-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathexample_code.py
345 lines (290 loc) · 17.5 KB
/
example_code.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
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
import os
os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'
import tensorflow as tf
import numpy as np
from pathlib import Path
import argparse
import sys
import pickle
tf.get_logger().setLevel('ERROR')
# set root path
root_path = str(Path(__file__).parent)
sys.path.append(root_path)
from args import BasicConfig
import sionna
from sionna.ofdm.channel_estimation import BaseChannelEstimator, LSChannelEstimator, LMMSEInterpolator
# ----------------------------- ARGUMENT CONFIG ----------------------------- #
parser = argparse.ArgumentParser()
# path config
parser.add_argument("--num_bs_ant", type=int, default=4, required=False)
parser.add_argument("--fft_size", type=int, default=48, required=False)
parser.add_argument("--num_bits_per_symbol", type=int, default=4, required=False)
parser.add_argument("--num_pilots", type=int, default=2, required=False) # i.e., the number of symbols in each TTI that are used for pilots
parser.add_argument("--batch_size", type=int, default=32, required=False)
# evaluation config
parser.add_argument("--min_ebNo", type=float, default=-5, required=False)
parser.add_argument("--max_ebNo", type=float, default=10, required=False)
parser.add_argument("--num_ebNo_points", type=int, default=16, required=False)
parser.add_argument("--max_mc_iter", type=int, default=100, required=False)
# channel config
parser.add_argument('--model_type', type=str, default='TDL', required=False)
parser.add_argument('--PDP_mode', type=str, default='B', required=False)
parser.add_argument('--delay_spread', type=float, default=100, required=False)
parser.add_argument('--min_speed', type=float, default=10.0, required=False)
parser.add_argument('--max_speed', type=float, default=10.0, required=False)
parser.add_argument('--delta_delay_spread', type=float, default=0.0, required=False)
run_args = parser.parse_args()
print(run_args)
class BasicConfig():
def __init__(self,
num_bs_ant: int = 4,
fft_size: int = 48,
num_bits_per_symbol: int = 4,
pilot_ofdm_symbol_indices: list = [2,11]
):
super().__init__()
# Default system parameters
# self._PDP_list = ['A', 'B', 'C', 'D', 'E']
self._cyclic_prefix_length = 6
self._pilot_ofdm_symbol_indices = pilot_ofdm_symbol_indices # two pilot configuration: the 2nd and 11th OFDM symbols are pilots
self._num_ut_ant = 1
self._num_bs_ant = num_bs_ant
self._carrier_frequency = 4e9
self._subcarrier_spacing = 15e3
self._fft_size = fft_size # number of subcarriers = 48 = 4 PRBs
self._num_ofdm_symbols = 14 # per slot
self._num_streams_per_tx = 1
self._dc_null = True
self._num_guard_carriers = [5, 6]
self._pilot_pattern = "kronecker"
self._num_bits_per_symbol = num_bits_per_symbol # 16QAM
self._coderate = 658/1024
# Required system components
self._sm = sionna.mimo.StreamManagement(np.array([[1]]),
self._num_streams_per_tx)
self._rg = sionna.ofdm.ResourceGrid(num_ofdm_symbols=self._num_ofdm_symbols,
fft_size=self._fft_size,
subcarrier_spacing = self._subcarrier_spacing,
num_tx=1,
num_streams_per_tx=self._num_streams_per_tx,
cyclic_prefix_length=self._cyclic_prefix_length,
num_guard_carriers=self._num_guard_carriers,
dc_null=self._dc_null,
pilot_pattern=self._pilot_pattern,
pilot_ofdm_symbol_indices=self._pilot_ofdm_symbol_indices)
# all the databits carried by the resource grid with size `fft_size`x`num_ofdm_symbols` form a single codeword.
self._n = int(self._rg.num_data_symbols * self._num_bits_per_symbol) # Codeword length.
self._k = int(self._n*self._coderate) # Number of information bits per codeword
self._ut_array = sionna.channel.tr38901.Antenna(polarization="single",
polarization_type="V",
antenna_pattern="38.901",
carrier_frequency=self._carrier_frequency)
self._bs_array = sionna.channel.tr38901.AntennaArray(num_rows=1,
num_cols= int(self._num_bs_ant/2),
polarization="dual",
polarization_type="VH",
antenna_pattern="38.901",
carrier_frequency=self._carrier_frequency)
self._frequencies = sionna.channel.subcarrier_frequencies(self._rg.fft_size, self._rg.subcarrier_spacing)
# Apply channel frequency response
self._channel_freq = sionna.channel.ApplyOFDMChannel(add_awgn=True)
self._binary_source = sionna.utils.BinarySource() # seed = none
self._encoder = sionna.fec.ldpc.encoding.LDPC5GEncoder(self._k, self._n)
self._mapper = sionna.mapping.Mapper("qam", self._num_bits_per_symbol)
self._rg_mapper = sionna.ofdm.ResourceGridMapper(self._rg)
# know the pilot positions on the resource grid; first estimate the channel based on pilots; then interpolation
self._lmmse_equ = sionna.ofdm.LMMSEEqualizer(self._rg, self._sm)
self._remove_nulled_scs = sionna.ofdm.RemoveNulledSubcarriers(self._rg)
def set_channel_models( self,
model_type: str= 'TDL',
PDP_mode: str = 'A', # choose from ['A', 'B', 'C', 'D', 'E']
delay_spread: float = 30e-9,
min_speed: float = 0.0,
max_speed: float = 0.0,
delta_delay_spread: float = 0.0):
self._PDP = PDP_mode
self._delay_spread = delay_spread # this should be in the unit of 's'...
self._min_speed = min_speed
self._max_speed = max_speed
self._model_type = model_type
# ---------------------------------------------------------------------------- #
if self._model_type == 'TDL':
# no spatial correlation is considered
self._comm_channel_model = sionna.channel.tr38901.TDL(model=self._PDP,
delay_spread=self._delay_spread,
carrier_frequency=self._carrier_frequency,
min_speed=self._min_speed,
max_speed=self._max_speed,
num_rx_ant=self._num_bs_ant,
num_tx_ant=self._num_ut_ant) # set random_seed to generate random uniform doppler, phi, and theta
elif model_type == 'CDL':
self._comm_channel_model = sionna.channel.tr38901.CDL(model=self._PDP,
delay_spread=self._delay_spread,
carrier_frequency=self._carrier_frequency,
ut_array=self._ut_array,
bs_array=self._bs_array,
direction="uplink",
min_speed=self._min_speed,
max_speed=self._max_speed)
else:
raise ValueError("model_type must be either TDL or CDL")
# ---------------------------------------------------------------------------- #
class _EvalBsRx(tf.keras.Model):
def __init__(self,
perfect_csi: bool,
config: BasicConfig,
lmmse_order: str = None,
freq_cov_mat: np.ndarray = None,
time_cov_mat: np.ndarray = None,
space_cov_mat: np.ndarray = None,
coded: bool = True,
int_method: str = 'nn',
det_method: str = 'lmmse'):
super(_EvalBsRx, self).__init__(name='_EvalBsRx')
self._perfect_csi = perfect_csi
self._config = config
self._removed_null_subc = sionna.ofdm.RemoveNulledSubcarriers(self._config._rg)
self._coded = coded
# --------------------------------- Detection -------------------------------- #
# Channel estimators
if int_method == 'nn':
self._channel_estimator = LSChannelEstimator(self._config._rg, interpolation_type='nn')
elif int_method == 'lin':
self._channel_estimator = LSChannelEstimator(self._config._rg, interpolation_type='lin')
elif int_method == 'lmmse':
freq_cov_mat = tf.constant(freq_cov_mat, tf.complex64)
time_cov_mat = tf.constant(time_cov_mat, tf.complex64)
space_cov_mat = tf.constant(space_cov_mat, tf.complex64)
lmmse_int_freq_first = LMMSEInterpolator(self._config._rg.pilot_pattern, time_cov_mat, freq_cov_mat, space_cov_mat, order=lmmse_order)
self._channel_estimator = LSChannelEstimator(self._config._rg, interpolator=lmmse_int_freq_first)
if coded:
self._demapper = sionna.mapping.Demapper("app", "qam", self._config._num_bits_per_symbol, hard_out = False) # output LLR for each bits
self._decoder = sionna.fec.ldpc.decoding.LDPC5GDecoder(self._config._encoder, hard_out=True)
else:
self._demapper = sionna.mapping.Demapper("app", "qam", self._config._num_bits_per_symbol, hard_out = True) # output hard bits!!
@tf.function
def call(self,
batch_size: int,
ebno_db: float):
# -------------------------------- Transmitter ------------------------------- #
batch_N0 = sionna.utils.ebnodb2no(ebno_db,
self._config._num_bits_per_symbol,
self._config._coderate,
self._config._rg)
if self._coded == True:
# use the outer encoder during training
b = self._config._binary_source([batch_size, 1, self._config._num_streams_per_tx, self._config._k])
tx_codeword_bits = self._config._encoder(b)
else:
#to reduce the computational complexity, the outer encoder (and decoder) are not used at training
b = self._config._binary_source([batch_size, 1, self._config._num_streams_per_tx, self._config._n])
tx_codeword_bits = b
batch_x = self._config._mapper(tx_codeword_bits)
batch_x_rg = self._config._rg_mapper(batch_x)
# ---------------------------- Through the Channel --------------------------- #
cir = self._config._comm_channel_model(batch_size, self._config._rg.num_ofdm_symbols, 1/self._config._rg.ofdm_symbol_duration)
batch_h_freq = sionna.channel.cir_to_ofdm_channel(self._config._frequencies, *cir, normalize=True) # this is real channel freq response
batch_y = self._config._channel_freq([batch_x_rg, batch_h_freq, batch_N0])
# ------------------------------- channel estimation ------------------------------- #
if self._perfect_csi:
batch_h_hat = self._removed_null_subc(batch_h_freq)
batch_err_var = 0.0
else:
batch_h_hat, batch_err_var = self._channel_estimator([batch_y, batch_N0]) # why here do not need pilot config
# ------ Detection(direct detection/equalization + demapping) + Decoding ------ #
batch_x_hat, batch_no_eff = self._config._lmmse_equ([batch_y, batch_h_hat, batch_err_var, batch_N0])
if self._coded:
batch_llr = self._demapper([batch_x_hat, batch_no_eff])
b_hat = self._decoder(batch_llr)
else:
b_hat = self._demapper([batch_x_hat, batch_no_eff])
return b, b_hat
class EvalBsRx(tf.keras.Model):
def __init__(self,
perfect_csi: bool,
config: BasicConfig, # need to tune the link-level config
ebNo_dB_range: np.ndarray,
result_save_path: list,
batch_size: int=64,
num_target_block_errors: int=1000,
max_mc_iter: int=200,
lmmse_order: str = 't-f-s',
coded: bool=True,
int_method: str='nn',
det_method: str='lmmse'):
super(EvalBsRx, self).__init__(name='EvalBsRx')
self._eval_obj = _EvalBsRx(perfect_csi=perfect_csi,
config=config,
lmmse_order = lmmse_order,
coded = coded,
int_method=int_method,
det_method=det_method)
self._config = config
self._ebNo_dB_range = ebNo_dB_range
self._result_save_path = result_save_path
self._batch_size = batch_size
self._num_target_block_errors = num_target_block_errors
self._max_mc_iter = max_mc_iter
self._batch_size = batch_size
def eval(self):
ber, bler = sionna.utils.sim_ber(self._eval_obj,
self._ebNo_dB_range,
batch_size=self._batch_size,
num_target_block_errors=self._num_target_block_errors,
max_mc_iter=self._max_mc_iter,
verbose=True)
ber = ber.numpy()
bler = bler.numpy()
with open(self._result_save_path[0], "wb") as file:
pickle.dump(ber, file)
with open(self._result_save_path[1], "wb") as file:
pickle.dump(bler, file)
return ber, bler
# ---------------------------------------------------------------------------- #
if __name__ == "__main__":
eval_result_save_folder = root_path + "/Baselines/" + run_args.model_type + run_args.PDP_mode + "_{:d}Nr_{:d}fft_{:d}bits_{:d}pilots".format(run_args.num_bs_ant, run_args.fft_size, run_args.num_bits_per_symbol, run_args.num_pilots)
if not os.path.exists(eval_result_save_folder):
os.makedirs(eval_result_save_folder)
if run_args.num_pilots == 2:
pilot_ofdm_symbol_indices = [2,11]
elif run_args.num_pilots == 1:
pilot_ofdm_symbol_indices = [2]
link_config = BasicConfig(num_bs_ant=run_args.num_bs_ant,
fft_size=run_args.fft_size,
num_bits_per_symbol=run_args.num_bits_per_symbol,
pilot_ofdm_symbol_indices=pilot_ofdm_symbol_indices)
link_config.set_channel_models(model_type=run_args.model_type,
PDP_mode=run_args.PDP_mode,
delay_spread=run_args.delay_spread, # this should be in the unit of second
min_speed=run_args.min_speed,
max_speed=run_args.max_speed,
delta_delay_spread=run_args.delta_delay_spread)
# ---------------------------------------------------------------------------- #
eval_obj = EvalBsRx(perfect_csi=False,
config = link_config,
ebNo_dB_range=np.linspace(run_args.min_ebNo,
run_args.max_ebNo,
run_args.num_ebNo_points),
batch_size=run_args.batch_size,
result_save_path=[str(Path(eval_result_save_folder).joinpath('lmmse_ber.pkl')),
str(Path(eval_result_save_folder).joinpath('lmmse_bler.pkl'))],
max_mc_iter=run_args.max_mc_iter,
coded=True,
int_method='nn')
# ---------------------------------------------------------------------------- #
ideal_eval_obj = EvalBsRx(perfect_csi=True,
config = link_config,
ebNo_dB_range=np.linspace(run_args.min_ebNo,
run_args.max_ebNo,
run_args.num_ebNo_points),
batch_size=run_args.batch_size,
result_save_path=[str(Path(eval_result_save_folder).joinpath('ideal_ber.pkl')),
str(Path(eval_result_save_folder).joinpath('ideal_bler.pkl'))],
max_mc_iter=run_args.max_mc_iter,
coded=True,
int_method='nn')
# ------------------------------ Run evaluations ----------------------------- #
lmmse_ber, lmmse_bler = eval_obj.eval()
ideal_ber, ideal_bler = ideal_eval_obj.eval()
print(lmmse_ber, lmmse_bler)
print(ideal_ber, ideal_bler)