GNU Radio 3.6.5.1 C++ API
gr_pfb_clock_sync_fff.h
Go to the documentation of this file.
1 /* -*- c++ -*- */
2 /*
3  * Copyright 2009,2010 Free Software Foundation, Inc.
4  *
5  * This file is part of GNU Radio
6  *
7  * GNU Radio is free software; you can redistribute it and/or modify
8  * it under the terms of the GNU General Public License as published by
9  * the Free Software Foundation; either version 3, or (at your option)
10  * any later version.
11  *
12  * GNU Radio is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
15  * GNU General Public License for more details.
16  *
17  * You should have received a copy of the GNU General Public License
18  * along with GNU Radio; see the file COPYING. If not, write to
19  * the Free Software Foundation, Inc., 51 Franklin Street,
20  * Boston, MA 02110-1301, USA.
21  */
22 
23 
24 #ifndef INCLUDED_GR_PFB_CLOCK_SYNC_FFF_H
25 #define INCLUDED_GR_PFB_CLOCK_SYNC_FFF_H
26 
27 #include <gr_core_api.h>
28 #include <gr_block.h>
29 
33  const std::vector<float> &taps,
34  unsigned int filter_size=32,
35  float init_phase=0,
36  float max_rate_deviation=1.5);
37 
38 class gr_fir_fff;
39 
40 /*!
41  * \brief Timing synchronizer using polyphase filterbanks
42  *
43  * This block performs timing synchronization for PAM signals by
44  * minimizing the derivative of the filtered signal, which in turn
45  * maximizes the SNR and minimizes ISI.
46  *
47  * This approach works by setting up two filterbanks; one filterbank
48  * contains the signal's pulse shaping matched filter (such as a root
49  * raised cosine filter), where each branch of the filterbank contains
50  * a different phase of the filter. The second filterbank contains
51  * the derivatives of the filters in the first filterbank. Thinking of
52  * this in the time domain, the first filterbank contains filters that
53  * have a sinc shape to them. We want to align the output signal to be
54  * sampled at exactly the peak of the sinc shape. The derivative of
55  * the sinc contains a zero at the maximum point of the sinc (sinc(0)
56  * = 1, sinc(0)' = 0). Furthermore, the region around the zero point
57  * is relatively linear. We make use of this fact to generate the
58  * error signal.
59  *
60  * If the signal out of the derivative filters is d_i[n] for the ith
61  * filter, and the output of the matched filter is x_i[n], we
62  * calculate the error as: e[n] = (Re{x_i[n]} * Re{d_i[n]} +
63  * Im{x_i[n]} * Im{d_i[n]}) / 2.0 This equation averages the error in
64  * the real and imaginary parts. There are two reasons we multiply by
65  * the signal itself. First, if the symbol could be positive or
66  * negative going, but we want the error term to always tell us to go
67  * in the same direction depending on which side of the zero point we
68  * are on. The sign of x_i[n] adjusts the error term to do
69  * this. Second, the magnitude of x_i[n] scales the error term
70  * depending on the symbol's amplitude, so larger signals give us a
71  * stronger error term because we have more confidence in that
72  * symbol's value. Using the magnitude of x_i[n] instead of just the
73  * sign is especially good for signals with low SNR.
74  *
75  * The error signal, e[n], gives us a value proportional to how far
76  * away from the zero point we are in the derivative signal. We want
77  * to drive this value to zero, so we set up a second order loop. We
78  * have two variables for this loop; d_k is the filter number in the
79  * filterbank we are on and d_rate is the rate which we travel through
80  * the filters in the steady state. That is, due to the natural clock
81  * differences between the transmitter and receiver, d_rate represents
82  * that difference and would traverse the filter phase paths to keep
83  * the receiver locked. Thinking of this as a second-order PLL, the
84  * d_rate is the frequency and d_k is the phase. So we update d_rate
85  * and d_k using the standard loop equations based on two error
86  * signals, d_alpha and d_beta. We have these two values set based on
87  * each other for a critically damped system, so in the block
88  * constructor, we just ask for "gain," which is d_alpha while d_beta
89  * is equal to (gain^2)/4.
90  *
91  * The block's parameters are:
92  *
93  * \li \p sps: The clock sync block needs to know the number of samples per
94  * symbol, because it defaults to return a single point representing
95  * the symbol. The sps can be any positive real number and does not
96  * need to be an integer.
97  *
98  * \li \p loop_bw: The loop bandwidth is used to set the gain of the
99  * inner control loop (see:
100  * http://gnuradio.squarespace.com/blog/2011/8/13/control-loop-gain-values.html).
101  * This should be set small (a value of around 2pi/100 is suggested in
102  * that blog post as the step size for the number of radians around
103  * the unit circle to move relative to the error).
104  *
105  * \li \p taps: One of the most important parameters for this block is
106  * the taps of the filter. One of the benefits of this algorithm is
107  * that you can put the matched filter in here as the taps, so you get
108  * both the matched filter and sample timing correction in one go. So
109  * create your normal matched filter. For a typical digital
110  * modulation, this is a root raised cosine filter. The number of taps
111  * of this filter is based on how long you expect the channel to be;
112  * that is, how many symbols do you want to combine to get the current
113  * symbols energy back (there's probably a better way of stating
114  * that). It's usually 5 to 10 or so. That gives you your filter, but
115  * now we need to think about it as a filter with different phase
116  * profiles in each filter. So take this number of taps and multiply
117  * it by the number of filters. This is the number you would use to
118  * create your prototype filter. When you use this in the PFB
119  * filerbank, it segments these taps into the filterbanks in such a
120  * way that each bank now represents the filter at different phases,
121  * equally spaced at 2pi/N, where N is the number of filters.
122  *
123  * \li \p filter_size (default=32): The number of filters can also be
124  * set and defaults to 32. With 32 filters, you get a good enough
125  * resolution in the phase to produce very small, almost unnoticeable,
126  * ISI. Going to 64 filters can reduce this more, but after that
127  * there is very little gained for the extra complexity.
128  *
129  * \li \p init_phase (default=0): The initial phase is another
130  * settable parameter and refers to the filter path the algorithm
131  * initially looks at (i.e., d_k starts at init_phase). This value
132  * defaults to zero, but it might be useful to start at a different
133  * phase offset, such as the mid-point of the filters.
134  *
135  * \li \p max_rate_deviation (default=1.5): The next parameter is the
136  * max_rate_devitation, which defaults to 1.5. This is how far we
137  * allow d_rate to swing, positive or negative, from 0. Constraining
138  * the rate can help keep the algorithm from walking too far away to
139  * lock during times when there is no signal.
140  *
141  * \li \p osps: note that unlike the ccf version of this algorithm,
142  * this block does \a not have a setting for the number of output
143  * samples per symbol. This is mostly because it should not be
144  * necessary as the reason for having multiple output sps is to
145  * perform equalization and the equalizers will take in complex
146  * numbers in order to do magnitude and phase correction.
147  */
148 
150 {
151  private:
152  /*!
153  * Build the polyphase filterbank timing synchronizer.
154  * \param sps (double) The number of samples per second in the incoming signal
155  * \param gain (float) The alpha gain of the control loop; beta = (gain^2)/4 by default.
156  * \param taps (vector<int>) The filter taps.
157  * \param filter_size (uint) The number of filters in the filterbank (default = 32).
158  * \param init_phase (float) The initial phase to look at, or which filter to start
159  * with (default = 0).
160  * \param max_rate_deviation (float) Distance from 0 d_rate can get (default = 1.5).
161  *
162  */
164  const std::vector<float> &taps,
165  unsigned int filter_size,
166  float init_phase,
167  float max_rate_deviation);
168 
169  bool d_updated;
170  double d_sps;
171  double d_sample_num;
172  float d_alpha;
173  float d_beta;
174  int d_nfilters;
175  std::vector<gr_fir_fff*> d_filters;
176  std::vector<gr_fir_fff*> d_diff_filters;
177  std::vector< std::vector<float> > d_taps;
178  std::vector< std::vector<float> > d_dtaps;
179  float d_k;
180  float d_rate;
181  float d_rate_i;
182  float d_rate_f;
183  float d_max_dev;
184  int d_filtnum;
185  int d_taps_per_filter;
186 
187  /*!
188  * Build the polyphase filterbank timing synchronizer.
189  */
190  gr_pfb_clock_sync_fff (double sps, float gain,
191  const std::vector<float> &taps,
192  unsigned int filter_size,
193  float init_phase,
194  float max_rate_deviation);
195 
196  void create_diff_taps(const std::vector<float> &newtaps,
197  std::vector<float> &difftaps);
198 
199 public:
201 
202  /*!
203  * Resets the filterbank's filter taps with the new prototype filter
204  */
205  void set_taps (const std::vector<float> &taps,
206  std::vector< std::vector<float> > &ourtaps,
207  std::vector<gr_fir_fff*> &ourfilter);
208 
209  /*!
210  * Returns the taps of the matched filter
211  */
212  std::vector<float> channel_taps(int channel);
213 
214  /*!
215  * Returns the taps in the derivative filter
216  */
217  std::vector<float> diff_channel_taps(int channel);
218 
219  /*!
220  * Print all of the filterbank taps to screen.
221  */
222  void print_taps();
223 
224  /*!
225  * Print all of the filterbank taps of the derivative filter to screen.
226  */
227  void print_diff_taps();
228 
229  /*!
230  * Set the gain value alpha for the control loop
231  */
232  void set_alpha(float alpha)
233  {
234  d_alpha = alpha;
235  }
236 
237  /*!
238  * Set the gain value beta for the control loop
239  */
240  void set_beta(float beta)
241  {
242  d_beta = beta;
243  }
244 
245  /*!
246  * Set the maximum deviation from 0 d_rate can have
247  */
248  void set_max_rate_deviation(float m)
249  {
250  d_max_dev = m;
251  }
252 
253  bool check_topology(int ninputs, int noutputs);
254 
255  int general_work (int noutput_items,
256  gr_vector_int &ninput_items,
257  gr_vector_const_void_star &input_items,
258  gr_vector_void_star &output_items);
259 };
260 
261 #endif