def test_sample_and_hold_ff(self):
src_data1 = tuple([i for i in range(10)])
src_data2 = (0, 0, 1, 1, 0, 1, 0, 0, 1, 1)
expected_result = (0, 0, 2, 3, 3, 5, 5, 5, 8, 9)
src1 = gr.vector_source_f(src_data1)
src2 = gr.vector_source_b(src_data2)
op = gr.sample_and_hold_ff()
dst = gr.vector_sink_f()
self.tb.connect(src1, op)
self.tb.connect(src2, op)
self.tb.connect(op, dst)
self.tb.run()
result_data = dst.data()
self.assertFloatTuplesAlmostEqual(expected_result, result_data)
def test_sample_and_hold_ff(self):
src_data1 = tuple([i for i in range(10)])
src_data2 = (0, 0, 1, 1, 0, 1, 0, 0, 1, 1)
expected_result = (0, 0, 2, 3, 3, 5, 5, 5, 8, 9)
src1 = gr.vector_source_f(src_data1)
src2 = gr.vector_source_b(src_data2)
op = gr.sample_and_hold_ff()
dst = gr.vector_sink_f()
self.tb.connect(src1, op)
self.tb.connect(src2, op)
self.tb.connect(op, dst)
self.tb.run()
result_data = dst.data()
self.assertFloatTuplesAlmostEqual(expected_result, result_data)
def __init__(self, fft_length, cp_length, logging=False):
"""
OFDM synchronization using PN Correlation:
T. M. Schmidl and D. C. Cox, "Robust Frequency and Timing
Synchonization for OFDM," IEEE Trans. Communications, vol. 45,
no. 12, 1997.
"""
gr.hier_block2.__init__(self, "ofdm_sync_pn",
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)
# PN Sync
# 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 corr output
if 1:
moving_sum_taps = [1.0 for i in range(fft_length//2)]
self.moving_sum_filter = gr.fir_filter_ccf(1,moving_sum_taps)
else:
moving_sum_taps = [complex(1.0,0.0) for i in range(fft_length//2)]
self.moving_sum_filter = gr.fft_filter_ccc(1,moving_sum_taps)
# Create a moving sum filter for the input
self.inputmag2 = gr.complex_to_mag_squared()
# Modified by Yong (12.06.27)
#movingsum2_taps = [1.0 for i in range(fft_length//2)]
movingsum2_taps = [0.5 for i in range(fft_length)]
if 1:
self.inputmovingsum = gr.fir_filter_fff(1,movingsum2_taps)
else:
self.inputmovingsum = gr.fft_filter_fff(1,movingsum2_taps)
self.square = gr.multiply_ff()
self.normalize = gr.divide_ff()
# Get magnitude (peaks) and angle (phase/freq error)
self.c2mag = gr.complex_to_mag_squared()
self.angle = gr.complex_to_arg()
self.sample_and_hold = gr.sample_and_hold_ff()
#ML measurements input to sampler block and detect
self.sub1 = gr.add_const_ff(-1)
self.pk_detect = gr.peak_detector_fb(0.20, 0.20, 30, 0.001)
#self.pk_detect = gr.peak_detector2_fb(9)
self.connect(self, self.input)
# Calculate the frequency offset from the correlation of the preamble
self.connect(self.input, self.delay)
self.connect(self.input, (self.corr,0))
self.connect(self.delay, self.conjg)
self.connect(self.conjg, (self.corr,1))
self.connect(self.corr, self.moving_sum_filter)
self.connect(self.moving_sum_filter, self.c2mag)
self.connect(self.moving_sum_filter, self.angle)
self.connect(self.angle, (self.sample_and_hold,0))
# Get the power of the input signal to normalize the output of the correlation
self.connect(self.input, self.inputmag2, self.inputmovingsum)
self.connect(self.inputmovingsum, (self.square,0))
self.connect(self.inputmovingsum, (self.square,1))
self.connect(self.square, (self.normalize,1))
self.connect(self.c2mag, (self.normalize,0))
# Create a moving sum filter for the corr output
matched_filter_taps = [1.0/cp_length for i in range(cp_length)]
self.matched_filter = gr.fir_filter_fff(1,matched_filter_taps)
self.connect(self.normalize, self.matched_filter)
self.connect(self.matched_filter, self.sub1, self.pk_detect)
#self.connect(self.matched_filter, self.pk_detect)
self.connect(self.pk_detect, (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.pk_detect, (self,1))
if logging:
self.connect(self.matched_filter, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-mf_f.dat"))
self.connect(self.c2mag, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-nominator_f.dat"))
self.connect(self.square, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-denominator_f.dat"))
self.connect(self.normalize, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-theta_f.dat"))
self.connect(self.angle, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-epsilon_f.dat"))
self.connect(self.pk_detect, gr.file_sink(gr.sizeof_char, "ofdm_sync_pn-peaks_b.dat"))
self.connect(self.sample_and_hold, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-sample_and_hold_f.dat"))
self.connect(self.input, gr.file_sink(gr.sizeof_gr_complex, "ofdm_sync_pn-input_c.dat"))
def __init__(self, fft_length, cp_length, half_sync, logging=False):
gr.hier_block2.__init__(
self,
"ofdm_sync_pn",
gr.io_signature(1, 1, gr.sizeof_gr_complex), # Input signature
gr.io_signature3(
3,
3, # Output signature
gr.sizeof_gr_complex, # delayed input
gr.sizeof_float, # fine frequency offset
gr.sizeof_char # timing indicator
))
if half_sync:
period = fft_length / 2
window = fft_length / 2
else: # full symbol
period = fft_length + cp_length
window = fft_length # makes the plateau cp_length long
# Calculate the frequency offset from the correlation of the preamble
x_corr = gr.multiply_cc()
self.connect(self, gr.conjugate_cc(), (x_corr, 0))
self.connect(self, gr.delay(gr.sizeof_gr_complex, period), (x_corr, 1))
P_d = gr.moving_average_cc(window, 1.0)
self.connect(x_corr, P_d)
P_d_angle = gr.complex_to_arg()
self.connect(P_d, P_d_angle)
# Get the power of the input signal to normalize the output of the correlation
R_d = gr.moving_average_ff(window, 1.0)
self.connect(self, gr.complex_to_mag_squared(), R_d)
R_d_squared = gr.multiply_ff() # this is retarded
self.connect(R_d, (R_d_squared, 0))
self.connect(R_d, (R_d_squared, 1))
M_d = gr.divide_ff()
self.connect(P_d, gr.complex_to_mag_squared(), (M_d, 0))
self.connect(R_d_squared, (M_d, 1))
# Now we need to detect peak of M_d
# NOTE: replaced fir_filter with moving_average for clarity
# the peak is up to cp_length long, but noisy, so average it out
#matched_filter_taps = [1.0/cp_length for i in range(cp_length)]
#matched_filter = gr.fir_filter_fff(1, matched_filter_taps)
matched_filter = gr.moving_average_ff(cp_length, 1.0 / cp_length)
# NOTE: the look_ahead parameter doesn't do anything
# these parameters are kind of magic, increase 1 and 2 (==) to be more tolerant
#peak_detect = raw.peak_detector_fb(0.55, 0.55, 30, 0.001)
peak_detect = raw.peak_detector_fb(0.25, 0.25, 30, 0.001)
# NOTE: gr.peak_detector_fb is broken!
#peak_detect = gr.peak_detector_fb(0.55, 0.55, 30, 0.001)
#peak_detect = gr.peak_detector_fb(0.45, 0.45, 30, 0.001)
#peak_detect = gr.peak_detector_fb(0.30, 0.30, 30, 0.001)
# offset by -1
self.connect(M_d, matched_filter, gr.add_const_ff(-1), peak_detect)
# peak_detect indicates the time M_d is highest, which is the end of the symbol.
# We should try to sample in the middle of the plateau!!
# FIXME until we figure out how to do this, just offset by cp_length/2
offset = 6 #cp_length/2
# nco(t) = P_d_angle(t-offset) sampled at peak_detect(t)
# modulate input(t - fft_length) by nco(t)
# signal to sample input(t) at t-offset
#
# We can't delay by < 0 so instead:
# input is delayed by fft_length
# P_d_angle is delayed by offset
# signal to sample is delayed by fft_length - offset
#
phi = gr.sample_and_hold_ff()
self.connect(peak_detect, (phi, 1))
self.connect(P_d_angle, gr.delay(gr.sizeof_float, offset), (phi, 0))
#self.connect(P_d_angle, matched_filter2, (phi,0)) # why isn't this better?!?
# FIXME: we add fft_length delay so that the preamble is nco corrected too
# BUT is this buffering worth it? consider implementing sync as a proper block
# delay the input signal to follow the frequency offset signal
self.connect(self, gr.delay(gr.sizeof_gr_complex,
(fft_length + offset)), (self, 0))
self.connect(phi, (self, 1))
self.connect(peak_detect, (self, 2))
if logging:
self.connect(matched_filter,
gr.file_sink(gr.sizeof_float, "sync-mf.dat"))
self.connect(M_d, gr.file_sink(gr.sizeof_float, "sync-M.dat"))
self.connect(P_d_angle,
gr.file_sink(gr.sizeof_float, "sync-angle.dat"))
self.connect(peak_detect,
gr.file_sink(gr.sizeof_char, "sync-peaks.datb"))
self.connect(phi, gr.file_sink(gr.sizeof_float, "sync-phi.dat"))
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 = filter.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 = filter.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"))
def __init__(self, fft_length, cp_length, snr, kstime, logging):
''' Maximum Likelihood OFDM synchronizer:
J. van de Beek, M. Sandell, and P. O. Borjesson, "ML Estimation
of Time and Frequency Offset in OFDM Systems," IEEE Trans.
Signal Processing, vol. 45, no. 7, pp. 1800-1805, 1997.
'''
gr.hier_block2.__init__(self, "ofdm_sync_ml",
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)
SNR = 10.0**(snr/10.0)
rho = SNR / (SNR + 1.0)
symbol_length = fft_length + cp_length
# ML Sync
# Energy Detection from ML Sync
self.connect(self, self.input)
# Create a delay line
self.delay = gr.delay(gr.sizeof_gr_complex, fft_length)
self.connect(self.input, self.delay)
# magnitude squared blocks
self.magsqrd1 = gr.complex_to_mag_squared()
self.magsqrd2 = gr.complex_to_mag_squared()
self.adder = gr.add_ff()
moving_sum_taps = [rho/2 for i in range(cp_length)]
self.moving_sum_filter = gr.fir_filter_fff(1,moving_sum_taps)
self.connect(self.input,self.magsqrd1)
self.connect(self.delay,self.magsqrd2)
self.connect(self.magsqrd1,(self.adder,0))
self.connect(self.magsqrd2,(self.adder,1))
self.connect(self.adder,self.moving_sum_filter)
# Correlation from ML Sync
self.conjg = gr.conjugate_cc();
self.mixer = gr.multiply_cc();
movingsum2_taps = [1.0 for i in range(cp_length)]
self.movingsum2 = gr.fir_filter_ccf(1,movingsum2_taps)
# Correlator data handler
self.c2mag = gr.complex_to_mag()
self.angle = gr.complex_to_arg()
self.connect(self.input,(self.mixer,1))
self.connect(self.delay,self.conjg,(self.mixer,0))
self.connect(self.mixer,self.movingsum2,self.c2mag)
self.connect(self.movingsum2,self.angle)
# ML Sync output arg, need to find maximum point of this
self.diff = gr.sub_ff()
self.connect(self.c2mag,(self.diff,0))
self.connect(self.moving_sum_filter,(self.diff,1))
#ML measurements input to sampler block and detect
self.f2c = gr.float_to_complex()
self.pk_detect = gr.peak_detector_fb(0.2, 0.25, 30, 0.0005)
self.sample_and_hold = gr.sample_and_hold_ff()
# use the sync loop values to set the sampler and the NCO
# self.diff = theta
# self.angle = epsilon
self.connect(self.diff, self.pk_detect)
# The DPLL corrects for timing differences between CP correlations
use_dpll = 0
if use_dpll:
self.dpll = gr.dpll_bb(float(symbol_length),0.01)
self.connect(self.pk_detect, self.dpll)
self.connect(self.dpll, (self.sample_and_hold,1))
else:
self.connect(self.pk_detect, (self.sample_and_hold,1))
self.connect(self.angle, (self.sample_and_hold,0))
################################
# correlate against known symbol
# This gives us the same timing signal as the PN sync block only on the preamble
# we don't use the signal generated from the CP correlation because we don't want
# to readjust the timing in the middle of the packet or we ruin the equalizer settings.
kstime = [k.conjugate() for k in kstime]
kstime.reverse()
self.kscorr = gr.fir_filter_ccc(1, kstime)
self.corrmag = gr.complex_to_mag_squared()
self.div = gr.divide_ff()
# The output signature of the correlation has a few spikes because the rest of the
# system uses the repeated preamble symbol. It needs to work that generically if
# anyone wants to use this against a WiMAX-like signal since it, too, repeats.
# The output theta of the correlator above is multiplied with this correlation to
# identify the proper peak and remove other products in this cross-correlation
self.threshold_factor = 0.1
self.slice = gr.threshold_ff(self.threshold_factor, self.threshold_factor, 0)
self.f2b = gr.float_to_char()
self.b2f = gr.char_to_float()
self.mul = gr.multiply_ff()
# Normalize the power of the corr output by the energy. This is not really needed
# and could be removed for performance, but it makes for a cleaner signal.
# if this is removed, the threshold value needs adjustment.
self.connect(self.input, self.kscorr, self.corrmag, (self.div,0))
self.connect(self.moving_sum_filter, (self.div,1))
self.connect(self.div, (self.mul,0))
self.connect(self.pk_detect, self.b2f, (self.mul,1))
self.connect(self.mul, self.slice)
# Set output signals
# Output 0: fine frequency correction value
# Output 1: timing signal
self.connect(self.sample_and_hold, (self,0))
self.connect(self.slice, self.f2b, (self,1))
if logging:
self.connect(self.moving_sum_filter, gr.file_sink(gr.sizeof_float, "ofdm_sync_ml-energy_f.dat"))
self.connect(self.diff, gr.file_sink(gr.sizeof_float, "ofdm_sync_ml-theta_f.dat"))
self.connect(self.angle, gr.file_sink(gr.sizeof_float, "ofdm_sync_ml-epsilon_f.dat"))
self.connect(self.corrmag, gr.file_sink(gr.sizeof_float, "ofdm_sync_ml-corrmag_f.dat"))
self.connect(self.kscorr, gr.file_sink(gr.sizeof_gr_complex, "ofdm_sync_ml-kscorr_c.dat"))
self.connect(self.div, gr.file_sink(gr.sizeof_float, "ofdm_sync_ml-div_f.dat"))
self.connect(self.mul, gr.file_sink(gr.sizeof_float, "ofdm_sync_ml-mul_f.dat"))
self.connect(self.slice, gr.file_sink(gr.sizeof_float, "ofdm_sync_ml-slice_f.dat"))
self.connect(self.pk_detect, gr.file_sink(gr.sizeof_char, "ofdm_sync_ml-peaks_b.dat"))
if use_dpll:
self.connect(self.dpll, gr.file_sink(gr.sizeof_char, "ofdm_sync_ml-dpll_b.dat"))
self.connect(self.sample_and_hold, gr.file_sink(gr.sizeof_float, "ofdm_sync_ml-sample_and_hold_f.dat"))
self.connect(self.input, gr.file_sink(gr.sizeof_gr_complex, "ofdm_sync_ml-input_c.dat"))
def __init__(self, fft_length, cp_length, half_sync, kstime, ks1time, threshold, logging=False):
gr.hier_block2.__init__(
self,
"ofdm_sync_pn",
gr.io_signature(1, 1, gr.sizeof_gr_complex), # Input signature
gr.io_signature3(
3,
3, # Output signature
gr.sizeof_gr_complex, # delayed input
gr.sizeof_float, # fine frequency offset
gr.sizeof_char, # timing indicator
),
)
if half_sync:
period = fft_length / 2
window = fft_length / 2
else: # full symbol
period = fft_length + cp_length
window = fft_length # makes the plateau cp_length long
# Calculate the frequency offset from the correlation of the preamble
x_corr = gr.multiply_cc()
self.connect(self, gr.conjugate_cc(), (x_corr, 0))
self.connect(self, gr.delay(gr.sizeof_gr_complex, period), (x_corr, 1))
P_d = gr.moving_average_cc(window, 1.0)
self.connect(x_corr, P_d)
# offset by -1
phi = gr.sample_and_hold_ff()
self.corrmag = gr.complex_to_mag_squared()
P_d_angle = gr.complex_to_arg()
self.connect(P_d, P_d_angle, (phi, 0))
cross_correlate = 1
if cross_correlate == 1:
# cross-correlate with the known symbol
kstime = [k.conjugate() for k in kstime]
kstime.reverse()
self.crosscorr_filter = gr.fir_filter_ccc(1, kstime)
"""
self.f2b = gr.float_to_char()
self.slice = gr.threshold_ff(threshold, threshold, 0, fft_length)
#self.connect(self, self.crosscorr_filter, self.corrmag, self.slice, self.f2b)
self.connect(self.f2b, (phi,1))
self.connect(self.f2b, (self,2))
self.connect(self.f2b, gr.file_sink(gr.sizeof_char, "ofdm_f2b.dat"))
"""
# new method starts here - only crosscorrelate and use peak_detect block #
peak_detect = gr.peak_detector_fb(100, 100, 30, 0.001)
self.corrmag1 = gr.complex_to_mag_squared()
self.connect(self, self.crosscorr_filter, self.corrmag, peak_detect)
self.connect(peak_detect, (phi, 1))
self.connect(peak_detect, (self, 2))
self.connect(peak_detect, gr.file_sink(gr.sizeof_char, "sync-peaks_b.dat"))
self.connect(self.corrmag, gr.file_sink(gr.sizeof_float, "ofdm_corrmag.dat"))
self.connect(self, gr.delay(gr.sizeof_gr_complex, (fft_length)), (self, 0))
else:
# Get the power of the input signal to normalize the output of the correlation
R_d = gr.moving_average_ff(window, 1.0)
self.connect(self, gr.complex_to_mag_squared(), R_d)
R_d_squared = gr.multiply_ff() # this is retarded
self.connect(R_d, (R_d_squared, 0))
self.connect(R_d, (R_d_squared, 1))
M_d = gr.divide_ff()
self.connect(P_d, gr.complex_to_mag_squared(), (M_d, 0))
self.connect(R_d_squared, (M_d, 1))
# Now we need to detect peak of M_d
matched_filter = gr.moving_average_ff(cp_length, 1.0 / cp_length)
peak_detect = gr.peak_detector_fb(0.25, 0.25, 30, 0.001)
self.connect(M_d, matched_filter, gr.add_const_ff(-1), peak_detect)
offset = cp_length / 2 # cp_length/2
self.connect(peak_detect, (phi, 1))
self.connect(peak_detect, (self, 2))
self.connect(P_d_angle, gr.delay(gr.sizeof_float, offset), (phi, 0))
self.connect(
self, gr.delay(gr.sizeof_gr_complex, (fft_length + offset)), (self, 0)
) # delay the input to follow the freq offset
self.connect(peak_detect, gr.delay(gr.sizeof_char, (fft_length + offset)), (self, 2))
self.connect(peak_detect, gr.file_sink(gr.sizeof_char, "sync-peaks_b.dat"))
self.connect(matched_filter, gr.file_sink(gr.sizeof_float, "sync-mf.dat"))
self.connect(phi, (self, 1))
if logging:
self.connect(matched_filter, gr.file_sink(gr.sizeof_float, "sync-mf.dat"))
self.connect(M_d, gr.file_sink(gr.sizeof_float, "sync-M.dat"))
self.connect(P_d_angle, gr.file_sink(gr.sizeof_float, "sync-angle.dat"))
self.connect(peak_detect, gr.file_sink(gr.sizeof_char, "sync-peaks.datb"))
self.connect(phi, gr.file_sink(gr.sizeof_float, "sync-phi.dat"))
def __init__(self, fft_length, cp_length, logging=False):
"""
OFDM synchronization using PN Correlation:
T. M. Schmidl and D. C. Cox, "Robust Frequency and Timing
Synchonization for OFDM," IEEE Trans. Communications, vol. 45,
no. 12, 1997.
"""
gr.hier_block2.__init__(self, "ofdm_sync_pn",
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)
# PN Sync
# 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 corr output
if 1:
moving_sum_taps = [1.0 for i in range(fft_length//2)]
self.moving_sum_filter = gr.fir_filter_ccf(1,moving_sum_taps)
else:
moving_sum_taps = [complex(1.0,0.0) for i in range(fft_length//2)]
self.moving_sum_filter = gr.fft_filter_ccc(1,moving_sum_taps)
# Create a moving sum filter for the input
self.inputmag2 = gr.complex_to_mag_squared()
movingsum2_taps = [1.0 for i in range(fft_length//2)]
if 1:
self.inputmovingsum = gr.fir_filter_fff(1,movingsum2_taps)
else:
self.inputmovingsum = gr.fft_filter_fff(1,movingsum2_taps)
self.square = gr.multiply_ff()
self.normalize = gr.divide_ff()
# Get magnitude (peaks) and angle (phase/freq error)
self.c2mag = gr.complex_to_mag_squared()
self.angle = gr.complex_to_arg()
self.sample_and_hold = gr.sample_and_hold_ff()
#ML measurements input to sampler block and detect
#self.sub1 = gr.add_const_ff(-1)
self.sub1 = gr.add_const_ff(0)
self.pk_detect = gr.peak_detector_fb(0.20, 0.20, 30, 0.001)
#self.pk_detect = gr.peak_detector2_fb(9)
self.connect(self, self.input)
# Calculate the frequency offset from the correlation of the preamble
self.connect(self.input, self.delay)
self.connect(self.input, (self.corr,0))
self.connect(self.delay, self.conjg)
self.connect(self.conjg, (self.corr,1))
self.connect(self.corr, self.moving_sum_filter)
self.connect(self.moving_sum_filter, self.c2mag)
self.connect(self.moving_sum_filter, self.angle)
self.connect(self.angle, (self.sample_and_hold,0))
# Get the power of the input signal to normalize the output of the correlation
self.connect(self.input, self.inputmag2, self.inputmovingsum)
self.connect(self.inputmovingsum, (self.square,0))
self.connect(self.inputmovingsum, (self.square,1))
self.connect(self.square, (self.normalize,1))
self.connect(self.c2mag, (self.normalize,0))
# Create a moving sum filter for the corr output
matched_filter_taps = [1.0/cp_length for i in range(cp_length)]
self.matched_filter = gr.fir_filter_fff(1,matched_filter_taps)
self.connect(self.normalize, self.matched_filter)
self.connect(self.matched_filter, self.sub1, self.pk_detect)
#self.connect(self.matched_filter, self.pk_detect)
self.connect(self.pk_detect, (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.pk_detect, (self,1))
if logging:
self.connect(self.matched_filter, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-mf_f.dat"))
self.connect(self.normalize, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-theta_f.dat"))
self.connect(self.angle, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-epsilon_f.dat"))
self.connect(self.pk_detect, gr.file_sink(gr.sizeof_char, "ofdm_sync_pn-peaks_b.dat"))
self.connect(self.sample_and_hold, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-sample_and_hold_f.dat"))
self.connect(self.input, gr.file_sink(gr.sizeof_gr_complex, "ofdm_sync_pn-input_c.dat"))
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"))
def __init__(self, fft_length, cp_length, half_sync, logging=False):
gr.hier_block2.__init__(self, "ofdm_sync_pn",
gr.io_signature(1, 1, gr.sizeof_gr_complex), # Input signature
gr.io_signature3(3, 3, # Output signature
gr.sizeof_gr_complex, # delayed input
gr.sizeof_float, # fine frequency offset
gr.sizeof_char # timing indicator
))
if half_sync:
period = fft_length/2
window = fft_length/2
else: # full symbol
period = fft_length + cp_length
window = fft_length # makes the plateau cp_length long
# Calculate the frequency offset from the correlation of the preamble
x_corr = gr.multiply_cc()
self.connect(self, gr.conjugate_cc(), (x_corr, 0))
self.connect(self, gr.delay(gr.sizeof_gr_complex, period), (x_corr, 1))
P_d = gr.moving_average_cc(window, 1.0)
self.connect(x_corr, P_d)
P_d_angle = gr.complex_to_arg()
self.connect(P_d, P_d_angle)
# Get the power of the input signal to normalize the output of the correlation
R_d = gr.moving_average_ff(window, 1.0)
self.connect(self, gr.complex_to_mag_squared(), R_d)
R_d_squared = gr.multiply_ff() # this is retarded
self.connect(R_d, (R_d_squared, 0))
self.connect(R_d, (R_d_squared, 1))
M_d = gr.divide_ff()
self.connect(P_d, gr.complex_to_mag_squared(), (M_d, 0))
self.connect(R_d_squared, (M_d, 1))
# Now we need to detect peak of M_d
# NOTE: replaced fir_filter with moving_average for clarity
# the peak is up to cp_length long, but noisy, so average it out
#matched_filter_taps = [1.0/cp_length for i in range(cp_length)]
#matched_filter = gr.fir_filter_fff(1, matched_filter_taps)
matched_filter = gr.moving_average_ff(cp_length, 1.0/cp_length)
# NOTE: the look_ahead parameter doesn't do anything
# these parameters are kind of magic, increase 1 and 2 (==) to be more tolerant
#peak_detect = raw.peak_detector_fb(0.55, 0.55, 30, 0.001)
peak_detect = raw.peak_detector_fb(0.25, 0.25, 30, 0.001)
# NOTE: gr.peak_detector_fb is broken!
#peak_detect = gr.peak_detector_fb(0.55, 0.55, 30, 0.001)
#peak_detect = gr.peak_detector_fb(0.45, 0.45, 30, 0.001)
#peak_detect = gr.peak_detector_fb(0.30, 0.30, 30, 0.001)
# offset by -1
self.connect(M_d, matched_filter, gr.add_const_ff(-1), peak_detect)
# peak_detect indicates the time M_d is highest, which is the end of the symbol.
# We should try to sample in the middle of the plateau!!
# FIXME until we figure out how to do this, just offset by cp_length/2
offset = 6 #cp_length/2
# nco(t) = P_d_angle(t-offset) sampled at peak_detect(t)
# modulate input(t - fft_length) by nco(t)
# signal to sample input(t) at t-offset
#
# We can't delay by < 0 so instead:
# input is delayed by fft_length
# P_d_angle is delayed by offset
# signal to sample is delayed by fft_length - offset
#
phi = gr.sample_and_hold_ff()
self.connect(peak_detect, (phi,1))
self.connect(P_d_angle, gr.delay(gr.sizeof_float, offset), (phi,0))
#self.connect(P_d_angle, matched_filter2, (phi,0)) # why isn't this better?!?
# FIXME: we add fft_length delay so that the preamble is nco corrected too
# BUT is this buffering worth it? consider implementing sync as a proper block
# delay the input signal to follow the frequency offset signal
self.connect(self, gr.delay(gr.sizeof_gr_complex, (fft_length+offset)), (self,0))
self.connect(phi, (self,1))
self.connect(peak_detect, (self,2))
if logging:
self.connect(matched_filter, gr.file_sink(gr.sizeof_float, "sync-mf.dat"))
self.connect(M_d, gr.file_sink(gr.sizeof_float, "sync-M.dat"))
self.connect(P_d_angle, gr.file_sink(gr.sizeof_float, "sync-angle.dat"))
self.connect(peak_detect, gr.file_sink(gr.sizeof_char, "sync-peaks.datb"))
self.connect(phi, gr.file_sink(gr.sizeof_float, "sync-phi.dat"))
def __init__(self, fft_length, cp_length, kstime, threshold, threshold_type, threshold_gap, logging=False):
"""
OFDM synchronization using PN Correlation:
T. M. Schmidl and D. C. Cox, "Robust Frequency and Timing
Synchonization for OFDM," IEEE Trans. Communications, vol. 45,
no. 12, 1997.
"""
gr.hier_block2.__init__(self, "ofdm_sync_pn",
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)
# PN Sync
# 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 corr output
if 1:
moving_sum_taps = [1.0 for i in range(fft_length//2)]
self.moving_sum_filter = gr.fir_filter_ccf(1,moving_sum_taps)
else:
moving_sum_taps = [complex(1.0,0.0) for i in range(fft_length//2)]
self.moving_sum_filter = gr.fft_filter_ccc(1,moving_sum_taps)
# Create a moving sum filter for the input
self.inputmag2 = gr.complex_to_mag_squared()
movingsum2_taps = [1.0 for i in range(fft_length//2)]
#movingsum2_taps = [0.5 for i in range(fft_length*4)] #apurv - implementing Veljo's suggestion, when pause b/w packets
if 1:
self.inputmovingsum = gr.fir_filter_fff(1,movingsum2_taps)
else:
self.inputmovingsum = gr.fft_filter_fff(1,movingsum2_taps)
self.square = gr.multiply_ff()
self.normalize = gr.divide_ff()
# Get magnitude (peaks) and angle (phase/freq error)
self.c2mag = gr.complex_to_mag_squared()
self.angle = gr.complex_to_arg()
self.sample_and_hold = gr.sample_and_hold_ff()
#ML measurements input to sampler block and detect
self.sub1 = gr.add_const_ff(-1)
self.pk_detect = gr.peak_detector_fb(0.20, 0.20, 30, 0.001) #apurv - implementing Veljo's suggestion, when pause b/w packets
self.connect(self, self.input)
# Calculate the frequency offset from the correlation of the preamble
self.connect(self.input, self.delay)
self.connect(self.input, (self.corr,0))
self.connect(self.delay, self.conjg)
self.connect(self.conjg, (self.corr,1))
self.connect(self.corr, self.moving_sum_filter)
#self.connect(self.moving_sum_filter, self.c2mag)
self.connect(self.moving_sum_filter, self.angle)
self.connect(self.angle, (self.sample_and_hold,0)) # apurv--
#self.connect(self.angle, gr.delay(gr.sizeof_float, offset), (self.sample_and_hold, 0)) #apurv++
cross_correlate = 1
if cross_correlate==1:
# cross-correlate with the known symbol
kstime = [k.conjugate() for k in kstime]
kstime.reverse()
self.crosscorr_filter = gr.fir_filter_ccc(1, kstime)
# get the magnitude #
self.corrmag = gr.complex_to_mag_squared()
self.f2b = gr.float_to_char()
self.threshold_factor = threshold #0.0012 #0.012 #0.0015
if 0:
self.slice = gr.threshold_ff(self.threshold_factor, self.threshold_factor, 0, fft_length)
else:
#thresholds = [self.threshold_factor, 9e-5]
self.slice = gr.threshold_ff(threshold, threshold, 0, fft_length, threshold_type, threshold_gap)
self.connect(self.input, self.crosscorr_filter, self.corrmag, self.slice, self.f2b)
# some debug dump #
self.connect(self.corrmag, gr.file_sink(gr.sizeof_float, "ofdm_corrmag.dat"))
#self.connect(self.f2b, gr.file_sink(gr.sizeof_char, "ofdm_f2b.dat"))
self.connect(self.f2b, (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.pk_detect, (self,1)) #removed
#self.connect(self.f2b, gr.delay(gr.sizeof_char, 1), (self, 1))
self.connect(self.f2b, (self, 1))
if logging:
self.connect(self.matched_filter, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-mf_f.dat"))
self.connect(self.normalize, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-theta_f.dat"))
self.connect(self.angle, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-epsilon_f.dat"))
self.connect(self.pk_detect, gr.file_sink(gr.sizeof_char, "ofdm_sync_pn-peaks_b.dat"))
self.connect(self.sample_and_hold, gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-sample_and_hold_f.dat"))
self.connect(self.input, gr.file_sink(gr.sizeof_gr_complex, "ofdm_sync_pn-input_c.dat"))
def __init__(self,
fft_length,
cp_length,
kstime,
threshold,
threshold_type,
threshold_gap,
logging=False):
"""
OFDM synchronization using PN Correlation:
T. M. Schmidl and D. C. Cox, "Robust Frequency and Timing
Synchonization for OFDM," IEEE Trans. Communications, vol. 45,
no. 12, 1997.
"""
gr.hier_block2.__init__(
self,
"ofdm_sync_pn",
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)
# PN Sync
# 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 corr output
if 1:
moving_sum_taps = [1.0 for i in range(fft_length // 2)]
self.moving_sum_filter = gr.fir_filter_ccf(1, moving_sum_taps)
else:
moving_sum_taps = [
complex(1.0, 0.0) for i in range(fft_length // 2)
]
self.moving_sum_filter = gr.fft_filter_ccc(1, moving_sum_taps)
# Create a moving sum filter for the input
self.inputmag2 = gr.complex_to_mag_squared()
movingsum2_taps = [1.0 for i in range(fft_length // 2)]
#movingsum2_taps = [0.5 for i in range(fft_length*4)] #apurv - implementing Veljo's suggestion, when pause b/w packets
if 1:
self.inputmovingsum = gr.fir_filter_fff(1, movingsum2_taps)
else:
self.inputmovingsum = gr.fft_filter_fff(1, movingsum2_taps)
self.square = gr.multiply_ff()
self.normalize = gr.divide_ff()
# Get magnitude (peaks) and angle (phase/freq error)
self.c2mag = gr.complex_to_mag_squared()
self.angle = gr.complex_to_arg()
self.sample_and_hold = gr.sample_and_hold_ff()
#ML measurements input to sampler block and detect
self.sub1 = gr.add_const_ff(-1)
self.pk_detect = gr.peak_detector_fb(
0.20, 0.20, 30, 0.001
) #apurv - implementing Veljo's suggestion, when pause b/w packets
self.connect(self, self.input)
# Calculate the frequency offset from the correlation of the preamble
self.connect(self.input, self.delay)
self.connect(self.input, (self.corr, 0))
self.connect(self.delay, self.conjg)
self.connect(self.conjg, (self.corr, 1))
self.connect(self.corr, self.moving_sum_filter)
#self.connect(self.moving_sum_filter, self.c2mag)
self.connect(self.moving_sum_filter, self.angle)
self.connect(self.angle, (self.sample_and_hold, 0)) # apurv--
#self.connect(self.angle, gr.delay(gr.sizeof_float, offset), (self.sample_and_hold, 0)) #apurv++
cross_correlate = 1
if cross_correlate == 1:
# cross-correlate with the known symbol
kstime = [k.conjugate() for k in kstime]
kstime.reverse()
self.crosscorr_filter = gr.fir_filter_ccc(1, kstime)
# get the magnitude #
self.corrmag = gr.complex_to_mag_squared()
self.f2b = gr.float_to_char()
self.threshold_factor = threshold #0.0012 #0.012 #0.0015
if 0:
self.slice = gr.threshold_ff(self.threshold_factor,
self.threshold_factor, 0,
fft_length)
else:
#thresholds = [self.threshold_factor, 9e-5]
self.slice = gr.threshold_ff(threshold, threshold, 0,
fft_length, threshold_type,
threshold_gap)
self.connect(self.input, self.crosscorr_filter, self.corrmag,
self.slice, self.f2b)
# some debug dump #
self.connect(self.corrmag,
gr.file_sink(gr.sizeof_float, "ofdm_corrmag.dat"))
#self.connect(self.f2b, gr.file_sink(gr.sizeof_char, "ofdm_f2b.dat"))
self.connect(self.f2b, (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.pk_detect, (self,1)) #removed
#self.connect(self.f2b, gr.delay(gr.sizeof_char, 1), (self, 1))
self.connect(self.f2b, (self, 1))
if logging:
self.connect(
self.matched_filter,
gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-mf_f.dat"))
self.connect(
self.normalize,
gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-theta_f.dat"))
self.connect(
self.angle,
gr.file_sink(gr.sizeof_float, "ofdm_sync_pn-epsilon_f.dat"))
self.connect(
self.pk_detect,
gr.file_sink(gr.sizeof_char, "ofdm_sync_pn-peaks_b.dat"))
self.connect(
self.sample_and_hold,
gr.file_sink(gr.sizeof_float,
"ofdm_sync_pn-sample_and_hold_f.dat"))
self.connect(
self.input,
gr.file_sink(gr.sizeof_gr_complex, "ofdm_sync_pn-input_c.dat"))
def __init__(self, mode, debug=False):
"""
OFDM time and coarse frequency synchronisation for DAB
@param mode DAB mode (1-4)
@param debug if True: write data streams out to files
"""
if mode < 1 or mode > 4:
raise ValueError, "Invalid DAB mode: " + str(
mode) + " (modes 1-4 exist)"
# get the correct DAB parameters
dp = parameters.dab_parameters(mode)
rp = parameters.receiver_parameters(mode)
gr.hier_block2.__init__(
self,
"ofdm_sync_dab",
gr.io_signature(1, 1, gr.sizeof_gr_complex), # input signature
gr.io_signature2(2, 2, gr.sizeof_gr_complex,
gr.sizeof_char)) # output signature
# workaround for a problem that prevents connecting more than one block directly (see trac ticket #161)
self.input = gr.kludge_copy(gr.sizeof_gr_complex)
self.connect(self, self.input)
#
# null-symbol detection
#
# (outsourced to detect_zero.py)
self.ns_detect = detect_null.detect_null(dp.ns_length, debug)
self.connect(self.input, self.ns_detect)
#
# fine frequency synchronisation
#
# the code for fine frequency synchronisation is adapted from
# ofdm_sync_ml.py; it abuses the cyclic prefix to find the fine
# frequency error, as suggested in "ML Estimation of Timing and
# Frequency Offset in OFDM Systems", by Jan-Jaap van de Beek,
# Magnus Sandell, Per Ola Börjesson, see
# http://www.sm.luth.se/csee/sp/research/report/bsb96r.html
self.ffs_delay = gr.delay(gr.sizeof_gr_complex, dp.fft_length)
self.ffs_conj = gr.conjugate_cc()
self.ffs_mult = gr.multiply_cc()
# self.ffs_moving_sum = gr.fir_filter_ccf(1, [1]*dp.cp_length)
self.ffs_moving_sum = dab_swig.moving_sum_cc(dp.cp_length)
self.ffs_angle = gr.complex_to_arg()
self.ffs_angle_scale = gr.multiply_const_ff(1. / dp.fft_length)
self.ffs_delay_sample_and_hold = gr.delay(
gr.sizeof_char,
dp.symbol_length) # sample the value at the end of the symbol ..
self.ffs_sample_and_hold = gr.sample_and_hold_ff()
self.ffs_delay_input_for_correction = gr.delay(
gr.sizeof_gr_complex, dp.symbol_length
) # by delaying the input, we can use the ff offset estimation from the first symbol to correct the first symbol itself
self.ffs_nco = gr.frequency_modulator_fc(
1) # ffs_sample_and_hold directly outputs phase error per sample
self.ffs_mixer = gr.multiply_cc()
# calculate fine frequency error
self.connect(self.input, self.ffs_conj, self.ffs_mult)
self.connect(self.input, self.ffs_delay, (self.ffs_mult, 1))
self.connect(self.ffs_mult, self.ffs_moving_sum, self.ffs_angle)
# only use the value from the first half of the first symbol
self.connect(self.ffs_angle, self.ffs_angle_scale,
(self.ffs_sample_and_hold, 0))
self.connect(self.ns_detect, self.ffs_delay_sample_and_hold,
(self.ffs_sample_and_hold, 1))
# do the correction
self.connect(self.ffs_sample_and_hold, self.ffs_nco,
(self.ffs_mixer, 0))
self.connect(self.input, self.ffs_delay_input_for_correction,
(self.ffs_mixer, 1))
# output - corrected signal and start of DAB frames
self.connect(self.ffs_mixer, (self, 0))
self.connect(self.ffs_delay_sample_and_hold, (self, 1))
if debug:
self.connect(
self.ffs_angle,
gr.file_sink(gr.sizeof_float,
"debug/ofdm_sync_dab_ffs_angle.dat"))
self.connect(
self.ffs_sample_and_hold,
gr.multiply_const_ff(1. / (dp.T * 2 * pi)),
gr.file_sink(gr.sizeof_float,
"debug/ofdm_sync_dab_fine_freq_err_f.dat"))
self.connect(
self.ffs_mixer,
gr.file_sink(gr.sizeof_gr_complex,
"debug/ofdm_sync_dab_fine_freq_corrected_c.dat"))
def __init__(self, mode, debug=False): """ OFDM time and coarse frequency synchronisation for DAB @param mode DAB mode (1-4) @param debug if True: write data streams out to files """ if mode<1 or mode>4: raise ValueError, "Invalid DAB mode: "+str(mode)+" (modes 1-4 exist)" # get the correct DAB parameters dp = parameters.dab_parameters(mode) rp = parameters.receiver_parameters(mode) gr.hier_block2.__init__(self,"ofdm_sync_dab", gr.io_signature(1, 1, gr.sizeof_gr_complex), # input signature gr.io_signature2(2, 2, gr.sizeof_gr_complex, gr.sizeof_char)) # output signature # workaround for a problem that prevents connecting more than one block directly (see trac ticket #161) self.input = gr.kludge_copy(gr.sizeof_gr_complex) self.connect(self, self.input) # # null-symbol detection # # (outsourced to detect_zero.py) self.ns_detect = detect_null.detect_null(dp.ns_length, debug) self.connect(self.input, self.ns_detect) # # fine frequency synchronisation # # the code for fine frequency synchronisation is adapted from # ofdm_sync_ml.py; it abuses the cyclic prefix to find the fine # frequency error, as suggested in "ML Estimation of Timing and # Frequency Offset in OFDM Systems", by Jan-Jaap van de Beek, # Magnus Sandell, Per Ola Börjesson, see # http://www.sm.luth.se/csee/sp/research/report/bsb96r.html self.ffs_delay = gr.delay(gr.sizeof_gr_complex, dp.fft_length) self.ffs_conj = gr.conjugate_cc() self.ffs_mult = gr.multiply_cc() # self.ffs_moving_sum = gr.fir_filter_ccf(1, [1]*dp.cp_length) self.ffs_moving_sum = dab_swig.moving_sum_cc(dp.cp_length) self.ffs_angle = gr.complex_to_arg() self.ffs_angle_scale = gr.multiply_const_ff(1./dp.fft_length) self.ffs_delay_sample_and_hold = gr.delay(gr.sizeof_char, dp.symbol_length) # sample the value at the end of the symbol .. self.ffs_sample_and_hold = gr.sample_and_hold_ff() self.ffs_delay_input_for_correction = gr.delay(gr.sizeof_gr_complex, dp.symbol_length) # by delaying the input, we can use the ff offset estimation from the first symbol to correct the first symbol itself self.ffs_nco = gr.frequency_modulator_fc(1) # ffs_sample_and_hold directly outputs phase error per sample self.ffs_mixer = gr.multiply_cc() # calculate fine frequency error self.connect(self.input, self.ffs_conj, self.ffs_mult) self.connect(self.input, self.ffs_delay, (self.ffs_mult, 1)) self.connect(self.ffs_mult, self.ffs_moving_sum, self.ffs_angle) # only use the value from the first half of the first symbol self.connect(self.ffs_angle, self.ffs_angle_scale, (self.ffs_sample_and_hold, 0)) self.connect(self.ns_detect, self.ffs_delay_sample_and_hold, (self.ffs_sample_and_hold, 1)) # do the correction self.connect(self.ffs_sample_and_hold, self.ffs_nco, (self.ffs_mixer, 0)) self.connect(self.input, self.ffs_delay_input_for_correction, (self.ffs_mixer, 1)) # output - corrected signal and start of DAB frames self.connect(self.ffs_mixer, (self, 0)) self.connect(self.ffs_delay_sample_and_hold, (self, 1)) if debug: self.connect(self.ffs_angle, gr.file_sink(gr.sizeof_float, "debug/ofdm_sync_dab_ffs_angle.dat")) self.connect(self.ffs_sample_and_hold, gr.multiply_const_ff(1./(dp.T*2*pi)), gr.file_sink(gr.sizeof_float, "debug/ofdm_sync_dab_fine_freq_err_f.dat")) self.connect(self.ffs_mixer, gr.file_sink(gr.sizeof_gr_complex, "debug/ofdm_sync_dab_fine_freq_corrected_c.dat"))
