generate_RML2016.04c.py

#!/usr/bin/env python
from transmitters import transmitters
from source_alphabet import source_alphabet
import timeseries_slicer
import analyze_stats
from gnuradio import channels, gr, blocks
import numpy as np
import numpy.fft, cPickle, gzip

'''
Generate dataset with dynamic channel model across range of SNRs
'''

apply_channel = True
output = {}
min_length = 9e9
snr_vals = range(-20,20,2)
for snr in snr_vals:
    for alphabet_type in transmitters.keys():
        print alphabet_type
        for i,mod_type in enumerate(transmitters[alphabet_type]):
            print "running test", i,mod_type

            tx_len = int(10e3)
            if mod_type.modname == "QAM64":
                tx_len = int(30e3)
            if mod_type.modname == "QAM16":
                tx_len = int(20e3)
            src = source_alphabet(alphabet_type, tx_len, True)
            mod = mod_type()
            fD = 1
            delays = [0.0, 0.9, 1.7]
            mags = [1, 0.8, 0.3]
            ntaps = 8
            noise_amp = 10**(-snr/10.0)
            print noise_amp
            #noise_amp = 0.1
            chan = channels.dynamic_channel_model( 200e3, 0.01, 1e2, 0.01, 1e3, 8, fD, True, 4, delays, mags, ntaps, noise_amp, 0x1337 )

            snk = blocks.vector_sink_c()

            tb = gr.top_block()

            # connect blocks
            if apply_channel:
                tb.connect(src, mod, chan, snk)
            else:
                tb.connect(src, mod, snk)
            tb.run()

            modulated_vector = np.array(snk.data(), dtype=np.complex64)
            if len(snk.data()) < min_length:
                min_length = len(snk.data())
                min_length_mod = mod_type
            output[(mod_type.modname, snr)] = modulated_vector

print "min length mod is %s with %i samples" % (min_length_mod, min_length)
# trim the beginning and ends, and make all mods have equal number of samples
start_indx = 100
fin_indx = min_length-100
for mod, snr in output:
 output[(mod,snr)] = output[(mod,snr)][start_indx:fin_indx]
X = timeseries_slicer.slice_timeseries_dict(output, 128, 64, 1000)
cPickle.dump( X, file("RML2014.04c_dict.dat", "wb" ) )
X = np.vstack(X.values())
cPickle.dump( X, file("RML2016.04c.dat", "wb" ) )