File: fastfir.cpp

package info (click to toggle)
cutesdr 1.20-4
  • links: PTS
  • area: main
  • in suites: bookworm, bullseye, forky, sid, trixie
  • size: 2,848 kB
  • sloc: cpp: 18,902; makefile: 21; sh: 5
file content (325 lines) | stat: -rw-r--r-- 10,921 bytes parent folder | download | duplicates (3)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
//////////////////////////////////////////////////////////////////////
// fastfir.cpp: implementation of the CFastFIR class.
//
//  This class implements a FIR Bandpass filter using a FFT convolution algorithm
//The filter is complex and is specified with 3 parameters:
// sample frequency, Hicut and Lowcut frequency
//
//Uses FFT overlap and save method of implementing the FIR.
//For best performance use FIR size   4*FIR <= FFT <= 8*FIR
//If need output to be power of 2 then FIR must = 1/2FFT size
//
// History:
//	2010-09-15  Initial creation MSW
//	2011-03-27  Initial release
//	2011-11-03  Fixed m_pFFTOverlapBuf initialization bug
//	2012-08-06	Fixed m_pWindowTbl sizing problem
//	2013-07-28  Added single/double precision math macros
//////////////////////////////////////////////////////////////////////
//==========================================================================================
// + + +   This Software is released under the "Simplified BSD License"  + + +
//Copyright 2010 Moe Wheatley. All rights reserved.
//
//Redistribution and use in source and binary forms, with or without modification, are
//permitted provided that the following conditions are met:
//
//   1. Redistributions of source code must retain the above copyright notice, this list of
//	  conditions and the following disclaimer.
//
//   2. Redistributions in binary form must reproduce the above copyright notice, this list
//	  of conditions and the following disclaimer in the documentation and/or other materials
//	  provided with the distribution.
//
//THIS SOFTWARE IS PROVIDED BY Moe Wheatley ``AS IS'' AND ANY EXPRESS OR IMPLIED
//WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND
//FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL Moe Wheatley OR
//CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
//CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
//SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
//ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
//NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
//ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
//The views and conclusions contained in the software and documentation are those of the
//authors and should not be interpreted as representing official policies, either expressed
//or implied, of Moe Wheatley.
//==========================================================================================
#include "dsp/fastfir.h"
#include <QDebug>
#include <math.h>
#include "interface/perform.h"
#include <QDir>
#include <QFile>


//////////////////////////////////////////////////////////////////////
// Local Defines
//////////////////////////////////////////////////////////////////////
#define CONV_FFT_SIZE 2048	//must be power of 2
#define CONV_FIR_SIZE 1025	//must be <= FFT size. Make 1/2 +1 if want
							//output to be in power of 2


#define CONV_INBUF_SIZE (CONV_FFT_SIZE+CONV_FIR_SIZE-1)


//////////////////////////////////////////////////////////////////////
// Construction/Destruction
//////////////////////////////////////////////////////////////////////

CFastFIR::CFastFIR()
{
int i;
	m_pWindowTbl = NULL;
	m_pFFTBuf = NULL;
	m_pFFTOverlapBuf = NULL;
	m_pFilterCoef = NULL;
	//allocate internal buffer space on Heap
	m_pWindowTbl = new TYPEREAL[CONV_FIR_SIZE];
	m_pFilterCoef = new TYPECPX[CONV_FFT_SIZE];
	m_pFFTBuf = new TYPECPX[CONV_FFT_SIZE];
	m_pFFTOverlapBuf = new TYPECPX[CONV_FIR_SIZE];

	if(!m_pWindowTbl || !m_pFilterCoef || !m_pFFTBuf || !m_pFFTOverlapBuf)
	{
		//major poblems if memory fails here
		return;
	}
	m_InBufInPos = (CONV_FIR_SIZE - 1);
	for( i=0; i<CONV_FFT_SIZE; i++)
	{
		m_pFFTBuf[i].re = 0.0;
		m_pFFTBuf[i].im = 0.0;
	}
#if 1
	//create Blackman-Nuttall window function for windowed sinc low pass filter design
	for( i=0; i<CONV_FIR_SIZE; i++)
	{
		m_pWindowTbl[i] = (0.3635819
			- 0.4891775*MCOS( (K_2PI*i)/(CONV_FIR_SIZE-1) )
			+ 0.1365995*MCOS( (2.0*K_2PI*i)/(CONV_FIR_SIZE-1) )
			- 0.0106411*MCOS( (3.0*K_2PI*i)/(CONV_FIR_SIZE-1) ) );
		m_pFFTOverlapBuf[i].re = 0.0;
		m_pFFTOverlapBuf[i].im = 0.0;
	}
#endif
#if 0
	//create Blackman-Harris window function for windowed sinc low pass filter design
	for( i=0; i<CONV_FIR_SIZE; i++)
	{
		m_pWindowTbl[i] = (0.35875
			- 0.48829*MCOS( (K_2PI*i)/(CONV_FIR_SIZE-1) )
			+ 0.14128*MCOS( (2.0*K_2PI*i)/(CONV_FIR_SIZE-1) )
			- 0.01168*MCOS( (3.0*K_2PI*i)/(CONV_FIR_SIZE-1) ) );
		m_pFFTOverlapBuf[i].re = 0.0;
		m_pFFTOverlapBuf[i].im = 0.0;
	}
#endif
#if 0
	//create Nuttall window function for windowed sinc low pass filter design
	for( i=0; i<CONV_FIR_SIZE; i++)
	{
		m_pWindowTbl[i] = (0.355768
			- 0.487396*MCOS( (K_2PI*i)/(CONV_FIR_SIZE-1) )
			+ 0.144232*MCOS( (2.0*K_2PI*i)/(CONV_FIR_SIZE-1) )
			- 0.012604*MCOS( (3.0*K_2PI*i)/(CONV_FIR_SIZE-1) ) );
		m_pFFTOverlapBuf[i].re = 0.0;
		m_pFFTOverlapBuf[i].im = 0.0;
	}
#endif
	m_Fft.SetFFTParams(CONV_FFT_SIZE, false, 0.0, 1.0);
	m_FLoCut = -1.0;
	m_FHiCut = 1.0;
	m_Offset = 1.0;
	m_SampleRate = 1.0;
}

CFastFIR::~CFastFIR()
{
	FreeMemory();

}

//////////////////////////////////////////////////////////////////////
//delete all heap memory
//////////////////////////////////////////////////////////////////////
void CFastFIR::FreeMemory()
{
	if(m_pWindowTbl)
	{
		delete m_pWindowTbl;
		m_pWindowTbl = NULL;
	}
	if(m_pFFTOverlapBuf)
	{
		delete m_pFFTOverlapBuf;
		m_pFFTOverlapBuf = NULL;
	}
	if(m_pFilterCoef)
	{
		delete m_pFilterCoef;
		m_pFilterCoef = NULL;
	}
	if(m_pFFTBuf)
	{
		delete m_pFFTBuf;
		m_pFFTBuf = NULL;
	}
}

//////////////////////////////////////////////////////////////////////
//  Call to setup filter parameters
// SampleRate in Hz
// FLowcut is low cutoff frequency of filter in Hz
// FHicut is high cutoff frequency of filter in Hz
// Offset is the CW tone offset frequency
// cutoff frequencies range from -SampleRate/2 to +SampleRate/2
//  HiCut must be greater than LowCut
//		example to make 2700Hz USB filter:
//	SetupParameters( 100, 2800, 0, 48000);
//////////////////////////////////////////////////////////////////////
void CFastFIR::SetupParameters( TYPEREAL FLoCut, TYPEREAL FHiCut,
								TYPEREAL Offset, TYPEREAL SampleRate)
{
int i;
	if( (FLoCut==m_FLoCut) && (FHiCut==m_FHiCut) &&
		(Offset==m_Offset) && (SampleRate==m_SampleRate) )
	{
		return;		//return if no changes
	}
	m_FLoCut = FLoCut;
	m_FHiCut = FHiCut;
	m_Offset = Offset;
	m_SampleRate = SampleRate;

	FLoCut += Offset;
	FHiCut += Offset;
	//sanity check on filter parameters
	if( (FLoCut >= FHiCut) ||
		(FLoCut >= SampleRate/2.0) ||
		(FLoCut <= -SampleRate/2.0) ||
		(FHiCut >= SampleRate/2.0) ||
		(FHiCut <= -SampleRate/2.0) )
	{
		qDebug()<<"Filter Parameter error";
		return;
	}
//qDebug()<<"FLowCut="<<FLoCut<<"FHiCut="<<FHiCut<<"SampleRate="<<SampleRate;
	m_Mutex.lock();
	//calculate some normalized filter parameters
	TYPEREAL nFL = FLoCut/SampleRate;
	TYPEREAL nFH = FHiCut/SampleRate;
	TYPEREAL nFc = (nFH-nFL)/2.0;		//prototype LP filter cutoff
	TYPEREAL nFs = K_2PI*(nFH+nFL)/2.0;		//2 PI times required frequency shift (FHiCut+FLoCut)/2
	TYPEREAL fCenter = 0.5*(TYPEREAL)(CONV_FIR_SIZE-1);	//floating point center index of FIR filter

	for(i=0; i<CONV_FFT_SIZE; i++)		//zero pad entire coefficient buffer to FFT size
	{
		m_pFilterCoef[i].re = 0.0;
		m_pFilterCoef[i].im = 0.0;
	}

	//create LP FIR windowed sinc, sin(x)/x complex LP filter coefficients
	for(i=0; i<CONV_FIR_SIZE; i++)
	{
		TYPEREAL x = (TYPEREAL)i - fCenter;
		TYPEREAL z;
		if( (TYPEREAL)i == fCenter )	//deal with odd size filter singularity where sin(0)/0==1
			z = 2.0 * nFc;
		else
			z = (TYPEREAL)MSIN(K_2PI*x*nFc)/(K_PI*x) * m_pWindowTbl[i];

		//shift lowpass filter coefficients in frequency by (hicut+lowcut)/2 to form bandpass filter anywhere in range
		// (also scales by 1/FFTsize since inverse FFT routine scales by FFTsize)
		m_pFilterCoef[i].re  =  z * MCOS(nFs * x)/(TYPEREAL)CONV_FFT_SIZE;
		m_pFilterCoef[i].im = z * MSIN(nFs * x)/(TYPEREAL)CONV_FFT_SIZE;
	}

#if 0		//debug hack to write m_pFilterCoef to a file for analysis
	QDir::setCurrent("d:/");
	QFile File;
	File.setFileName("lpcoef.txt");
	if(File.open(QIODevice::WriteOnly))
	{
		qDebug()<<"file Opened OK";
		char Buf[256];
		for( i=0; i<CONV_FIR_SIZE; i++)
		{
			sprintf( Buf, "%19.12g %19.12g\r\n", (TYPEREAL)CONV_FFT_SIZE*m_pFilterCoef[i].re, (TYPEREAL)CONV_FFT_SIZE*m_pFilterCoef[i].im);
			File.write(Buf);
		}
	}
	else
		qDebug()<<"file Failed to Open";

#endif
	//convert FIR coefficients to frequency domain by taking forward FFT
	m_Fft.FwdFFT(m_pFilterCoef);
	m_Mutex.unlock();



}

///////////////////////////////////////////////////////////////////////////////
//   Process 'InLength' complex samples in 'InBuf'.
//  returns number of complex samples placed in OutBuf
//number of samples returned in general will not be equal to the number of
//input samples due to FFT block size processing.
//600ns/samp
///////////////////////////////////////////////////////////////////////////////
int CFastFIR::ProcessData(int InLength, TYPECPX* InBuf, TYPECPX* OutBuf)
{
int i = 0;
int j;
int len = InLength;
int outpos = 0;
	if( !InLength)	//if nothing to do
		return 0;
//StartPerformance();
	m_Mutex.lock();
	while(len--)
	{
		j = m_InBufInPos - (CONV_FFT_SIZE - CONV_FIR_SIZE + 1) ;
		if(j >= 0 )
		{	//keep copy of last CONV_FIR_SIZE-1 samples for overlap save
			m_pFFTOverlapBuf[j] = InBuf[i];
		}
		m_pFFTBuf[m_InBufInPos++] = InBuf[i++];
		if(m_InBufInPos >= CONV_FFT_SIZE)
		{	//perform FFT -> complexMultiply by FIR coefficients -> inverse FFT on filled FFT input buffer
			m_Fft.FwdFFT(m_pFFTBuf);
			CpxMpy(CONV_FFT_SIZE, m_pFilterCoef, m_pFFTBuf, m_pFFTBuf);
			m_Fft.RevFFT(m_pFFTBuf);
			for(j=(CONV_FIR_SIZE-1); j<CONV_FFT_SIZE; j++)
			{	//copy FFT output into OutBuf minus CONV_FIR_SIZE-1 samples at beginning
				OutBuf[outpos++] = m_pFFTBuf[j];
			}
			for(j=0; j<(CONV_FIR_SIZE - 1);j++)
			{	//copy overlap buffer into start of fft input buffer
				m_pFFTBuf[j] = m_pFFTOverlapBuf[j];
			}
			//reset input position to data start position of fft input buffer
			m_InBufInPos = CONV_FIR_SIZE - 1;
		}
	}
	m_Mutex.unlock();
//StopPerformance(InLength);
	return outpos;	//return number of output samples processed and placed in OutBuf
}

///////////////////////////////////////////////////////////////////////////////
//   Complex multiply N point array m with src and place in dest.  
// src and dest can be the same buffer.
///////////////////////////////////////////////////////////////////////////////
inline void CFastFIR::CpxMpy(int N, TYPECPX* m, TYPECPX* src, TYPECPX* dest)
{
	for(int i=0; i<N; i++)
	{
		TYPEREAL sr = src[i].re;
		TYPEREAL si = src[i].im;
		dest[i].re = m[i].re * sr - m[i].im * si;
		dest[i].im = m[i].re * si + m[i].im * sr;
	}
}