forked from uclacores/dual_path_network
-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathfrm_train_generator.py
166 lines (130 loc) · 6.95 KB
/
frm_train_generator.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
import numpy as np
from keras.utils import Sequence
from frm_dataset_creator2 import create_sample_fast
class train_generator(Sequence):
'Generates data for Keras'
def __init__(self, samples_per_epoch, batch_size,pkt_size,max_sps,mod_list,
sps_rng,pulse_ebw_list,timing_offset_rng,fading_spread_rng,freq_err_rng,phase_err_rng,snr_rng):
self.first_run = True
self.batch_size = batch_size
self.pkt_size = pkt_size
self.max_sps = max_sps
self.mod_list = mod_list
self.sps_rng = sps_rng
self.pulse_ebw_list = pulse_ebw_list
self.timing_offset_rng = timing_offset_rng
self.fading_spread_rng = fading_spread_rng
self.freq_err_rng = freq_err_rng
self.phase_err_rng = phase_err_rng
self.snr_rng = snr_rng
self.samples_per_epoch = samples_per_epoch
self.n_mods = len(mod_list)
def __len__(self):
return self.samples_per_epoch // self.batch_size
def __getitem__(self, index):
if self.first_run:
np.random.seed()
self.first_run = False
batch_size = self.batch_size
pkt_size = self.pkt_size
mod_list = self.mod_list
max_sps = self.max_sps
comb_v = np.zeros((batch_size,pkt_size,2))
carrier_v = np.zeros((batch_size,pkt_size,2))
fading_v = np.zeros((batch_size,pkt_size,2))
clean_v = np.zeros((batch_size,pkt_size,2))
timing_v = np.zeros((batch_size,pkt_size,2))
raw_v = np.zeros((batch_size,pkt_size,2))
mod_v = np.zeros((batch_size,len(mod_list)))
samp_weights_v = np.zeros((batch_size,))
mod_index = np.random.choice(self.n_mods,(batch_size,)).astype(np.int_)
mod_v[range(batch_size),mod_index] = 1
sps = np.random.uniform(self.sps_rng[0],self.sps_rng[1],(batch_size,))
pulse_ebw = np.random.choice(self.pulse_ebw_list,(batch_size,))
timing_offset = np.random.uniform(self.timing_offset_rng[0],self.timing_offset_rng[1],(batch_size,))
fading_spread = np.random.uniform(self.fading_spread_rng[0],self.fading_spread_rng[1],(batch_size,))
freq_err = np.random.uniform(self.freq_err_rng[0],self.freq_err_rng[1],(batch_size,))
phase_err = np.random.uniform(self.phase_err_rng[0],self.phase_err_rng[1],(batch_size,))
snr = np.random.uniform(self.snr_rng[0],self.snr_rng[1],(batch_size,))
if 'cpfsk' in mod_list:
cpfsk_loc = mod_index == mod_list.index('cpfsk')
else:
cpfsk_loc = np.zeros_like(mod_index,dtype='bool')
if 'gmsk' in mod_list:
gmsk_loc = mod_index == mod_list.index('gmsk')
else:
gmsk_loc = np.zeros_like(mod_index,dtype='bool')
non_linear_mod = np.logical_or( gmsk_loc,cpfsk_loc)
samp_weights_v = 1-np.exp(-0.4 *10**((snr)/10))
freq_v = (2*np.pi*freq_err)
for i in range(batch_size):
mod = self.mod_list[mod_index[i]]
op = create_sample_fast( mod = mod,pkt_len = self.pkt_size,sps=sps[i],pulse_ebw = pulse_ebw[i],
timing_offset = timing_offset[i],
fading_spread = fading_spread[i],
freq_err = freq_err[i], phase_err =phase_err[i],
snr = snr[i], max_sps = self.max_sps,
complex_fading=True, freq_in_hz = True,
seed = None)
comb_v[i] ,carrier_v[i],fading_v[i],clean_v[i],timing_v[i],raw_v[i],coeff = op
timing_step_v = np.floor(max_sps/sps)
timing_offNum_v = np.round(timing_offset*max_sps)
return ([comb_v], [freq_v,fading_v,fading_v,clean_v,timing_step_v,timing_offNum_v,mod_v],
[np.ones((batch_size,)),np.ones((batch_size,)),samp_weights_v,
samp_weights_v,samp_weights_v * np.logical_not(non_linear_mod),
samp_weights_v * np.logical_not(non_linear_mod),np.ones((batch_size,))])
class train_generator_mod(Sequence):
'Generates data for Keras'
def __init__(self, samples_per_epoch, batch_size,pkt_size,max_sps,mod_list,
sps_rng,pulse_ebw_list,timing_offset_rng,fading_spread_rng,freq_err_rng,phase_err_rng,snr_rng):
self.first_run = True
self.batch_size = batch_size
self.pkt_size = pkt_size
self.max_sps = max_sps
self.mod_list = mod_list
self.sps_rng = sps_rng
self.pulse_ebw_list = pulse_ebw_list
self.timing_offset_rng = timing_offset_rng
self.fading_spread_rng = fading_spread_rng
self.freq_err_rng = freq_err_rng
self.phase_err_rng = phase_err_rng
self.snr_rng = snr_rng
self.samples_per_epoch = samples_per_epoch
self.n_mods = len(mod_list)
def __len__(self):
return self.samples_per_epoch // self.batch_size
def __getitem__(self, index):
if self.first_run:
np.random.seed()
self.first_run = False
batch_size = self.batch_size
pkt_size = self.pkt_size
max_sps = self.max_sps
mod_list = self.mod_list
comb_v = np.zeros((batch_size,pkt_size,2))
carrier_v = np.zeros((batch_size,pkt_size,2))
fading_v = np.zeros((batch_size,pkt_size,2))
clean_v = np.zeros((batch_size,pkt_size,2))
timing_v = np.zeros((batch_size,pkt_size,2))
raw_v = np.zeros((batch_size,pkt_size,2))
mod_v = np.zeros((batch_size,len(mod_list)))
mod_index = np.random.choice(self.n_mods,(batch_size,)).astype(np.int_)
mod_v[range(batch_size),mod_index] = 1
sps = np.random.uniform(self.sps_rng[0],self.sps_rng[1],(batch_size,))
pulse_ebw = np.random.choice(self.pulse_ebw_list,(batch_size,))
timing_offset = np.random.uniform(self.timing_offset_rng[0],self.timing_offset_rng[1],(batch_size,))
fading_spread = np.random.uniform(self.fading_spread_rng[0],self.fading_spread_rng[1],(batch_size,))
freq_err = np.random.uniform(self.freq_err_rng[0],self.freq_err_rng[1],(batch_size,))
phase_err = np.random.uniform(self.phase_err_rng[0],self.phase_err_rng[1],(batch_size,))
snr = np.random.uniform(self.snr_rng[0],self.snr_rng[1],(batch_size,))
for i in range(batch_size):
mod = self.mod_list[mod_index[i]]
op = create_sample_fast( mod = mod,pkt_len = self.pkt_size,sps=sps[i],pulse_ebw = pulse_ebw[i],
timing_offset = timing_offset[i],
fading_spread = fading_spread[i],
freq_err = freq_err[i], phase_err =phase_err[i],
snr = snr[i], max_sps = self.max_sps,
complex_fading=True, freq_in_hz = True,
seed = None)
comb_v[i] ,carrier_v[i],fading_v[i],clean_v[i],timing_v[i],raw_v[i],coeff = op
return [comb_v], [mod_v]