def test_sf_tag(self):
constA = [-3.0, -1.0, 1.0, 3]
constB = [12.0, -12.0, 6.0, -6]
src_data = (0, 1, 2, 3, 3, 2, 1, 0)
expected_result = (-3, -1, 1, 3,
-6, 6, -12, 12)
first_tag = gr.tag_t()
first_tag.key = pmt.intern("set_symbol_table")
first_tag.value = pmt.init_f32vector(len(constA), constA)
first_tag.offset = 0
second_tag = gr.tag_t()
second_tag.key = pmt.intern("set_symbol_table")
second_tag.value = pmt.init_f32vector(len(constB), constB)
second_tag.offset = 4
src = blocks.vector_source_s(src_data, False, 1, [first_tag, second_tag])
op = digital.chunks_to_symbols_sf(constB)
dst = blocks.vector_sink_f()
self.tb.connect(src, op)
self.tb.connect(op, dst)
self.tb.run()
actual_result = dst.data()
self.assertEqual(expected_result, actual_result)
def test_sf_tag(self):
constA = [-3.0, -1.0, 1.0, 3]
constB = [12.0, -12.0, 6.0, -6]
src_data = (0, 1, 2, 3, 3, 2, 1, 0)
expected_result = [-3, -1, 1, 3, -6, 6, -12, 12]
first_tag = gr.tag_t()
first_tag.key = pmt.intern("set_symbol_table")
first_tag.value = pmt.init_f32vector(len(constA), constA)
first_tag.offset = 0
second_tag = gr.tag_t()
second_tag.key = pmt.intern("set_symbol_table")
second_tag.value = pmt.init_f32vector(len(constB), constB)
second_tag.offset = 4
src = blocks.vector_source_s(src_data, False, 1,
[first_tag, second_tag])
op = digital.chunks_to_symbols_sf(constB)
dst = blocks.vector_sink_f()
self.tb.connect(src, op)
self.tb.connect(op, dst)
self.tb.run()
actual_result = dst.data()
self.assertEqual(expected_result, actual_result)
def run_test (f,Kb,bitspersymbol,K,dimensionality,tot_constellation,N0,seed):
tb = gr.top_block ()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head (gr.sizeof_short,Kb/16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(bitspersymbol,gr.GR_MSB_FIRST) # unpack shorts to symbols compatible with the FSM input cardinality
enc = trellis.encoder_ss(f,0) # initial state = 0
# essentially here we implement the combination of modulation and channel as a memoryless modulation (the memory induced by the channel is hidden in the FSM)
mod = digital.chunks_to_symbols_sf(tot_constellation,dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN,math.sqrt(N0/2),seed)
# RX
metrics = trellis.metrics_f(f.O(),dimensionality,tot_constellation,digital.TRELLIS_EUCLIDEAN) # data preprocessing to generate metrics for Viterbi
va = trellis.viterbi_s(f,K,0,-1) # Put -1 if the Initial/Final states are not set.
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol,gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s();
tb.connect (src,src_head,s2fsmi,enc,mod)
tb.connect (mod,(add,0))
tb.connect (noise,(add,1))
tb.connect (add,metrics)
tb.connect (metrics,va,fsmi2s,dst)
tb.run()
ntotal = dst.ntotal ()
nright = dst.nright ()
runlength = dst.runlength ()
#print ntotal,nright,runlength
return (ntotal,ntotal-nright)
def run_test (fo,fi,interleaver,Kb,bitspersymbol,K,dimensionality,constellation,Es,N0,IT,seed):
tb = gr.top_block ()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short,Kb/16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(bitspersymbol,gr.GR_MSB_FIRST) # unpack shorts to symbols compatible with the outer FSM input cardinality
enc = trellis.sccc_encoder_ss(fo,0,fi,0,interleaver,K)
mod = digital.chunks_to_symbols_sf(constellation,dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN,math.sqrt(N0/2),seed)
# RX
dec = trellis.sccc_decoder_combined_fs(fo,0,-1,fi,0,-1,interleaver,K,IT,trellis.TRELLIS_MIN_SUM,dimensionality,constellation,digital.TRELLIS_EUCLIDEAN,1.0)
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol,gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s()
#tb.connect (src,src_head,s2fsmi,enc_out,inter,enc_in,mod)
tb.connect (src,src_head,s2fsmi,enc,mod)
tb.connect (mod,(add,0))
tb.connect (noise,(add,1))
#tb.connect (add,head)
#tb.connect (tail,fsmi2s,dst)
tb.connect (add,dec,fsmi2s,dst)
tb.run()
#print enc_out.ST(), enc_in.ST()
ntotal = dst.ntotal ()
nright = dst.nright ()
runlength = dst.runlength ()
return (ntotal,ntotal-nright)
def run_test (f,Kb,bitspersymbol,K,dimensionality,constellation,N0,seed):
tb = gr.top_block ()
# TX
#packet = [0]*Kb
#for i in range(Kb-1*16): # last 16 bits = 0 to drive the final state to 0
#packet[i] = random.randint(0, 1) # random 0s and 1s
#src = blocks.vector_source_s(packet,False)
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short,Kb/16) # packet size in shorts
#b2s = blocks.unpacked_to_packed_ss(1,gr.GR_MSB_FIRST) # pack bits in shorts
s2fsmi = blocks.packed_to_unpacked_ss(bitspersymbol,gr.GR_MSB_FIRST) # unpack shorts to symbols compatible with the FSM input cardinality
enc = trellis.encoder_ss(f,0) # initial state = 0
mod = digital.chunks_to_symbols_sf(constellation,dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN,math.sqrt(N0/2),seed)
# RX
metrics = trellis.metrics_f(f.O(),dimensionality,constellation,digital.TRELLIS_EUCLIDEAN) # data preprocessing to generate metrics for Viterbi
va = trellis.viterbi_s(f,K,0,-1) # Put -1 if the Initial/Final states are not set.
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol,gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
#s2b = blocks.packed_to_unpacked_ss(1,gr.GR_MSB_FIRST) # unpack shorts to bits
#dst = blocks.vector_sink_s();
dst = blocks.check_lfsr_32k_s()
tb.connect (src,src_head,s2fsmi,enc,mod)
#tb.connect (src,b2s,s2fsmi,enc,mod)
tb.connect (mod,(add,0))
tb.connect (noise,(add,1))
tb.connect (add,metrics)
tb.connect (metrics,va,fsmi2s,dst)
#tb.connect (metrics,va,fsmi2s,s2b,dst)
tb.run()
# A bit of cheating: run the program once and print the
# final encoder state..
# Then put it as the last argument in the viterbi block
#print "final state = " , enc.ST()
ntotal = dst.ntotal ()
nright = dst.nright ()
runlength = dst.runlength ()
#ntotal = len(packet)
#if len(dst.data()) != ntotal:
#print "Error: not enough data\n"
#nright = 0;
#for i in range(ntotal):
#if packet[i]==dst.data()[i]:
#nright=nright+1
#else:
#print "Error in ", i
return (ntotal,ntotal-nright)
def run_test(f, Kb, bitspersymbol, K, channel, modulation, dimensionality,
tot_constellation, N0, seed):
tb = gr.top_block()
L = len(channel)
# TX
# this for loop is TOO slow in python!!!
packet = [0] * (K + 2 * L)
random.seed(seed)
for i in range(len(packet)):
packet[i] = random.randint(0, 2**bitspersymbol - 1) # random symbols
for i in range(L): # first/last L symbols set to 0
packet[i] = 0
packet[len(packet) - i - 1] = 0
src = blocks.vector_source_s(packet, False)
mod = digital.chunks_to_symbols_sf(modulation[1], modulation[0])
# CHANNEL
isi = filter.fir_filter_fff(1, channel)
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN, math.sqrt(N0 / 2), seed)
# RX
skip = blocks.skiphead(
gr.sizeof_float, L
) # skip the first L samples since you know they are coming from the L zero symbols
#metrics = trellis.metrics_f(f.O(),dimensionality,tot_constellation,digital.TRELLIS_EUCLIDEAN) # data preprocessing to generate metrics for Viterbi
#va = trellis.viterbi_s(f,K+L,0,0) # Put -1 if the Initial/Final states are not set.
va = trellis.viterbi_combined_s(
f, K + L, 0, 0, dimensionality, tot_constellation,
digital.TRELLIS_EUCLIDEAN
) # using viterbi_combined_s instead of metrics_f/viterbi_s allows larger packet lengths because metrics_f is complaining for not being able to allocate large buffers. This is due to the large f.O() in this application...
dst = blocks.vector_sink_s()
tb.connect(src, mod)
tb.connect(mod, isi, (add, 0))
tb.connect(noise, (add, 1))
#tb.connect (add,metrics)
#tb.connect (metrics,va,dst)
tb.connect(add, skip, va, dst)
tb.run()
data = dst.data()
ntotal = len(data) - L
nright = 0
for i in range(ntotal):
if packet[i + L] == data[i]:
nright = nright + 1
#else:
#print "Error in ", i
return (ntotal, ntotal - nright)
def run_test(f, Kb, bitspersymbol, K, channel, modulation, dimensionality, tot_constellation, N0, seed):
tb = gr.top_block()
L = len(channel)
# TX
# this for loop is TOO slow in python!!!
packet = [0] * (K + 2 * L)
random.seed(seed)
for i in range(len(packet)):
packet[i] = random.randint(0, 2 ** bitspersymbol - 1) # random symbols
for i in range(L): # first/last L symbols set to 0
packet[i] = 0
packet[len(packet) - i - 1] = 0
src = blocks.vector_source_s(packet, False)
mod = digital.chunks_to_symbols_sf(modulation[1], modulation[0])
# CHANNEL
isi = filter.fir_filter_fff(1, channel)
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN, math.sqrt(N0 / 2), seed)
# RX
skip = blocks.skiphead(
gr.sizeof_float, L
) # skip the first L samples since you know they are coming from the L zero symbols
# metrics = trellis.metrics_f(f.O(),dimensionality,tot_constellation,digital.TRELLIS_EUCLIDEAN) # data preprocessing to generate metrics for Viterbi
# va = trellis.viterbi_s(f,K+L,0,0) # Put -1 if the Initial/Final states are not set.
va = trellis.viterbi_combined_s(
f, K + L, 0, 0, dimensionality, tot_constellation, digital.TRELLIS_EUCLIDEAN
) # using viterbi_combined_s instead of metrics_f/viterbi_s allows larger packet lengths because metrics_f is complaining for not being able to allocate large buffers. This is due to the large f.O() in this application...
dst = blocks.vector_sink_s()
tb.connect(src, mod)
tb.connect(mod, isi, (add, 0))
tb.connect(noise, (add, 1))
# tb.connect (add,metrics)
# tb.connect (metrics,va,dst)
tb.connect(add, skip, va, dst)
tb.run()
data = dst.data()
ntotal = len(data) - L
nright = 0
for i in range(ntotal):
if packet[i + L] == data[i]:
nright = nright + 1
# else:
# print "Error in ", i
return (ntotal, ntotal - nright)
def __init__(self, ts, factor, alpha, samp_rate, freqs):
gr.hier_block2.__init__(
self, "freq_timing_estimator_hier",
gr.io_signature(1, 1, gr.sizeof_gr_complex*1),
gr.io_signaturev(3, 3, [gr.sizeof_char*1, gr.sizeof_float*1, gr.sizeof_float*1]),
)
##################################################
# Parameters
##################################################
self.ts = ts
self.factor = factor
self.alpha = alpha
self.samp_rate = samp_rate
self.freqs = freqs
self.n = n = len(freqs)
##################################################
# Blocks
##################################################
self._filter=[0]*self.n
self._c2mag2=[0]*self.n
for i in range(self.n):
self._filter[i]= filter.freq_xlating_fir_filter_ccc(1, (numpy.conjugate(self.ts[::-1])), self.freqs[i], self.samp_rate)
self._c2mag2[i] = blocks.complex_to_mag_squared(1)
self.blocks_max = blocks.max_ff(1)
self.blocks_peak_detector = blocks.peak_detector_fb(self.factor, self.factor, 0, self.alpha)
self.blocks_argmax = blocks.argmax_fs(1)
self.blocks_null_sink = blocks.null_sink(gr.sizeof_short*1)
self.digital_chunks_to_symbols = digital.chunks_to_symbols_sf((freqs), 1)
self.blocks_sample_and_hold = blocks.sample_and_hold_ff()
##################################################
# Connections
##################################################
for i in range(self.n):
self.connect((self, 0), (self._filter[i], 0))
self.connect((self._filter[i], 0), (self._c2mag2[i], 0))
self.connect((self._c2mag2[i], 0), (self.blocks_max, i))
self.connect((self._c2mag2[i], 0), (self.blocks_argmax, i))
self.connect((self.blocks_max, 0), (self.blocks_peak_detector, 0))
self.connect((self.blocks_peak_detector, 0), (self, 0))
self.connect((self.blocks_argmax, 0), (self.blocks_null_sink, 0))
self.connect((self.blocks_argmax, 1), (self.digital_chunks_to_symbols, 0))
self.connect((self.digital_chunks_to_symbols, 0), (self.blocks_sample_and_hold, 0))
self.connect((self.blocks_peak_detector, 0), (self.blocks_sample_and_hold, 1))
self.connect((self.blocks_sample_and_hold, 0), (self, 1))
self.connect((self.blocks_max, 0), (self, 2))
def run_test(fo, fi, interleaver, Kb, bitspersymbol, K, dimensionality,
constellation, N0, seed):
tb = gr.top_block()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short, Kb / 16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(
bitspersymbol, gr.GR_MSB_FIRST
) # unpack shorts to symbols compatible with the outer FSM input cardinality
enc_out = trellis.encoder_ss(fo, 0) # initial state = 0
inter = trellis.permutation(interleaver.K(), interleaver.INTER(), 1,
gr.sizeof_short)
enc_in = trellis.encoder_ss(fi, 0) # initial state = 0
mod = digital.chunks_to_symbols_sf(constellation, dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN, math.sqrt(N0 / 2), seed)
# RX
metrics_in = trellis.metrics_f(
fi.O(), dimensionality, constellation, digital.TRELLIS_EUCLIDEAN
) # data preprocessing to generate metrics for innner Viterbi
gnd = blocks.vector_source_f([0], True)
siso_in = trellis.siso_f(
fi, K, 0, -1, True, False, trellis.TRELLIS_MIN_SUM
) # Put -1 if the Initial/Final states are not set.
deinter = trellis.permutation(interleaver.K(), interleaver.DEINTER(),
fi.I(), gr.sizeof_float)
va_out = trellis.viterbi_s(
fo, K, 0, -1) # Put -1 if the Initial/Final states are not set.
fsmi2s = blocks.unpacked_to_packed_ss(
bitspersymbol, gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s()
tb.connect(src, src_head, s2fsmi, enc_out, inter, enc_in, mod)
tb.connect(mod, (add, 0))
tb.connect(noise, (add, 1))
tb.connect(add, metrics_in)
tb.connect(gnd, (siso_in, 0))
tb.connect(metrics_in, (siso_in, 1))
tb.connect(siso_in, deinter, va_out, fsmi2s, dst)
tb.run()
ntotal = dst.ntotal()
nright = dst.nright()
runlength = dst.runlength()
return (ntotal, ntotal - nright)
def run_test(f, Kb, bitspersymbol, K, dimensionality, constellation, N0, seed):
tb = gr.top_block()
# TX
numpy.random.seed(-seed)
packet = numpy.random.randint(0, 2, Kb) # create Kb random bits
packet[Kb - 10:Kb] = 0
packet[0:Kb] = 0
src = blocks.vector_source_s(packet.tolist(), False)
b2s = blocks.unpacked_to_packed_ss(1,
gr.GR_MSB_FIRST) # pack bits in shorts
s2fsmi = blocks.packed_to_unpacked_ss(
bitspersymbol, gr.GR_MSB_FIRST
) # unpack shorts to symbols compatible with the FSM input cardinality
enc = trellis.encoder_ss(f, 0) # initial state = 0
mod = digital.chunks_to_symbols_sf(constellation, dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN, math.sqrt(N0 / 2),
long(seed))
# RX
va = trellis.viterbi_combined_fs(
f, K, 0, 0, dimensionality, constellation, digital.TRELLIS_EUCLIDEAN
) # Put -1 if the Initial/Final states are not set.
fsmi2s = blocks.unpacked_to_packed_ss(
bitspersymbol, gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
s2b = blocks.packed_to_unpacked_ss(
1, gr.GR_MSB_FIRST) # unpack shorts to bits
dst = blocks.vector_sink_s()
tb.connect(src, b2s, s2fsmi, enc, mod)
tb.connect(mod, (add, 0))
tb.connect(noise, (add, 1))
tb.connect(add, va, fsmi2s, s2b, dst)
tb.run()
# A bit of cheating: run the program once and print the
# final encoder state..
# Then put it as the last argument in the viterbi block
#print "final state = " , enc.ST()
if len(dst.data()) != len(packet):
print "Error: not enough data:", len(dst.data()), len(packet)
ntotal = len(packet)
nwrong = sum(abs(packet - numpy.array(dst.data())))
return (ntotal, nwrong, abs(packet - numpy.array(dst.data())))
def test_sf_callback(self):
constA = [-3, -1, 1, 3]
constB = [12, -12, 6, -6]
src_data = (0, 1, 2, 3, 3, 2, 1, 0)
expected_result=(12, -12, 6, -6, -6, 6, -12, 12)
src = blocks.vector_source_s(src_data, False, 1, "")
op = digital.chunks_to_symbols_sf(constA)
op.set_symbol_table(constB)
dst = blocks.vector_sink_f()
self.tb.connect(src, op)
self.tb.connect(op, dst)
self.tb.run()
actual_result = dst.data()
self.assertEqual(expected_result, actual_result)
def test_sf_006(self):
const = [-3, -1, 1, 3]
src_data = (0, 1, 2, 3, 3, 2, 1, 0)
expected_result = [-3, -1, 1, 3, 3, 1, -1, -3]
src = blocks.vector_source_s(src_data)
op = digital.chunks_to_symbols_sf(const)
dst = blocks.vector_sink_f()
self.tb.connect(src, op)
self.tb.connect(op, dst)
self.tb.run()
actual_result = dst.data()
self.assertEqual(expected_result, actual_result)
def test_sf_callback(self):
constA = [-3, -1, 1, 3]
constB = [12, -12, 6, -6]
src_data = [0, 1, 2, 3, 3, 2, 1, 0]
expected_result = [12, -12, 6, -6, -6, 6, -12, 12]
src = blocks.vector_source_s(src_data, False, 1, [])
op = digital.chunks_to_symbols_sf(constA)
op.set_symbol_table(constB)
dst = blocks.vector_sink_f()
self.tb.connect(src, op)
self.tb.connect(op, dst)
self.tb.run()
actual_result = dst.data()
self.assertEqual(expected_result, actual_result)
def test_sf_006(self):
const = [-3, -1, 1, 3]
src_data = (0, 1, 2, 3, 3, 2, 1, 0)
expected_result = (-3, -1, 1, 3,
3, 1, -1, -3)
src = blocks.vector_source_s(src_data)
op = digital.chunks_to_symbols_sf(const)
dst = blocks.vector_sink_f()
self.tb.connect(src, op)
self.tb.connect(op, dst)
self.tb.run()
actual_result = dst.data()
self.assertEqual(expected_result, actual_result)
def run_test(fo, fi, interleaver, Kb, bitspersymbol, K, dimensionality,
constellation, Es, N0, IT, seed):
tb = gr.top_block()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short, Kb / 16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(
bitspersymbol, gr.GR_MSB_FIRST
) # unpack shorts to symbols compatible with the outer FSM input cardinality
#src = blocks.vector_source_s([0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1],False)
enc = trellis.pccc_encoder_ss(fo, 0, fi, 0, interleaver, K)
code = blocks.vector_sink_s()
mod = digital.chunks_to_symbols_sf(constellation, dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN, math.sqrt(N0 / 2), seed)
# RX
metrics_in = trellis.metrics_f(
fi.O() * fo.O(), dimensionality, constellation,
digital.TRELLIS_EUCLIDEAN
) # data preprocessing to generate metrics for innner SISO
scale = blocks.multiply_const_ff(1.0 / N0)
dec = trellis.pccc_decoder_s(fo, 0, -1, fi, 0, -1, interleaver, K, IT,
trellis.TRELLIS_MIN_SUM)
fsmi2s = blocks.unpacked_to_packed_ss(
bitspersymbol, gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s()
tb.connect(src, src_head, s2fsmi, enc, mod)
#tb.connect (src,enc,mod)
#tb.connect(enc,code)
tb.connect(mod, (add, 0))
tb.connect(noise, (add, 1))
tb.connect(add, metrics_in, scale, dec, fsmi2s, dst)
tb.run()
#print code.data()
ntotal = dst.ntotal()
nright = dst.nright()
runlength = dst.runlength()
return (ntotal, ntotal - nright)
def run_test (f,Kb,bitspersymbol,K,dimensionality,constellation,N0,seed,P):
tb = gr.top_block ()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short,Kb/16*P) # packet size in shorts
s2fsmi=blocks.packed_to_unpacked_ss(bitspersymbol,gr.GR_MSB_FIRST) # unpack shorts to symbols compatible with the FSM input cardinality
s2p = blocks.stream_to_streams(gr.sizeof_short,P) # serial to parallel
enc = trellis.encoder_ss(f,0) # initiali state = 0
mod = digital.chunks_to_symbols_sf(constellation,dimensionality)
# CHANNEL
add=[]
noise=[]
for i in range(P):
add.append(blocks.add_ff())
noise.append(analog.noise_source_f(analog.GR_GAUSSIAN,math.sqrt(N0/2),seed))
# RX
metrics = trellis.metrics_f(f.O(),dimensionality,constellation,digital.TRELLIS_EUCLIDEAN) # data preprocessing to generate metrics for Viterbi
va = trellis.viterbi_s(f,K,0,-1) # Put -1 if the Initial/Final states are not set.
p2s = blocks.streams_to_stream(gr.sizeof_short,P) # parallel to serial
fsmi2s=blocks.unpacked_to_packed_ss(bitspersymbol,gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s()
tb.connect (src,src_head,s2fsmi,s2p)
for i in range(P):
tb.connect ((s2p,i),(enc,i),(mod,i))
tb.connect ((mod,i),(add[i],0))
tb.connect (noise[i],(add[i],1))
tb.connect (add[i],(metrics,i))
tb.connect ((metrics,i),(va,i),(p2s,i))
tb.connect (p2s,fsmi2s,dst)
tb.run()
# A bit of cheating: run the program once and print the
# final encoder state.
# Then put it as the last argument in the viterbi block
#print "final state = " , enc.ST()
ntotal = dst.ntotal ()
nright = dst.nright ()
runlength = dst.runlength ()
return (ntotal,ntotal-nright)
def run_test(fo, fi, interleaver, Kb, bitspersymbol, K, channel, modulation,
dimensionality, tot_constellation, Es, N0, IT, seed):
tb = gr.top_block()
L = len(channel)
# TX
# this for loop is TOO slow in python!!!
packet = [0] * (K)
random.seed(seed)
for i in range(len(packet)):
packet[i] = random.randint(0, 2**bitspersymbol - 1) # random symbols
src = blocks.vector_source_s(packet, False)
enc_out = trellis.encoder_ss(fo, 0) # initial state = 0
inter = trellis.permutation(interleaver.K(), interleaver.INTER(), 1,
gr.sizeof_short)
mod = digital.chunks_to_symbols_sf(modulation[1], modulation[0])
# CHANNEL
isi = filter.fir_filter_fff(1, channel)
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN, math.sqrt(N0 / 2), seed)
# RX
(head, tail) = make_rx(tb, fo, fi, dimensionality, tot_constellation, K,
interleaver, IT, Es, N0, trellis.TRELLIS_MIN_SUM)
dst = blocks.vector_sink_s()
tb.connect(src, enc_out, inter, mod)
tb.connect(mod, isi, (add, 0))
tb.connect(noise, (add, 1))
tb.connect(add, head)
tb.connect(tail, dst)
tb.run()
data = dst.data()
ntotal = len(data)
nright = 0
for i in range(ntotal):
if packet[i] == data[i]:
nright = nright + 1
#else:
#print "Error in ", i
return (ntotal, ntotal - nright)
def run_test (f,Kb,bitspersymbol,K,dimensionality,constellation,N0,seed):
tb = gr.top_block ()
# TX
numpy.random.seed(-seed)
packet = numpy.random.randint(0,2,Kb) # create Kb random bits
packet[Kb-10:Kb]=0
packet[0:Kb]=0
src = blocks.vector_source_s(packet.tolist(),False)
b2s = blocks.unpacked_to_packed_ss(1,gr.GR_MSB_FIRST) # pack bits in shorts
s2fsmi = blocks.packed_to_unpacked_ss(bitspersymbol,gr.GR_MSB_FIRST) # unpack shorts to symbols compatible with the FSM input cardinality
enc = trellis.encoder_ss(f,0) # initial state = 0
mod = digital.chunks_to_symbols_sf(constellation,dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN,math.sqrt(N0 / 2),int(seed))
# RX
va = trellis.viterbi_combined_fs(f,K,0,0,dimensionality,constellation,digital.TRELLIS_EUCLIDEAN) # Put -1 if the Initial/Final states are not set.
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol,gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
s2b = blocks.packed_to_unpacked_ss(1,gr.GR_MSB_FIRST) # unpack shorts to bits
dst = blocks.vector_sink_s();
tb.connect (src,b2s,s2fsmi,enc,mod)
tb.connect (mod,(add,0))
tb.connect (noise,(add,1))
tb.connect (add,va,fsmi2s,s2b,dst)
tb.run()
# A bit of cheating: run the program once and print the
# final encoder state..
# Then put it as the last argument in the viterbi block
#print "final state = " , enc.ST()
if len(dst.data()) != len(packet):
print("Error: not enough data:", len(dst.data()), len(packet))
ntotal=len(packet)
nwrong = sum(abs(packet-numpy.array(dst.data())));
return (ntotal,nwrong,abs(packet-numpy.array(dst.data())))
def run_test(f, Kb, bitspersymbol, K, dimensionality, tot_constellation, N0,
seed):
tb = gr.top_block()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short, Kb / 16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(
bitspersymbol, gr.GR_MSB_FIRST
) # unpack shorts to symbols compatible with the FSM input cardinality
enc = trellis.encoder_ss(f, 0) # initial state = 0
# essentially here we implement the combination of modulation and channel as a memoryless modulation (the memory induced by the channel is hidden in the FSM)
mod = digital.chunks_to_symbols_sf(tot_constellation, dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN, math.sqrt(N0 / 2), seed)
# RX
metrics = trellis.metrics_f(
f.O(), dimensionality, tot_constellation, digital.TRELLIS_EUCLIDEAN
) # data preprocessing to generate metrics for Viterbi
va = trellis.viterbi_s(
f, K, 0, -1) # Put -1 if the Initial/Final states are not set.
fsmi2s = blocks.unpacked_to_packed_ss(
bitspersymbol, gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s()
tb.connect(src, src_head, s2fsmi, enc, mod)
tb.connect(mod, (add, 0))
tb.connect(noise, (add, 1))
tb.connect(add, metrics)
tb.connect(metrics, va, fsmi2s, dst)
tb.run()
ntotal = dst.ntotal()
nright = dst.nright()
runlength = dst.runlength()
#print ntotal,nright,runlength
return (ntotal, ntotal - nright)
def run_test(fo, fi, interleaver, Kb, bitspersymbol, K, dimensionality,
constellation, Es, N0, IT, seed):
tb = gr.top_block()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short, Kb / 16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(
bitspersymbol, gr.GR_MSB_FIRST
) # unpack shorts to symbols compatible with the outer FSM input cardinality
enc_out = trellis.encoder_ss(fo, 0) # initial state = 0
inter = trellis.permutation(interleaver.K(), interleaver.INTER(), 1,
gr.sizeof_short)
enc_in = trellis.encoder_ss(fi, 0) # initial state = 0
mod = digital.chunks_to_symbols_sf(constellation, dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN, math.sqrt(N0 / 2), seed)
# RX
(head, tail) = make_rx(tb, fo, fi, dimensionality, constellation, K,
interleaver, IT, Es, N0, trellis.TRELLIS_MIN_SUM)
#(head,tail) = make_rx(tb,fo,fi,dimensionality,constellation,K,interleaver,IT,Es,N0,trellis.TRELLIS_SUM_PRODUCT)
fsmi2s = blocks.unpacked_to_packed_ss(
bitspersymbol, gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s()
tb.connect(src, src_head, s2fsmi, enc_out, inter, enc_in, mod)
tb.connect(mod, (add, 0))
tb.connect(noise, (add, 1))
tb.connect(add, head)
tb.connect(tail, fsmi2s, dst)
tb.run()
#print enc_out.ST(), enc_in.ST()
ntotal = dst.ntotal()
nright = dst.nright()
runlength = dst.runlength()
return (ntotal, ntotal - nright)
def run_test(fo, fi, interleaver, Kb, bitspersymbol, K, dimensionality,
constellation, Es, N0, IT, seed):
tb = gr.top_block()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short, Kb / 16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(
bitspersymbol, gr.GR_MSB_FIRST
) # unpack shorts to symbols compatible with the outer FSM input cardinality
enc = trellis.sccc_encoder_ss(fo, 0, fi, 0, interleaver, K)
mod = digital.chunks_to_symbols_sf(constellation, dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN, math.sqrt(N0 / 2), seed)
# RX
dec = trellis.sccc_decoder_combined_fs(fo, 0, -1, fi, 0, -1, interleaver,
K, IT, trellis.TRELLIS_MIN_SUM,
dimensionality, constellation,
digital.TRELLIS_EUCLIDEAN, 1.0)
fsmi2s = blocks.unpacked_to_packed_ss(
bitspersymbol, gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s()
#tb.connect (src,src_head,s2fsmi,enc_out,inter,enc_in,mod)
tb.connect(src, src_head, s2fsmi, enc, mod)
tb.connect(mod, (add, 0))
tb.connect(noise, (add, 1))
#tb.connect (add,head)
#tb.connect (tail,fsmi2s,dst)
tb.connect(add, dec, fsmi2s, dst)
tb.run()
#print enc_out.ST(), enc_in.ST()
ntotal = dst.ntotal()
nright = dst.nright()
runlength = dst.runlength()
return (ntotal, ntotal - nright)
def run_test (fo,fi,interleaver,Kb,bitspersymbol,K,channel,modulation,dimensionality,tot_constellation,Es,N0,IT,seed):
tb = gr.top_block ()
L = len(channel)
# TX
# this for loop is TOO slow in python!!!
packet = [0]*(K)
random.seed(seed)
for i in range(len(packet)):
packet[i] = random.randint(0, 2**bitspersymbol - 1) # random symbols
src = blocks.vector_source_s(packet,False)
enc_out = trellis.encoder_ss(fo,0) # initial state = 0
inter = trellis.permutation(interleaver.K(),interleaver.INTER(),1,gr.sizeof_short)
mod = digital.chunks_to_symbols_sf(modulation[1],modulation[0])
# CHANNEL
isi = filter.fir_filter_fff(1,channel)
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN,math.sqrt(N0/2),seed)
# RX
(head,tail) = make_rx(tb,fo,fi,dimensionality,tot_constellation,K,interleaver,IT,Es,N0,trellis.TRELLIS_MIN_SUM)
dst = blocks.vector_sink_s();
tb.connect (src,enc_out,inter,mod)
tb.connect (mod,isi,(add,0))
tb.connect (noise,(add,1))
tb.connect (add,head)
tb.connect (tail,dst)
tb.run()
data = dst.data()
ntotal = len(data)
nright=0
for i in range(ntotal):
if packet[i]==data[i]:
nright=nright+1
#else:
#print "Error in ", i
return (ntotal,ntotal-nright)
def run_test (f,Kb,bitspersymbol,K,dimensionality,constellation,N0,seed):
tb = gr.top_block ()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short,Kb/16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(bitspersymbol,gr.GR_MSB_FIRST) # unpack shorts to symbols compatible with the FSM input cardinality
enc = trellis.encoder_ss(f,0) # initial state = 0
mod = digital.chunks_to_symbols_sf(constellation,dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN,math.sqrt(N0/2),seed)
# RX
va = trellis.viterbi_combined_fs(f,K,0,-1,dimensionality,constellation,digital.TRELLIS_EUCLIDEAN) # Put -1 if the Initial/Final states are not set.
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol,gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s();
tb.connect (src,src_head,s2fsmi,enc,mod)
tb.connect (mod,(add,0))
tb.connect (noise,(add,1))
tb.connect (add,va,fsmi2s,dst)
tb.run()
# A bit of cheating: run the program once and print the
# final encoder state..
# Then put it as the last argument in the viterbi block
#print "final state = " , enc.ST()
ntotal = dst.ntotal ()
nright = dst.nright ()
runlength = dst.runlength ()
return (ntotal,ntotal-nright)
def run_test (fo,fi,interleaver,Kb,bitspersymbol,K,dimensionality,constellation,Es,N0,IT,seed):
tb = gr.top_block ()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head (gr.sizeof_short,Kb/16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(bitspersymbol,gr.GR_MSB_FIRST) # unpack shorts to symbols compatible with the outer FSM input cardinality
#src = blocks.vector_source_s([0,0,0,0,0,0,0,0,0,0,0,0,1,1,1,1],False)
enc = trellis.pccc_encoder_ss(fo,0,fi,0,interleaver,K)
code = blocks.vector_sink_s()
mod = digital.chunks_to_symbols_sf(constellation,dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN,math.sqrt(N0/2),seed)
# RX
metrics_in = trellis.metrics_f(fi.O()*fo.O(),dimensionality,constellation,digital.TRELLIS_EUCLIDEAN) # data preprocessing to generate metrics for innner SISO
scale = blocks.multiply_const_ff(1.0/N0)
dec = trellis.pccc_decoder_s(fo,0,-1,fi,0,-1,interleaver,K,IT,trellis.TRELLIS_MIN_SUM)
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol,gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s()
tb.connect (src,src_head,s2fsmi,enc,mod)
#tb.connect (src,enc,mod)
#tb.connect(enc,code)
tb.connect (mod,(add,0))
tb.connect (noise,(add,1))
tb.connect (add,metrics_in,scale,dec,fsmi2s,dst)
tb.run()
#print code.data()
ntotal = dst.ntotal ()
nright = dst.nright ()
runlength = dst.runlength ()
return (ntotal,ntotal-nright)
def run_test(fo, fi, interleaver, Kb, bitspersymbol, K, dimensionality, tot_constellation, Es, N0, IT, seed):
tb = gr.top_block()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short, Kb / 16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(
bitspersymbol, gr.GR_MSB_FIRST
) # unpack shorts to symbols compatible with the iouter FSM input cardinality
enc_out = trellis.encoder_ss(fo, 0) # initial state = 0
inter = trellis.permutation(interleaver.K(), interleaver.INTER(), 1, gr.sizeof_short)
enc_in = trellis.encoder_ss(fi, 0) # initial state = 0
# essentially here we implement the combination of modulation and channel as a memoryless modulation (the memory induced by the channel is hidden in the innner FSM)
mod = digital.chunks_to_symbols_sf(tot_constellation, dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN, math.sqrt(N0 / 2), seed)
# RX
(head, tail) = make_rx(
tb, fo, fi, dimensionality, tot_constellation, K, interleaver, IT, Es, N0, trellis.TRELLIS_MIN_SUM
)
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol, gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s()
tb.connect(src, src_head, s2fsmi, enc_out, inter, enc_in, mod)
tb.connect(mod, (add, 0))
tb.connect(noise, (add, 1))
tb.connect(add, head)
tb.connect(tail, fsmi2s, dst)
tb.run()
ntotal = dst.ntotal()
nright = dst.nright()
runlength = dst.runlength()
# print ntotal,nright,runlength
return (ntotal, ntotal - nright)
def run_test (f,Kb,bitspersymbol,K,dimensionality,constellation,N0,seed):
tb = gr.top_block ()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head(gr.sizeof_short,Kb/16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(bitspersymbol,gr.GR_MSB_FIRST) # unpack shorts to symbols compatible with the FSM input cardinality
enc = trellis.encoder_ss(f,0) # initial state = 0
mod = digital.chunks_to_symbols_sf(constellation,dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN,math.sqrt(N0/2),seed)
# RX
va = trellis.viterbi_combined_fs(f,K,0,-1,dimensionality,constellation,digital.TRELLIS_EUCLIDEAN) # Put -1 if the Initial/Final states are not set.
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol,gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s();
tb.connect (src,src_head,s2fsmi,enc,mod)
tb.connect (mod,(add,0))
tb.connect (noise,(add,1))
tb.connect (add,va,fsmi2s,dst)
tb.run()
# A bit of cheating: run the program once and print the
# final encoder state..
# Then put it as the last argument in the viterbi block
#print "final state = " , enc.ST()
ntotal = dst.ntotal ()
nright = dst.nright ()
runlength = dst.runlength ()
return (ntotal,ntotal-nright)
def run_test (fo,fi,interleaver,Kb,bitspersymbol,K,dimensionality,constellation,N0,seed):
tb = gr.top_block ()
# TX
src = blocks.lfsr_32k_source_s()
src_head = blocks.head (gr.sizeof_short,Kb/16) # packet size in shorts
s2fsmi = blocks.packed_to_unpacked_ss(bitspersymbol,gr.GR_MSB_FIRST) # unpack shorts to symbols compatible with the outer FSM input cardinality
enc_out = trellis.encoder_ss(fo,0) # initial state = 0
inter = trellis.permutation(interleaver.K(),interleaver.INTER(),1,gr.sizeof_short)
enc_in = trellis.encoder_ss(fi,0) # initial state = 0
mod = digital.chunks_to_symbols_sf(constellation,dimensionality)
# CHANNEL
add = blocks.add_ff()
noise = analog.noise_source_f(analog.GR_GAUSSIAN,math.sqrt(N0/2),seed)
# RX
metrics_in = trellis.metrics_f(fi.O(),dimensionality,constellation,digital.TRELLIS_EUCLIDEAN) # data preprocessing to generate metrics for innner Viterbi
gnd = blocks.vector_source_f([0],True);
siso_in = trellis.siso_f(fi,K,0,-1,True,False,trellis.TRELLIS_MIN_SUM) # Put -1 if the Initial/Final states are not set.
deinter = trellis.permutation(interleaver.K(),interleaver.DEINTER(),fi.I(),gr.sizeof_float)
va_out = trellis.viterbi_s(fo,K,0,-1) # Put -1 if the Initial/Final states are not set.
fsmi2s = blocks.unpacked_to_packed_ss(bitspersymbol,gr.GR_MSB_FIRST) # pack FSM input symbols to shorts
dst = blocks.check_lfsr_32k_s()
tb.connect (src,src_head,s2fsmi,enc_out,inter,enc_in,mod)
tb.connect (mod,(add,0))
tb.connect (noise,(add,1))
tb.connect (add,metrics_in)
tb.connect (gnd,(siso_in,0))
tb.connect (metrics_in,(siso_in,1))
tb.connect (siso_in,deinter,va_out,fsmi2s,dst)
tb.run()
ntotal = dst.ntotal ()
nright = dst.nright ()
runlength = dst.runlength ()
return (ntotal,ntotal-nright)
def __init__(self, seq1, seq2, factor, alpha, samp_rate, freqs):
"""
Description:
This block is functionally equivalent to the frequency_timing_estimator block, except from the fact that each filter is matched to a sequence that can be written as the kronecker product of seq1 and seq2.
Args:
seq1: sequence1 of kronecker filter, which is the given training sequence.
seq2: sequence2 of kronecker filter, which is the pulse for each training symbol.
factor: the rise and fall factors in peak detector, which is the factor determining when a peak has started and ended. In the peak detector, an average of the signal is calculated. When the value of the signal goes over factor*average, we start looking for a peak. When the value of the signal goes below factor*average, we stop looking for a peak. factor takes values in (0,1).
alpha: the smoothing factor of a moving average filter used in the peak detector takeng values in (0,1).
samp_rate: the sample rate of the system, which is used in the kronecker_filter.
freqs: the vector of center frequencies for each matched filter. Note that for a training sequence of length Nt, each matched filter can recover a sequence with normalized frequency offset ~ 1/(2Nt).
"""
gr.hier_block2.__init__(self,
"freq_timing_estimator",
gr.io_signature(1, 1, gr.sizeof_gr_complex*1),
gr.io_signaturev(3, 3, [gr.sizeof_char*1, gr.sizeof_float*1, gr.sizeof_float*1]),
)
##################################################
# Parameters
##################################################
self.seq1 = seq1
self.seq2 = seq2
self.factor = factor
self.alpha = alpha
self.samp_rate = samp_rate
self.freqs = freqs
self.n = n = len(freqs)
##################################################
# Blocks
##################################################
self._filter=[0]*self.n
self._c2mag2=[0]*self.n
for i in range(self.n):
self._filter[i]= cdma.kronecker_filter(seq1,seq2,samp_rate,self.freqs[i])
#self._filter[i]= filter.freq_xlating_fir_filter_ccc(1, (numpy.conjugate(self.ts[::-1])), self.freqs[i], self.samp_rate)
self._c2mag2[i] = blocks.complex_to_mag_squared(1)
self.blocks_max = blocks.max_ff(1)
self.blocks_peak_detector = blocks.peak_detector_fb(self.factor, self.factor, 0, self.alpha)
self.blocks_argmax = blocks.argmax_fs(1)
self.blocks_null_sink = blocks.null_sink(gr.sizeof_short*1)
self.digital_chunks_to_symbols = digital.chunks_to_symbols_sf((freqs), 1)
self.blocks_sample_and_hold = blocks.sample_and_hold_ff()
##################################################
# Connections
##################################################
for i in range(self.n):
self.connect((self, 0), (self._filter[i], 0))
self.connect((self._filter[i], 0), (self._c2mag2[i], 0))
self.connect((self._c2mag2[i], 0), (self.blocks_max, i))
self.connect((self._c2mag2[i], 0), (self.blocks_argmax, i))
self.connect((self.blocks_max, 0), (self.blocks_peak_detector, 0))
self.connect((self.blocks_peak_detector, 0), (self, 0))
self.connect((self.blocks_argmax, 0), (self.blocks_null_sink, 0))
self.connect((self.blocks_argmax, 1), (self.digital_chunks_to_symbols, 0))
self.connect((self.digital_chunks_to_symbols, 0), (self.blocks_sample_and_hold, 0))
self.connect((self.blocks_peak_detector, 0), (self.blocks_sample_and_hold, 1))
self.connect((self.blocks_sample_and_hold, 0), (self, 1))
self.connect((self.blocks_max, 0), (self, 2))
def __init__(self, seq1, seq2, factor, lookahead, alpha, freqs):
"""
Description:
frequency timing estimator class does frequency/timing acquisition from scratch.It uses a bank of parallel correlators at each specified frequency. It then takes the max abs value of all these and passes it through a peak detector to find timing.
Args:
seq1: sequence1 of kronecker filter, which is the given training sequence.
seq2: sequence2 of kronecker filter, which is the pulse for each training symbol.
factor: the rise and fall factors in peak detector, which is the factor determining when a peak has started and ended. In the peak detector, an average of the signal is calculated. When the value of the signal goes over factor*average, we start looking for a peak. When the value of the signal goes below factor*average, we stop looking for a peak.
alpha: the smoothing factor of a moving average filter used in the peak detector taking values in (0,1).
freqs: the vector of normalized center frequencies for each matched filter. Note that for a training sequence of length Nt, each matched filter can recover a sequence with normalized frequency offset ~ 1/(2Nt).
"""
gr.hier_block2.__init__(self,
"freq_timing_estimator",
gr.io_signature(1, 1, gr.sizeof_gr_complex*1),
gr.io_signaturev(3, 3, [gr.sizeof_char*1, gr.sizeof_float*1, gr.sizeof_float*1]),
)
##################################################
# Parameters
##################################################
self.seq1 = seq1
self.seq2 = seq2
self.factor = factor
self.lookahead = lookahead
self.alpha = alpha
self.freqs = freqs
self.n = n = len(freqs)
self.on = 1
##################################################
# Blocks
##################################################
self._filter=[0]*self.n
self._c2mag2=[0]*self.n
for i in range(self.n):
#self._filter[i]= cdma.kronecker_filter(seq1,seq2,1,self.freqs[i])
#self._filter[i]= filter.freq_xlating_fir_filter_ccc(1, numpy.kron(seq1,seq2), self.freqs[i], 1)
self._filter[i]= filter.freq_xlating_fft_filter_ccc(1, numpy.kron(seq1,seq2), self.freqs[i], 1)
self._c2mag2[i] = blocks.complex_to_mag_squared(1)
self.blocks_max = blocks.max_ff(1)
self.blocks_peak_detector = cdma.switched_peak_detector_fb(self.factor, self.factor, self.lookahead, self.alpha, self.on)
self.blocks_argmax = blocks.argmax_fs(1)
self.blocks_null_sink = blocks.null_sink(gr.sizeof_short*1)
self.digital_chunks_to_symbols = digital.chunks_to_symbols_sf((freqs), 1)
self.blocks_sample_and_hold = blocks.sample_and_hold_ff()
##################################################
# Connections
##################################################
for i in range(self.n):
self.connect((self, 0), (self._filter[i], 0))
self.connect((self._filter[i], 0), (self._c2mag2[i], 0))
self.connect((self._c2mag2[i], 0), (self.blocks_max, i))
self.connect((self._c2mag2[i], 0), (self.blocks_argmax, i))
self.connect((self.blocks_max, 0), (self.blocks_peak_detector, 0))
self.connect((self.blocks_peak_detector, 0), (self, 0))
self.connect((self.blocks_argmax, 0), (self.blocks_null_sink, 0))
self.connect((self.blocks_argmax, 1), (self.digital_chunks_to_symbols, 0))
self.connect((self.digital_chunks_to_symbols, 0), (self.blocks_sample_and_hold, 0))
self.connect((self.blocks_peak_detector, 0), (self.blocks_sample_and_hold, 1))
self.connect((self.blocks_sample_and_hold, 0), (self, 1))
self.connect((self.blocks_max, 0), (self, 2))
def __init__(self, ts, factor, alpha, samp_rate, freqs):
"""
Description:
This block is designed to perform frequency and timing acquisition for a known training sequence in the presense of frequency and timing offset and noise. Its input is a complex stream. It has three outputs:
1) a stream of flags (bytes) indicating the begining of the training sequence (to be used from subsequent blocks to "chop" the incoming stream,
2) a stream with the current acquired frequency offset, and
3) a stream with the current acquired peak of the matched filter
Internally, it consists of a user defined number of parallel matched filters (as many as the size of the freqs vector), each consistng of a frequency Xlating FIR filter with sample rate samp_rate, filter taps matched to the training sequence ts, and center frequency freqs[i]. The filter outputs are magnitude squared and passed through a max block and then through a peak detector.
Args:
ts: the training sequence. For example, in DSSS system, it's the chip-based spread training sequence.
factor: the rise and fall factors in peak detector, which is the factor determining when a peak has started and ended. In the peak detector, an average of the signal is calculated. When the value of the signal goes over factor*average, we start looking for a peak. When the value of the signal goes below factor*average, we stop looking for a peak. factor takes values in (0,1).
alpha: the smoothing factor of a moving average filter used in the peak detector takeng values in (0,1).
samp_rate: the sample rate of the system, which is used in the freq_xlating_fir_filter.
freqs: the vector of center frequencies for each matched filter. Note that for a training sequence of length Nt, each matched filter can recover a sequence with normalized frequency offset ~ 1/(2Nt).
"""
gr.hier_block2.__init__(
self, "freq_timing_estimator",
gr.io_signature(1, 1, gr.sizeof_gr_complex*1),
gr.io_signaturev(3, 3, [gr.sizeof_char*1, gr.sizeof_float*1, gr.sizeof_float*1]),
)
##################################################
# Parameters
##################################################
self.ts = ts
self.factor = factor
self.alpha = alpha
self.samp_rate = samp_rate
self.freqs = freqs
self.n = n = len(freqs)
##################################################
# Blocks
##################################################
self._filter=[0]*self.n
self._c2mag2=[0]*self.n
for i in range(self.n):
self._filter[i]= filter.freq_xlating_fir_filter_ccc(1, (numpy.conjugate(self.ts[::-1])), self.freqs[i], self.samp_rate)
self._c2mag2[i] = blocks.complex_to_mag_squared(1)
self.blocks_max = blocks.max_ff(1)
self.blocks_peak_detector = blocks.peak_detector_fb(self.factor, self.factor, 0, self.alpha)
self.blocks_argmax = blocks.argmax_fs(1)
self.blocks_null_sink = blocks.null_sink(gr.sizeof_short*1)
self.digital_chunks_to_symbols = digital.chunks_to_symbols_sf((freqs), 1)
self.blocks_sample_and_hold = blocks.sample_and_hold_ff()
##################################################
# Connections
##################################################
for i in range(self.n):
self.connect((self, 0), (self._filter[i], 0))
self.connect((self._filter[i], 0), (self._c2mag2[i], 0))
self.connect((self._c2mag2[i], 0), (self.blocks_max, i))
self.connect((self._c2mag2[i], 0), (self.blocks_argmax, i))
self.connect((self.blocks_max, 0), (self.blocks_peak_detector, 0))
self.connect((self.blocks_peak_detector, 0), (self, 0))
self.connect((self.blocks_argmax, 0), (self.blocks_null_sink, 0))
self.connect((self.blocks_argmax, 1), (self.digital_chunks_to_symbols, 0))
self.connect((self.digital_chunks_to_symbols, 0), (self.blocks_sample_and_hold, 0))
self.connect((self.blocks_peak_detector, 0), (self.blocks_sample_and_hold, 1))
self.connect((self.blocks_sample_and_hold, 0), (self, 1))
self.connect((self.blocks_max, 0), (self, 2))
def __init__(self, seq1, seq2, factor, alpha, freqs, debug_onoff, debug_port, prefix):
"""
Description:
frequency timing estimator class does frequency/timing acquisition from scratch.It uses a bank of parallel correlators at each specified frequency. It then takes the max abs value of all these and passes it through a peak detector to find timing.
Args:
seq1: sequence1 of kronecker filter, which is the given training sequence.
seq2: sequence2 of kronecker filter, which is the pulse for each training symbol.
factor: the rise and fall factors in peak detector, which is the factor determining when a peak has started and ended. In the peak detector, an average of the signal is calculated. When the value of the signal goes over factor*average, we start looking for a peak. When the value of the signal goes below factor*average, we stop looking for a peak.
alpha: the smoothing factor of a moving average filter used in the peak detector taking values in (0,1).
freqs: the vector of normalized center frequencies for each matched filter. Note that for a training sequence of length Nt, each matched filter can recover a sequence with normalized frequency offset ~ 1/(2Nt).
"""
gr.hier_block2.__init__(self,
"freq_timing_estimator",
gr.io_signature(1, 1, gr.sizeof_gr_complex*1),
gr.io_signaturev(3, 3, [gr.sizeof_char*1, gr.sizeof_float*1, gr.sizeof_float*1]),
)
##################################################
# Parameters
##################################################
self.seq1 = seq1
self.seq2 = seq2
self.factor = factor
self.alpha = alpha
self.freqs = freqs
self.n = n = len(freqs)
self.on = 1
self.debug_onoff = debug_onoff # 1: dump ports info to file 0: no debug output
self.debug_port = debug_port # 0-n_filt-1 is the output of each filter branck; n_filter is the output of maximum
self.prefix = prefix
##################################################
# Blocks
##################################################
self._filter=[0]*self.n
self._c2mag2=[0]*self.n
for i in range(self.n):
#self._filter[i]= cdma.kronecker_filter(seq1,seq2,1,self.freqs[i])
#self._filter[i]= filter.freq_xlating_fir_filter_ccc(1, numpy.kron(seq1,seq2), self.freqs[i], 1)
self._filter[i]= filter.freq_xlating_fft_filter_ccc(1, numpy.kron(seq1,seq2), self.freqs[i], 1)
self._c2mag2[i] = blocks.complex_to_mag_squared(1)
self.blocks_max = blocks.max_ff(1)
self.blocks_peak_detector = cdma.switched_peak_detector_fb(self.factor, self.factor, 0, self.alpha, self.on)
self.blocks_argmax = blocks.argmax_fs(1)
self.blocks_null_sink = blocks.null_sink(gr.sizeof_short*1)
self.digital_chunks_to_symbols = digital.chunks_to_symbols_sf((freqs), 1)
self.blocks_sample_and_hold = blocks.sample_and_hold_ff()
if self.debug_onoff == True:
num_of_file_sinks = len(self.debug_port)
self._filesink = [0]*num_of_file_sinks
for i in range(num_of_file_sinks):
if self.debug_port[i] == self.n:
filename = prefix+"max.dat"
else:
filename = prefix+"filter"+str(i)+".dat"
print filename
self._filesink[i] = blocks.file_sink(gr.sizeof_float*1, filename, False)
self._filesink[i].set_unbuffered(False)
# this is the block for bundling the outputs of each branch of filters and the input of peak detector
##################################################
# Connections
##################################################
for i in range(self.n):
self.connect((self, 0), (self._filter[i], 0))
self.connect((self._filter[i], 0), (self._c2mag2[i], 0))
self.connect((self._c2mag2[i], 0), (self.blocks_max, i))
self.connect((self._c2mag2[i], 0), (self.blocks_argmax, i))
self.connect((self.blocks_max, 0), (self.blocks_peak_detector, 0))
self.connect((self.blocks_peak_detector, 0), (self, 0))
self.connect((self.blocks_argmax, 0), (self.blocks_null_sink, 0))
self.connect((self.blocks_argmax, 1), (self.digital_chunks_to_symbols, 0))
self.connect((self.digital_chunks_to_symbols, 0), (self.blocks_sample_and_hold, 0))
self.connect((self.blocks_peak_detector, 0), (self.blocks_sample_and_hold, 1))
self.connect((self.blocks_sample_and_hold, 0), (self, 1))
self.connect((self.blocks_max, 0), (self, 2))
if self.debug_onoff == True:
for i in range(num_of_file_sinks):
port_index = self.debug_port[i]
if port_index == self.n:
self.connect((self.blocks_max, 0), (self._filesink[i], 0))
else:
self.connect((self._c2mag2[port_index], 0), (self._filesink[i], 0))
def __init__(self, seq1, seq2, factor, alpha, freqs):
"""
Description:
frequency timing estimator class does frequency/timing acquisition from scratch.It uses a bank of parallel correlators at each specified frequency. It then takes the max abs value of all these and passes it through a peak detector to find timing.
Args:
seq1: sequence1 of kronecker filter, which is the given training sequence.
seq2: sequence2 of kronecker filter, which is the pulse for each training symbol.
factor: the rise and fall factors in peak detector, which is the factor determining when a peak has started and ended. In the peak detector, an average of the signal is calculated. When the value of the signal goes over factor*average, we start looking for a peak. When the value of the signal goes below factor*average, we stop looking for a peak.
alpha: the smoothing factor of a moving average filter used in the peak detector taking values in (0,1).
freqs: the vector of normalized center frequencies for each matched filter. Note that for a training sequence of length Nt, each matched filter can recover a sequence with normalized frequency offset ~ 1/(2Nt).
"""
gr.hier_block2.__init__(
self,
"freq_timing_estimator",
gr.io_signature(1, 1, gr.sizeof_gr_complex * 1),
gr.io_signaturev(
3, 3,
[gr.sizeof_char * 1, gr.sizeof_float * 1, gr.sizeof_float * 1
]),
)
##################################################
# Parameters
##################################################
self.seq1 = seq1
self.seq2 = seq2
self.factor = factor
self.alpha = alpha
self.freqs = freqs
self.n = n = len(freqs)
self.on = 1
##################################################
# Blocks
##################################################
self._filter = [0] * self.n
self._c2mag2 = [0] * self.n
for i in range(self.n):
#self._filter[i]= cdma.kronecker_filter(seq1,seq2,1,self.freqs[i])
#self._filter[i]= filter.freq_xlating_fir_filter_ccc(1, numpy.kron(seq1,seq2), self.freqs[i], 1)
self._filter[i] = filter.freq_xlating_fft_filter_ccc(
1, numpy.kron(seq1, seq2), self.freqs[i], 1)
self._c2mag2[i] = blocks.complex_to_mag_squared(1)
self.blocks_max = blocks.max_ff(1)
self.blocks_peak_detector = cdma.switched_peak_detector_fb(
self.factor, self.factor, 0, self.alpha, self.on)
self.blocks_argmax = blocks.argmax_fs(1)
self.blocks_null_sink = blocks.null_sink(gr.sizeof_short * 1)
self.digital_chunks_to_symbols = digital.chunks_to_symbols_sf((freqs),
1)
self.blocks_sample_and_hold = blocks.sample_and_hold_ff()
##################################################
# Connections
##################################################
for i in range(self.n):
self.connect((self, 0), (self._filter[i], 0))
self.connect((self._filter[i], 0), (self._c2mag2[i], 0))
self.connect((self._c2mag2[i], 0), (self.blocks_max, i))
self.connect((self._c2mag2[i], 0), (self.blocks_argmax, i))
self.connect((self.blocks_max, 0), (self.blocks_peak_detector, 0))
self.connect((self.blocks_peak_detector, 0), (self, 0))
self.connect((self.blocks_argmax, 0), (self.blocks_null_sink, 0))
self.connect((self.blocks_argmax, 1),
(self.digital_chunks_to_symbols, 0))
self.connect((self.digital_chunks_to_symbols, 0),
(self.blocks_sample_and_hold, 0))
self.connect((self.blocks_peak_detector, 0),
(self.blocks_sample_and_hold, 1))
self.connect((self.blocks_sample_and_hold, 0), (self, 1))
self.connect((self.blocks_max, 0), (self, 2))
