]> git.gag.com Git - debian/gnuradio/blobdiff - gnuradio-core/src/python/gnuradio/blks2impl/ofdm_sync_pnac.py
Imported Upstream version 3.2.2
[debian/gnuradio] / gnuradio-core / src / python / gnuradio / blks2impl / ofdm_sync_pnac.py
diff --git a/gnuradio-core/src/python/gnuradio/blks2impl/ofdm_sync_pnac.py b/gnuradio-core/src/python/gnuradio/blks2impl/ofdm_sync_pnac.py
new file mode 100644 (file)
index 0000000..10a1259
--- /dev/null
@@ -0,0 +1,125 @@
+#!/usr/bin/env python
+#
+# Copyright 2007 Free Software Foundation, Inc.
+# 
+# This file is part of GNU Radio
+# 
+# GNU Radio 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 3, or (at your option)
+# any later version.
+# 
+# GNU Radio 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 GNU Radio; see the file COPYING.  If not, write to
+# the Free Software Foundation, Inc., 51 Franklin Street,
+# Boston, MA 02110-1301, USA.
+# 
+
+import math
+from numpy import fft
+from gnuradio import gr
+
+class ofdm_sync_pnac(gr.hier_block2):
+    def __init__(self, fft_length, cp_length, kstime, logging=False):
+        """
+        OFDM synchronization using PN Correlation and initial cross-correlation:
+        F. Tufvesson, O. Edfors, and M. Faulkner, "Time and Frequency Synchronization for OFDM using
+        PN-Sequency Preambles," IEEE Proc. VTC, 1999, pp. 2203-2207.
+
+        This implementation is meant to be a more robust version of the Schmidl and Cox receiver design.
+        By correlating against the preamble and using that as the input to the time-delayed correlation,
+        this circuit produces a very clean timing signal at the end of the preamble. The timing is 
+        more accurate and does not have the problem associated with determining the timing from the
+        plateau structure in the Schmidl and Cox.
+
+        This implementation appears to require that the signal is received with a normalized power or signal
+        scalling factor to reduce ambiguities intorduced from partial correlation of the cyclic prefix and
+        the peak detection. A better peak detection block might fix this.
+
+        Also, the cross-correlation falls apart as the frequency offset gets larger and completely fails
+        when an integer offset is introduced. Another thing to look at.
+        """
+
+       gr.hier_block2.__init__(self, "ofdm_sync_pnac",
+                               gr.io_signature(1, 1, gr.sizeof_gr_complex), # Input signature
+                                gr.io_signature2(2, 2, gr.sizeof_float, gr.sizeof_char)) # Output signature
+
+            
+        self.input = gr.add_const_cc(0)
+
+        symbol_length = fft_length + cp_length
+
+        # PN Sync with cross-correlation input
+
+        # cross-correlate with the known symbol
+        kstime = [k.conjugate() for k in kstime[0:fft_length//2]]
+        kstime.reverse()
+        self.crosscorr_filter = gr.fir_filter_ccc(1, kstime)
+        
+        # Create a delay line
+        self.delay = gr.delay(gr.sizeof_gr_complex, fft_length/2)
+
+        # Correlation from ML Sync
+        self.conjg = gr.conjugate_cc();
+        self.corr = gr.multiply_cc();
+
+        # Create a moving sum filter for the input
+        self.mag = gr.complex_to_mag_squared()
+        movingsum_taps = (fft_length//1)*[1.0,]
+        self.power = gr.fir_filter_fff(1,movingsum_taps)
+     
+        # Get magnitude (peaks) and angle (phase/freq error)
+        self.c2mag = gr.complex_to_mag_squared()
+        self.angle = gr.complex_to_arg()
+        self.compare = gr.sub_ff()
+        
+        self.sample_and_hold = gr.sample_and_hold_ff()
+
+        #ML measurements input to sampler block and detect
+        self.threshold = gr.threshold_ff(0,0,0)      # threshold detection might need to be tweaked
+        self.peaks = gr.float_to_char()
+
+        self.connect(self, self.input)
+
+        # Cross-correlate input signal with known preamble
+        self.connect(self.input, self.crosscorr_filter)
+
+        # use the output of the cross-correlation as input time-shifted correlation
+        self.connect(self.crosscorr_filter, self.delay)
+        self.connect(self.crosscorr_filter, (self.corr,0))
+        self.connect(self.delay, self.conjg)
+        self.connect(self.conjg, (self.corr,1))
+        self.connect(self.corr, self.c2mag)
+        self.connect(self.corr, self.angle)
+        self.connect(self.angle, (self.sample_and_hold,0))
+        
+        # Get the power of the input signal to compare against the correlation
+        self.connect(self.crosscorr_filter, self.mag, self.power)
+
+        # Compare the power to the correlator output to determine timing peak
+        # When the peak occurs, it peaks above zero, so the thresholder detects this
+        self.connect(self.c2mag, (self.compare,0))
+        self.connect(self.power, (self.compare,1))
+        self.connect(self.compare, self.threshold)
+        self.connect(self.threshold, self.peaks, (self.sample_and_hold,1))
+
+        # Set output signals
+        #    Output 0: fine frequency correction value
+        #    Output 1: timing signal
+        self.connect(self.sample_and_hold, (self,0))
+        self.connect(self.peaks, (self,1))
+
+        if logging:
+            self.connect(self.compare, gr.file_sink(gr.sizeof_float, "ofdm_sync_pnac-compare_f.dat"))
+            self.connect(self.c2mag, gr.file_sink(gr.sizeof_float, "ofdm_sync_pnac-theta_f.dat"))
+            self.connect(self.power, gr.file_sink(gr.sizeof_float, "ofdm_sync_pnac-inputpower_f.dat"))
+            self.connect(self.angle, gr.file_sink(gr.sizeof_float, "ofdm_sync_pnac-epsilon_f.dat"))
+            self.connect(self.threshold, gr.file_sink(gr.sizeof_float, "ofdm_sync_pnac-threshold_f.dat"))
+            self.connect(self.peaks, gr.file_sink(gr.sizeof_char, "ofdm_sync_pnac-peaks_b.dat"))
+            self.connect(self.sample_and_hold, gr.file_sink(gr.sizeof_float, "ofdm_sync_pnac-sample_and_hold_f.dat"))
+            self.connect(self.input, gr.file_sink(gr.sizeof_gr_complex, "ofdm_sync_pnac-input_c.dat"))