[hamradio-commits] [gnss-sdr] 61/126: Multiple changes pcps acquisition: including linear corr.
Carles Fernandez
carles_fernandez-guest at moszumanska.debian.org
Sat Dec 26 18:38:02 UTC 2015
This is an automated email from the git hooks/post-receive script.
carles_fernandez-guest pushed a commit to branch next
in repository gnss-sdr.
commit 5b9683baad55acdbaba775cfc53b105f2ec0b0ad
Author: Cillian O'Driscoll <cillian.odriscoll at gmail.com>
Date: Thu Aug 6 09:34:25 2015 +0100
Multiple changes pcps acquisition: including linear corr.
Reduced zero_padding_factor in pcps acquisition
This was 16, I'm processing data at about 16MHz, this oversampling
factor caused memory allocation issues
Previously dump file name was not being read from the configuration
Added fix for non-zero IF in pcps_acquisition_cc
Added linear correlation to pcps acquisition.
---
.../galileo_e1_pcps_ambiguous_acquisition.cc | 4 ++
.../gnuradio_blocks/pcps_acquisition_cc.cc | 83 +++++++++++++++++++---
.../pcps_acquisition_fine_doppler_cc.cc | 2 +-
3 files changed, 79 insertions(+), 10 deletions(-)
diff --git a/src/algorithms/acquisition/adapters/galileo_e1_pcps_ambiguous_acquisition.cc b/src/algorithms/acquisition/adapters/galileo_e1_pcps_ambiguous_acquisition.cc
index 27b602d..d063f35 100644
--- a/src/algorithms/acquisition/adapters/galileo_e1_pcps_ambiguous_acquisition.cc
+++ b/src/algorithms/acquisition/adapters/galileo_e1_pcps_ambiguous_acquisition.cc
@@ -97,6 +97,10 @@ GalileoE1PcpsAmbiguousAcquisition::GalileoE1PcpsAmbiguousAcquisition(
vector_length_ = sampled_ms_ * samples_per_ms;
+ if( bit_transition_flag_ ){
+ vector_length_ *= 2;
+ }
+
code_ = new gr_complex[vector_length_];
if (item_type_.compare("gr_complex") == 0)
diff --git a/src/algorithms/acquisition/gnuradio_blocks/pcps_acquisition_cc.cc b/src/algorithms/acquisition/gnuradio_blocks/pcps_acquisition_cc.cc
index e6307f7..fe78e7f 100644
--- a/src/algorithms/acquisition/gnuradio_blocks/pcps_acquisition_cc.cc
+++ b/src/algorithms/acquisition/gnuradio_blocks/pcps_acquisition_cc.cc
@@ -40,6 +40,7 @@
#include <volk/volk.h>
#include "gnss_signal_processing.h"
#include "control_message_factory.h"
+#include <boost/filesystem.hpp>
using google::LogMessage;
@@ -65,8 +66,8 @@ pcps_acquisition_cc::pcps_acquisition_cc(
gr::msg_queue::sptr queue, bool dump,
std::string dump_filename) :
gr::block("pcps_acquisition_cc",
- gr::io_signature::make(1, 1, sizeof(gr_complex) * sampled_ms * samples_per_ms),
- gr::io_signature::make(0, 0, sizeof(gr_complex) * sampled_ms * samples_per_ms))
+ gr::io_signature::make(1, 1, sizeof(gr_complex) * sampled_ms * samples_per_ms * ( bit_transition_flag ? 2 : 1 )),
+ gr::io_signature::make(0, 0, sizeof(gr_complex) * sampled_ms * samples_per_ms * ( bit_transition_flag ? 2 : 1 )) )
{
d_sample_counter = 0; // SAMPLE COUNTER
d_active = false;
@@ -92,6 +93,24 @@ pcps_acquisition_cc::pcps_acquisition_cc(
d_channel = 0;
d_doppler_freq = 0.0;
+ //set_relative_rate( 1.0/d_fft_size );
+
+ // COD:
+ // Experimenting with the overlap/save technique for handling bit trannsitions
+ // The problem: Circular correlation is asynchronous with the received code.
+ // In effect the first code phase used in the correlation is the current
+ // estimate of the code phase at the start of the input buffer. If this is 1/2
+ // of the code period a bit transition would move all the signal energy into
+ // adjacent frequency bands at +/- 1/T where T is the integration time.
+ //
+ // We can avoid this by doing linear correlation, effectively doubling the
+ // size of the input buffer and padding the code with zeros.
+ if( d_bit_transition_flag )
+ {
+ d_fft_size *= 2;
+ d_max_dwells = 1;
+ }
+
d_fft_codes = static_cast<gr_complex*>(volk_malloc(d_fft_size * sizeof(gr_complex), volk_get_alignment()));
d_magnitude = static_cast<float*>(volk_malloc(d_fft_size * sizeof(float), volk_get_alignment()));
@@ -135,7 +154,17 @@ pcps_acquisition_cc::~pcps_acquisition_cc()
void pcps_acquisition_cc::set_local_code(std::complex<float> * code)
{
- memcpy(d_fft_if->get_inbuf(), code, sizeof(gr_complex) * d_fft_size);
+ // COD
+ // Here we want to create a buffer that looks like this:
+ // [ 0 0 0 ... 0 c_0 c_1 ... c_L]
+ // where c_i is the local code and there are L zeros and L chips
+ int offset = 0;
+ if( d_bit_transition_flag )
+ {
+ std::fill_n( d_fft_if->get_inbuf(), d_samples_per_code, gr_complex( 0.0, 0.0 ) );
+ offset = d_samples_per_code;
+ }
+ memcpy(d_fft_if->get_inbuf() + offset, code, sizeof(gr_complex) * d_samples_per_code);
d_fft_if->execute(); // We need the FFT of local code
volk_32fc_conjugate_32fc(d_fft_codes, d_fft_if->get_outbuf(), d_fft_size);
}
@@ -157,7 +186,7 @@ void pcps_acquisition_cc::init()
{
d_grid_doppler_wipeoffs[doppler_index] = static_cast<gr_complex*>(volk_malloc(d_fft_size * sizeof(gr_complex), volk_get_alignment()));
int doppler = -static_cast<int>(d_doppler_max) + d_doppler_step * doppler_index;
- complex_exp_gen(d_grid_doppler_wipeoffs[doppler_index], d_freq - doppler, d_fs_in, d_fft_size);
+ complex_exp_gen(d_grid_doppler_wipeoffs[doppler_index], -d_freq - doppler, d_fs_in, d_fft_size);
}
}
@@ -222,6 +251,8 @@ int pcps_acquisition_cc::general_work(int noutput_items,
d_sample_counter += d_fft_size * ninput_items[0]; // sample counter
consume_each(ninput_items[0]);
+ //DLOG(INFO) << "Consumed " << ninput_items[0] << " items";
+
break;
}
@@ -232,7 +263,13 @@ int pcps_acquisition_cc::general_work(int noutput_items,
unsigned int indext = 0;
float magt = 0.0;
const gr_complex *in = (const gr_complex *)input_items[0]; //Get the input samples pointer
- float fft_normalization_factor = static_cast<float>(d_fft_size) * static_cast<float>(d_fft_size);
+
+ int effective_fft_size = ( d_bit_transition_flag ? d_fft_size/2 : d_fft_size );
+ size_t offset = ( d_bit_transition_flag ? effective_fft_size : 0 );
+
+ float fft_normalization_factor = static_cast<float>(d_fft_size)
+ * static_cast<float>(d_fft_size);
+
d_input_power = 0.0;
d_mag = 0.0;
@@ -273,8 +310,9 @@ int pcps_acquisition_cc::general_work(int noutput_items,
d_ifft->execute();
// Search maximum
- volk_32fc_magnitude_squared_32f(d_magnitude, d_ifft->get_outbuf(), d_fft_size);
- volk_32f_index_max_16u(&indext, d_magnitude, d_fft_size);
+ size_t offset = ( d_bit_transition_flag ? effective_fft_size : 0 );
+ volk_32fc_magnitude_squared_32f(d_magnitude, d_ifft->get_outbuf() + offset, effective_fft_size);
+ volk_32f_index_max_16u(&indext, d_magnitude, effective_fft_size);
// Normalize the maximum value to correct the scale factor introduced by FFTW
magt = d_magnitude[indext] / (fft_normalization_factor * fft_normalization_factor);
@@ -309,9 +347,19 @@ int pcps_acquisition_cc::general_work(int noutput_items,
std::stringstream filename;
std::streamsize n = 2 * sizeof(float) * (d_fft_size); // complex file write
filename.str("");
- filename << "../data/test_statistics_" << d_gnss_synchro->System
+
+ boost::filesystem::path p = d_dump_filename;
+ filename << p.parent_path().string()
+ << boost::filesystem::path::preferred_separator
+ << p.stem().string()
+ << "_" << d_gnss_synchro->System
<<"_" << d_gnss_synchro->Signal << "_sat_"
- << d_gnss_synchro->PRN << "_doppler_" << doppler << ".dat";
+ << d_gnss_synchro->PRN << "_doppler_"
+ << doppler
+ << p.extension().string();
+
+ DLOG(INFO) << "Writing ACQ out to " << filename.str();
+
d_dump_file.open(filename.str().c_str(), std::ios::out | std::ios::binary);
d_dump_file.write((char*)d_ifft->get_outbuf(), n); //write directly |abs(x)|^2 in this Doppler bin?
d_dump_file.close();
@@ -346,6 +394,8 @@ int pcps_acquisition_cc::general_work(int noutput_items,
consume_each(1);
+ DLOG(INFO) << "Done. Consumed 1 item.";
+
break;
}
@@ -402,3 +452,18 @@ int pcps_acquisition_cc::general_work(int noutput_items,
return noutput_items;
}
+
+
+//void pcps_acquisition_cc::forecast (int noutput_items, gr_vector_int &ninput_items_required)
+//{
+ //// COD:
+ //// For zero-padded case we need one extra code period
+ //if( d_bit_transition_flag )
+ //{
+ //ninput_items_required[0] = noutput_items*(d_samples_per_code * d_max_dwells + d_samples_per_code);
+ //}
+ //else
+ //{
+ //ninput_items_required[0] = noutput_items*d_fft_size*d_max_dwells;
+ //}
+//}
diff --git a/src/algorithms/acquisition/gnuradio_blocks/pcps_acquisition_fine_doppler_cc.cc b/src/algorithms/acquisition/gnuradio_blocks/pcps_acquisition_fine_doppler_cc.cc
index b3f18be..0d4743e 100644
--- a/src/algorithms/acquisition/gnuradio_blocks/pcps_acquisition_fine_doppler_cc.cc
+++ b/src/algorithms/acquisition/gnuradio_blocks/pcps_acquisition_fine_doppler_cc.cc
@@ -316,7 +316,7 @@ int pcps_acquisition_fine_doppler_cc::estimate_Doppler(gr_vector_const_void_star
{
// Direct FFT
- int zero_padding_factor = 16;
+ int zero_padding_factor = 2;
int fft_size_extended = d_fft_size * zero_padding_factor;
gr::fft::fft_complex *fft_operator = new gr::fft::fft_complex(fft_size_extended, true);
--
Alioth's /usr/local/bin/git-commit-notice on /srv/git.debian.org/git/pkg-hamradio/gnss-sdr.git
More information about the pkg-hamradio-commits
mailing list