def test_001_detect(self):
""" Send two bursts, with zeros in between, and check
they are both detected at the correct position and no
false alarms occur """
n_zeros = 15
fft_len = 32
cp_len = 4
sig_len = (fft_len + cp_len) * 10
sync_symbol = [(random.randint(0, 1) * 2) - 1
for x in range(fft_len / 2)] * 2
tx_signal = [0,] * n_zeros + \
sync_symbol[-cp_len:] + \
sync_symbol + \
[(random.randint(0, 1)*2)-1 for x in range(sig_len)]
tx_signal = tx_signal * 2
add = blocks.add_cc()
sync = digital.ofdm_sync_sc_cfb(fft_len, cp_len)
sink_freq = blocks.vector_sink_f()
sink_detect = blocks.vector_sink_b()
self.tb.connect(blocks.vector_source_c(tx_signal), (add, 0))
self.tb.connect(analog.noise_source_c(analog.GR_GAUSSIAN, .01),
(add, 1))
self.tb.connect(add, sync)
self.tb.connect((sync, 0), sink_freq)
self.tb.connect((sync, 1), sink_detect)
self.tb.run()
sig1_detect = sink_detect.data()[0:len(tx_signal) / 2]
sig2_detect = sink_detect.data()[len(tx_signal) / 2:]
self.assertTrue(
abs(sig1_detect.index(1) - (n_zeros + fft_len + cp_len)) < cp_len)
self.assertTrue(
abs(sig2_detect.index(1) - (n_zeros + fft_len + cp_len)) < cp_len)
self.assertEqual(numpy.sum(sig1_detect), 1)
self.assertEqual(numpy.sum(sig2_detect), 1)
def test_001_detect (self):
""" Send two bursts, with zeros in between, and check
they are both detected at the correct position and no
false alarms occur """
n_zeros = 15
fft_len = 32
cp_len = 4
sig_len = (fft_len + cp_len) * 10
sync_symbol = [(random.randint(0, 1)*2)-1 for x in range(fft_len/2)] * 2
tx_signal = [0,] * n_zeros + \
sync_symbol[-cp_len:] + \
sync_symbol + \
[(random.randint(0, 1)*2)-1 for x in range(sig_len)]
tx_signal = tx_signal * 2
add = blocks.add_cc()
sync = digital.ofdm_sync_sc_cfb(fft_len, cp_len)
sink_freq = blocks.vector_sink_f()
sink_detect = blocks.vector_sink_b()
self.tb.connect(blocks.vector_source_c(tx_signal), (add, 0))
self.tb.connect(analog.noise_source_c(analog.GR_GAUSSIAN, .01), (add, 1))
self.tb.connect(add, sync)
self.tb.connect((sync, 0), sink_freq)
self.tb.connect((sync, 1), sink_detect)
self.tb.run()
sig1_detect = sink_detect.data()[0:len(tx_signal)/2]
sig2_detect = sink_detect.data()[len(tx_signal)/2:]
self.assertTrue(abs(sig1_detect.index(1) - (n_zeros + fft_len + cp_len)) < cp_len)
self.assertTrue(abs(sig2_detect.index(1) - (n_zeros + fft_len + cp_len)) < cp_len)
self.assertEqual(numpy.sum(sig1_detect), 1)
self.assertEqual(numpy.sum(sig2_detect), 1)
def run_flow_graph(sync_sym1, sync_sym2, data_sym):
top_block = gr.top_block()
carr_offset = random.randint(-max_offset/2, max_offset/2) * 2
tx_data = shift_tuple(sync_sym1, carr_offset) + \
shift_tuple(sync_sym2, carr_offset) + \
shift_tuple(data_sym, carr_offset)
channel = [rand_range(min_chan_ampl, max_chan_ampl) * numpy.exp(1j * rand_range(0, 2 * numpy.pi)) for x in range(fft_len)]
src = blocks.vector_source_c(tx_data, False, fft_len)
chan = blocks.multiply_const_vcc(channel)
noise = analog.noise_source_c(analog.GR_GAUSSIAN, wgn_amplitude)
add = blocks.add_cc(fft_len)
chanest = digital.ofdm_chanest_vcvc(sync_sym1, sync_sym2, 1)
sink = blocks.vector_sink_c(fft_len)
top_block.connect(src, chan, (add, 0), chanest, sink)
top_block.connect(noise, blocks.stream_to_vector(gr.sizeof_gr_complex, fft_len), (add, 1))
top_block.run()
channel_est = None
carr_offset_hat = 0
rx_sym_est = [0,] * fft_len
tags = sink.tags()
for tag in tags:
if pmt.symbol_to_string(tag.key) == 'ofdm_sync_carr_offset':
carr_offset_hat = pmt.to_long(tag.value)
self.assertEqual(carr_offset, carr_offset_hat)
if pmt.symbol_to_string(tag.key) == 'ofdm_sync_chan_taps':
channel_est = shift_tuple(pmt.c32vector_elements(tag.value), carr_offset)
shifted_carrier_mask = shift_tuple(carrier_mask, carr_offset)
for i in range(fft_len):
if shifted_carrier_mask[i] and channel_est[i]:
self.assertAlmostEqual(channel[i], channel_est[i], places=0)
rx_sym_est[i] = (sink.data()[i] / channel_est[i]).real
return (carr_offset, list(shift_tuple(rx_sym_est, -carr_offset_hat)))
def __init__(self, constellation, f, N0=0.25, seed=-666L):
"""
constellation - a constellation object used for modulation.
f - a finite state machine specification used for coding.
N0 - noise level
seed - random seed
"""
super(trellis_tb, self).__init__()
# packet size in bits (make it multiple of 16 so it can be packed in a short)
packet_size = 1024 * 16
# bits per FSM input symbol
bitspersymbol = int(round(math.log(f.I()) /
math.log(2))) # bits per FSM input symbol
# packet size in trellis steps
K = packet_size / bitspersymbol
# TX
src = blocks.lfsr_32k_source_s()
# packet size in shorts
src_head = blocks.head(gr.sizeof_short, packet_size / 16)
# unpack shorts to symbols compatible with the FSM input cardinality
s2fsmi = blocks.packed_to_unpacked_ss(bitspersymbol, gr.GR_MSB_FIRST)
# initial FSM state = 0
enc = trellis.encoder_ss(f, 0)
mod = digital.chunks_to_symbols_sc(constellation.points(), 1)
# CHANNEL
add = blocks.add_cc()
noise = analog.noise_source_c(analog.GR_GAUSSIAN, math.sqrt(N0 / 2),
seed)
# RX
# data preprocessing to generate metrics for Viterbi
metrics = trellis.constellation_metrics_cf(constellation.base(),
digital.TRELLIS_EUCLIDEAN)
# Put -1 if the Initial/Final states are not set.
va = trellis.viterbi_s(f, K, 0, -1)
# pack FSM input symbols to shorts
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol, gr.GR_MSB_FIRST)
# check the output
self.dst = blocks.check_lfsr_32k_s()
self.connect(src, src_head, s2fsmi, enc, mod)
self.connect(mod, (add, 0))
self.connect(noise, (add, 1))
self.connect(add, metrics, va, fsmi2s, self.dst)
def __init__(self, noise_voltage, freq, timing):
gr.hier_block2.__init__(self, "channel_model",
gr.io_signature(1, 1, gr.sizeof_gr_complex),
gr.io_signature(1, 1, gr.sizeof_gr_complex))
timing_offset = filter.fractional_resampler_cc(0, timing)
noise_adder = blocks.add_cc()
noise = analog.noise_source_c(analog.GR_GAUSSIAN, noise_voltage, 0)
freq_offset = analog.sig_source_c(1, analog.GR_SIN_WAVE, freq, 1.0,
0.0)
mixer_offset = blocks.multiply_cc()
self.connect(self, timing_offset)
self.connect(timing_offset, (mixer_offset, 0))
self.connect(freq_offset, (mixer_offset, 1))
self.connect(mixer_offset, (noise_adder, 1))
self.connect(noise, (noise_adder, 0))
self.connect(noise_adder, self)
def run_flow_graph(sync_sym1, sync_sym2, data_sym):
top_block = gr.top_block()
carr_offset = random.randint(-max_offset / 2, max_offset / 2) * 2
tx_data = shift_tuple(sync_sym1, carr_offset) + \
shift_tuple(sync_sym2, carr_offset) + \
shift_tuple(data_sym, carr_offset)
channel = [
rand_range(min_chan_ampl, max_chan_ampl) *
numpy.exp(1j * rand_range(0, 2 * numpy.pi))
for x in range(fft_len)
]
src = blocks.vector_source_c(tx_data, False, fft_len)
chan = blocks.multiply_const_vcc(channel)
noise = analog.noise_source_c(analog.GR_GAUSSIAN, wgn_amplitude)
add = blocks.add_cc(fft_len)
chanest = digital.ofdm_chanest_vcvc(sync_sym1, sync_sym2, 1)
sink = blocks.vector_sink_c(fft_len)
top_block.connect(src, chan, (add, 0), chanest, sink)
top_block.connect(
noise, blocks.stream_to_vector(gr.sizeof_gr_complex, fft_len),
(add, 1))
top_block.run()
channel_est = None
carr_offset_hat = 0
rx_sym_est = [
0,
] * fft_len
tags = sink.tags()
for tag in tags:
if pmt.symbol_to_string(tag.key) == 'ofdm_sync_carr_offset':
carr_offset_hat = pmt.to_long(tag.value)
self.assertEqual(carr_offset, carr_offset_hat)
if pmt.symbol_to_string(tag.key) == 'ofdm_sync_chan_taps':
channel_est = shift_tuple(
pmt.c32vector_elements(tag.value), carr_offset)
shifted_carrier_mask = shift_tuple(carrier_mask, carr_offset)
for i in range(fft_len):
if shifted_carrier_mask[i] and channel_est[i]:
self.assertAlmostEqual(channel[i],
channel_est[i],
places=0)
rx_sym_est[i] = (sink.data()[i] / channel_est[i]).real
return (carr_offset, list(shift_tuple(rx_sym_est,
-carr_offset_hat)))
def __init__(self, constellation, f, N0=0.25, seed=-666L):
"""
constellation - a constellation object used for modulation.
f - a finite state machine specification used for coding.
N0 - noise level
seed - random seed
"""
super(trellis_tb, self).__init__()
# packet size in bits (make it multiple of 16 so it can be packed in a short)
packet_size = 1024 * 16
# bits per FSM input symbol
bitspersymbol = int(round(math.log(f.I()) / math.log(2))) # bits per FSM input symbol
# packet size in trellis steps
K = packet_size / bitspersymbol
# TX
src = blocks.lfsr_32k_source_s()
# packet size in shorts
src_head = blocks.head(gr.sizeof_short, packet_size / 16)
# unpack shorts to symbols compatible with the FSM input cardinality
s2fsmi = blocks.packed_to_unpacked_ss(bitspersymbol, gr.GR_MSB_FIRST)
# initial FSM state = 0
enc = trellis.encoder_ss(f, 0)
mod = digital.chunks_to_symbols_sc(constellation.points(), 1)
# CHANNEL
add = blocks.add_cc()
noise = analog.noise_source_c(analog.GR_GAUSSIAN, math.sqrt(N0 / 2), seed)
# RX
# data preprocessing to generate metrics for Viterbi
metrics = trellis.constellation_metrics_cf(constellation.base(), digital.TRELLIS_EUCLIDEAN)
# Put -1 if the Initial/Final states are not set.
va = trellis.viterbi_s(f, K, 0, -1)
# pack FSM input symbols to shorts
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol, gr.GR_MSB_FIRST)
# check the output
self.dst = blocks.check_lfsr_32k_s()
self.connect(src, src_head, s2fsmi, enc, mod)
self.connect(mod, (add, 0))
self.connect(noise, (add, 1))
self.connect(add, metrics, va, fsmi2s, self.dst)
def __init__(self, noise_voltage, freq, timing):
gr.hier_block2.__init__(
self,
"channel_model",
gr.io_signature(1, 1, gr.sizeof_gr_complex),
gr.io_signature(1, 1, gr.sizeof_gr_complex),
)
timing_offset = filter.fractional_interpolator_cc(0, timing)
noise_adder = blocks.add_cc()
noise = analog.noise_source_c(analog.GR_GAUSSIAN, noise_voltage, 0)
freq_offset = analog.sig_source_c(1, analog.GR_SIN_WAVE, freq, 1.0, 0.0)
mixer_offset = blocks.multiply_cc()
self.connect(self, timing_offset)
self.connect(timing_offset, (mixer_offset, 0))
self.connect(freq_offset, (mixer_offset, 1))
self.connect(mixer_offset, (noise_adder, 1))
self.connect(noise, (noise_adder, 0))
self.connect(noise_adder, self)
