|
|
int i;
|
|
double inv_N = 1.0 / (double)N;
|
|
double two_inv_N = 2.0 * inv_N;
|
|
double* x = new double[N * 2]; // (double *) malloc0 (N * sizeof (complex));
|
|
fftw_plan pfor = fftw_plan_dft_1d (N, (fftw_complex *) in,
|
|
(fftw_complex *) x, FFTW_FORWARD, FFTW_PATIENT);
|
|
fftw_plan prev = fftw_plan_dft_1d (N, (fftw_complex *) x,
|
|
(fftw_complex *) out, FFTW_BACKWARD, FFTW_PATIENT);
|
|
fftw_execute (pfor);
|
|
x[0] *= inv_N;
|
|
x[1] *= inv_N;
|
|
for (i = 1; i < N / 2; i++)
|
|
{
|
|
x[2 * i + 0] *= two_inv_N;
|
|
x[2 * i + 1] *= two_inv_N;
|
|
}
|
|
x[N + 0] *= inv_N;
|
|
x[N + 1] *= inv_N;
|
|
memset (&x[N + 2], 0, (N - 2) * sizeof (double));
|
|
fftw_execute (prev);
|
|
fftw_destroy_plan (prev);
|
|
fftw_destroy_plan (pfor);
|
|
delete[] x;
|
|
}
|
|
|
|
void FIR::mp_imp (int N, double* fir, double* mpfir, int pfactor, int polarity)
|
|
{
|
|
int i;
|
|
int size = N * pfactor;
|
|
double inv_PN = 1.0 / (double)size;
|
|
double* firpad = new double[size * 2]; // (double *) malloc0 (size * sizeof (complex));
|
|
double* firfreq = new double[size * 2]; // (double *) malloc0 (size * sizeof (complex));
|
|
double* mag = new double[size]; // (double *) malloc0 (size * sizeof (double));
|
|
double* ana = new double[size * 2]; // (double *) malloc0 (size * sizeof (complex));
|
|
double* impulse = new double[size * 2]; // (double *) malloc0 (size * sizeof (complex));
|
|
double* newfreq = new double[size * 2]; // (double *) malloc0 (size * sizeof (complex));
|
|
memcpy (firpad, fir, N * sizeof (dcomplex));
|
|
fftw_plan pfor = fftw_plan_dft_1d (size, (fftw_complex *) firpad,
|
|
(fftw_complex *) firfreq, FFTW_FORWARD, FFTW_PATIENT);
|
|
fftw_plan prev = fftw_plan_dft_1d (size, (fftw_complex *) newfreq,
|
|
(fftw_complex *) impulse, FFTW_BACKWARD, FFTW_PATIENT);
|
|
// print_impulse("orig_imp.txt", N, fir, 1, 0);
|
|
fftw_execute (pfor);
|
|
for (i = 0; i < size; i++)
|
|
{
|
|
mag[i] = sqrt (firfreq[2 * i + 0] * firfreq[2 * i + 0] + firfreq[2 * i + 1] * firfreq[2 * i + 1]) * inv_PN;
|
|
if (mag[i] > 0.0)
|
|
ana[2 * i + 0] = log (mag[i]);
|
|
else
|
|
ana[2 * i + 0] = log (1.0e-300);
|
|
}
|
|
analytic (size, ana, ana);
|
|
for (i = 0; i < size; i++)
|
|
{
|
|
newfreq[2 * i + 0] = + mag[i] * cos (ana[2 * i + 1]);
|
|
if (polarity)
|
|
newfreq[2 * i + 1] = + mag[i] * sin (ana[2 * i + 1]);
|
|
else
|
|
newfreq[2 * i + 1] = - mag[i] * sin (ana[2 * i + 1]);
|
|
}
|
|
fftw_execute (prev);
|
|
if (polarity)
|
|
memcpy (mpfir, &impulse[2 * (pfactor - 1) * N], N * sizeof (dcomplex));
|
|
else
|
|
memcpy (mpfir, impulse, N * sizeof (dcomplex));
|
|
// print_impulse("min_imp.txt", N, mpfir, 1, 0);
|
|
fftw_destroy_plan (prev);
|
|
fftw_destroy_plan (pfor);
|
|
delete[] (newfreq);
|
|
delete[] (impulse);
|
|
delete[] (ana);
|
|
delete[] (mag);
|
|
delete[] (firfreq);
|
|
delete[] (firpad);
|
|
}
|
|
|
|
// impulse response of a zero frequency filter comprising a cascade of two resonators,
|
|
// each followed by a detrending filter
|
|
double* FIR::zff_impulse(int nc, double scale)
|
|
{
|
|
// nc = number of coefficients (power of two)
|
|
int n_resdet = nc / 2 - 1; // size of single zero-frequency resonator with detrender
|
|
int n_dresdet = 2 * n_resdet - 1; // size of two cascaded units; when we convolve these we get 2 * n - 1 length
|
|
// allocate the single and make the values
|
|
double* resdet = new double[n_resdet]; // (double*)malloc0 (n_resdet * sizeof(double));
|
|
for (int i = 1, j = 0, k = n_resdet - 1; i < nc / 4; i++, j++, k--)
|
|
resdet[j] = resdet[k] = (double)(i * (i + 1) / 2);
|
|
resdet[nc / 4 - 1] = (double)(nc / 4 * (nc / 4 + 1) / 2);
|
|
// print_impulse ("resdet", n_resdet, resdet, 0, 0);
|
|
// allocate the double and complex versions and make the values
|
|
double* dresdet = new double[n_dresdet]; // (double*)malloc0 (n_dresdet * sizeof(double));
|
|
double div = (double)((nc / 2 + 1) * (nc / 2 + 1)); // calculate divisor
|
|
double* c_dresdet = new double[nc * 2]; // (double*)malloc0 (nc * sizeof(complex));
|
|
for (int n = 0; n < n_dresdet; n++) // convolve to make the cascade
|
|
{
|
|
for (int k = 0; k < n_resdet; k++)
|
|
if ((n - k) >= 0 && (n - k) < n_resdet)
|
|
dresdet[n] += resdet[k] * resdet[n - k];
|
|
dresdet[n] /= div;
|
|
c_dresdet[2 * n + 0] = dresdet[n] * scale;
|
|
c_dresdet[2 * n + 1] = 0.0;
|
|
}
|
|
// print_impulse("dresdet", n_dresdet, dresdet, 0, 0);
|
|
// print_impulse("c_dresdet", nc, c_dresdet, 1, 0);
|
|
delete[] (dresdet);
|
|
delete[] (resdet);
|
|
return c_dresdet;
|
|
}
|
|
|
|
} // namespace WDSP
|