mirror of
https://github.com/f4exb/sdrangel.git
synced 2024-11-03 15:31:15 -05:00
141 lines
5.4 KiB
C++
141 lines
5.4 KiB
C++
///////////////////////////////////////////////////////////////////////////////////
|
|
// Copyright (C) 2015 Edouard Griffiths, F4EXB //
|
|
// Copyright (C) 2020 Jon Beniston, M7RCE //
|
|
// //
|
|
// 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 as 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 V3 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/>. //
|
|
///////////////////////////////////////////////////////////////////////////////////
|
|
|
|
#ifndef INCLUDE_RAISEDCOSINE_H
|
|
#define INCLUDE_RAISEDCOSINE_H
|
|
|
|
#include <cmath>
|
|
#include "dsp/dsptypes.h"
|
|
|
|
// Raised-cosine low-pass filter for pulse shaping, without intersymbol interference (ISI)
|
|
// https://en.wikipedia.org/wiki/Raised-cosine_filter
|
|
// This could be optimised in to a polyphase filter, as samplesPerSymbol-1 inputs
|
|
// to filter() should be zero, as the data is upsampled to the sample rate
|
|
template <class Type> class RaisedCosine {
|
|
public:
|
|
RaisedCosine() : m_ptr(0) { }
|
|
|
|
// beta - roll-off factor
|
|
// symbolSpan - number of symbols over which the filter is spread
|
|
// samplesPerSymbol - number of samples per symbol
|
|
// normaliseUpsampledAmplitude - when true, scale the filter such that an upsampled
|
|
// (by samplesPerSymbol) bipolar sequence (E.g. [1 0 0 -1 0 0..]) has maximum
|
|
// output values close to (1,-1)
|
|
void create(double beta, int symbolSpan, int samplesPerSymbol, bool normaliseUpsampledAmplitude = false)
|
|
{
|
|
int nTaps = symbolSpan * samplesPerSymbol + 1;
|
|
int i, j;
|
|
|
|
// check constraints
|
|
if(!(nTaps & 1)) {
|
|
qDebug("Raised cosine filter has to have an odd number of taps");
|
|
nTaps++;
|
|
}
|
|
|
|
// make room
|
|
m_samples.resize(nTaps);
|
|
for(int i = 0; i < nTaps; i++)
|
|
m_samples[i] = 0;
|
|
m_ptr = 0;
|
|
m_taps.resize(nTaps / 2 + 1);
|
|
|
|
// calculate filter taps
|
|
for(i = 0; i < nTaps / 2 + 1; i++)
|
|
{
|
|
double t = (i - (nTaps / 2)) / (double)samplesPerSymbol;
|
|
double denominator = 1.0 - std::pow(2.0 * beta * t, 2.0);
|
|
double sinc;
|
|
|
|
if (denominator != 0.0)
|
|
{
|
|
if (t == 0)
|
|
sinc = 1.0;
|
|
else
|
|
sinc = sin(M_PI*t)/(M_PI*t);
|
|
m_taps[i] = sinc * (cos(M_PI*beta*t) / denominator) / (double)samplesPerSymbol;
|
|
}
|
|
else
|
|
m_taps[i] = beta * sin(M_PI/(2.0*beta)) / (2.0*samplesPerSymbol);
|
|
}
|
|
|
|
// normalize
|
|
if (!normaliseUpsampledAmplitude)
|
|
{
|
|
// normalize energy
|
|
double sum = 0;
|
|
for(i = 0; i < (int)m_taps.size() - 1; i++)
|
|
sum += std::pow(m_taps[i], 2.0) * 2;
|
|
sum += std::pow(m_taps[i], 2.0);
|
|
sum = std::sqrt(sum);
|
|
for(i = 0; i < (int)m_taps.size(); i++)
|
|
m_taps[i] /= sum;
|
|
}
|
|
else
|
|
{
|
|
// Calculate maximum output of filter, assuming upsampled bipolar input E.g. [1 0 0 -1 0 0..]
|
|
// This doesn't necessarily include the centre tap, so we try each offset
|
|
double maxGain = 0.0;
|
|
for (i = 0; i < samplesPerSymbol; i++)
|
|
{
|
|
double g = 0.0;
|
|
for (j = 0; j < (int)m_taps.size() - 1; j += samplesPerSymbol)
|
|
g += std::fabs(2.0 * m_taps[j]);
|
|
if ((i & 1) == 0)
|
|
g += std::fabs(m_taps[j]);
|
|
if (g > maxGain)
|
|
maxGain = g;
|
|
}
|
|
// Scale up so maximum out is 1
|
|
for(i = 0; i < (int)m_taps.size(); i++)
|
|
m_taps[i] /= maxGain;
|
|
}
|
|
}
|
|
|
|
Type filter(Type sample)
|
|
{
|
|
Type acc = 0;
|
|
unsigned int n_samples = m_samples.size();
|
|
unsigned int n_taps = m_taps.size() - 1;
|
|
unsigned int a = m_ptr;
|
|
unsigned int b = a == n_samples - 1 ? 0 : a + 1;
|
|
|
|
m_samples[m_ptr] = sample;
|
|
|
|
for (unsigned int i = 0; i < n_taps; ++i)
|
|
{
|
|
acc += (m_samples[a] + m_samples[b]) * m_taps[i];
|
|
|
|
a = (a == 0) ? n_samples - 1 : a - 1;
|
|
b = (b == n_samples - 1) ? 0 : b + 1;
|
|
}
|
|
|
|
acc += m_samples[a] * m_taps[n_taps];
|
|
|
|
m_ptr = (m_ptr == n_samples - 1) ? 0 : m_ptr + 1;
|
|
|
|
return acc;
|
|
}
|
|
|
|
private:
|
|
std::vector<Real> m_taps;
|
|
std::vector<Type> m_samples;
|
|
unsigned int m_ptr;
|
|
};
|
|
|
|
#endif // INCLUDE_RAISEDCOSINE_H
|