-
Notifications
You must be signed in to change notification settings - Fork 0
/
SU_Waterfilling.py
361 lines (255 loc) · 10.7 KB
/
SU_Waterfilling.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
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
import numpy as np
import numpy as np
# from Detector import Detector
import scipy.io as sio
from utils import *
from Dataset_Mu import *
from Mu_detector_np import *
from Mu_detector_coding import *
# from CommonCom import *
np.random.seed(667)
from utils_LDPC import *
class Mu_precode(object):
def __init__(self,params):
self.Nr = params['Nr'] # 接收天线数 32
self.Nt = params['Nt'] # 用户发送天线数 4
self.User = params['User'] #用户数
self.constellation = params['constellation']
# self.M = np.shape(params['constellation'])[0]
self.M = params['M'] #调制阶数
#仿真参数
self.Ns = self.Nt * self.User
self.SNR_dBs = np.arange(params['SNR_dB_min_test'], params['SNR_dB_max_test'], params['SNR_step_test'])
self.test_symbol = params['test_symbol']
def Waterfill3(self, H, sigma2):
"""
:param H:
:param sigma2:
:return:
"""
# sigma2 = sigma2/self.Nt
Rh = np.conj(H).T @ H
V, lam, Vh = np.linalg.svd(Rh)
# _,lam,_=np.linalg.svd(H)
P = np.zeros(self.Nt)
r = np.shape(lam)[0]
ignore = 0
sigma2 = 1/sigma2
while ignore < self.Nt:
left = self.Nt - ignore
temp = np.sum(1 / lam[:left])
mu = self.Nt / (r - ignore) * (1 + 1 / sigma2 * temp)
for i in range(left):
P[i] = mu - self.Nt / (sigma2 * lam[i])
if P[left - 1] < 0:
ignore += 1
P[left - 1] = 0
else:
break
L_send = self.Nt - ignore
P_eff = P[:L_send]
U_H, S_H, Vh_H = np.linalg.svd(H)
S_eff = np.zeros((self.Nr, L_send))
for i in range(L_send):
S_eff[i, i] = np.sqrt(P_eff[i]) * S_H[i]
H_eff = U_H @ S_eff
return H_eff, L_send, U_H, S_eff
def Waterfill2(self,H,sigma2):
"""
:param H:
:param sigma2:
:return:
"""
# sigma2 = sigma2/self.Nt
Rh = np.conj(H).T@H
V,lam,Vh = np.linalg.svd(Rh)
# _,lam,_=np.linalg.svd(H)
P = np.zeros(self.Nt)
r = np.shape(lam)[0]
ignore = 0
while ignore<self.Nt:
left = self.Nt-ignore
temp = np.sum(1/lam[:left])
mu = self.Nt/(r-ignore) * (1+1/sigma2* temp)
for i in range(left):
P[i] = mu - self.Nt/(sigma2*lam[i])
if P[left-1] < 0:
ignore+=1
P[left-1]= 0
else:
break
L_send = self.Nt - ignore
P_eff = P[:L_send]
U_H, S_H, Vh_H = np.linalg.svd(H)
S_eff = np.zeros((self.Nr, L_send))
for i in range(L_send):
S_eff[i, i] = np.sqrt(P_eff[i]) * S_H[i]
H_eff = U_H @ S_eff
return H_eff, L_send,U_H, S_eff
def Waterfill1(self,H,sigma2):
"""
:param H:
:param sigma2:
:return:
备注2021.6.5: 似乎有一些问题,注水过程重为何总功率改变??????
"""
Rn = 1/sigma2 * np.eye(self.Nr)
Rh = np.conj(H).T@Rn@H
V,Lam,Vh = np.linalg.svd(Rh)
P = np.zeros(self.Nt)
flag = False
sqrtrsum = np.sum(1 / np.sqrt(Lam))
rsum = self.Nt + np.sum(1 / Lam)
sqrtwater = sqrtrsum / rsum
for j in range(self.Nt):
P[j] = 1 / sqrtwater * 1 / np.sqrt(Lam[j]) - 1 / Lam[j]
if P[j] < 0:
flag = True
ignore = 0
while flag:
flag = False
sqrtrsum = sqrtrsum - 1 / np.sqrt(Lam[-1 - ignore])
rsum = rsum - 1 - 1 / Lam[-1 - ignore] # 带着ZF检测器一起算?
sqrtwater = sqrtrsum / rsum
for j in range(self.Nt):
P[j] = 1 / sqrtwater * 1 / np.sqrt(Lam[j]) - 1 / Lam[j]
if P[j] < 0:
P[j] = 0
if j < self.Nt - ignore - 1:
flag = True
ignore = ignore + 1
# 功率注水的对角矩阵:
# P = np.matmul(V,P)
L_send = self.Nt - ignore
P_eff = P[self.Nt-L_send:self.Nt]
U_H,S_H,Vt_H = np.linalg.svd(H)
S_eff = np.zeros((self.Nt,L_send))
for i in range(L_send):
S_eff[i,i] = np.sqrt(P_eff[i])*S_H[i]
H_eff = U_H@S_eff
return H_eff,L_send,
def MMSE(self,y,H,sigma2):
HTH = np.conj(H).T@H
HTH_inv = np.linalg.inv((np.matmul(np.conj(H).T, H)+ sigma2*np.eye(HTH.shape[0])))
Hty = np.conj(H).T@y
res = HTH_inv@Hty
# res = np.expand_dims(np.array(res), axis=0)
res = res.T
res = demodulate_np(res, self.constellation)
self.x_hat = res
return res
def svd_solve(self,y,U,L_send,P_eff):
y = np.squeeze(y)
temp = np.conj(U).T@y
# res = np.expand_dims(np.array(res), axis=0)
# P_show = P_eff
P_inv = np.linalg.pinv(P_eff)
# ob = np.zeros((1,L_send),dtype=np.complex)
# for i in range(L_send):
# ob[0,i] = temp[i]/P_eff[i,i]
res = P_inv@temp
# res =res.T
res = np.expand_dims(res,axis=0)
res = demodulate_np(res, self.constellation)
return res
def test_func_normal(self):
ser_all = []
for i in range(self.SNR_dBs.shape[0]):
ser = 0.
print('======================正在仿真SNR:%ddB================================' % (self.SNR_dBs[i]))
for j in range(self.test_symbol):
y = np.zeros((self.Nr, 1), dtype=np.complex)
H_all_list = []
x_all_list = []
sigma2_all = self.Nt * self.User / (np.power(10, self.SNR_dBs[i] / 10) )
for k in range(self.User):
########### 仅测试用################
Hr = np.random.randn(self.Nr, self.Nt) / np.sqrt(2)
Hi = np.random.randn(self.Nr, self.Nt) / np.sqrt(2)
H = Hr + 1j * Hi
# H = np.eye(2) + 1j * np.eye(2)
######################################
sigma2 = self.Nt / (np.power(10, self.SNR_dBs[
i] / 10)) # sigma2 = self.Nt / (np.power(10, snr / 10) * self.Nr)
H_eff, L_send = H, self.Nt
H_all_list.append(H_eff)
s_index = np.random.randint(low=0, high=len(self.constellation), size=[self.Nt, 1])
x = self.constellation[s_index]
noise = np.sqrt(sigma2 / 2) * np.random.randn(self.Nr, 1) + 1j * np.sqrt(
sigma2 / 2) * np.random.randn(self.Nr, 1)
y += H_eff @ x + noise
x_all_list.append(x)
H_all = np.concatenate(H_all_list, axis=1)
x_all = np.concatenate(x_all_list, axis=0)
x_hat = self.MMSE(y, H_all, sigma2_all)
x_hat = np.expand_dims(x_hat, axis=1)
ser += accuracy(x_all, x_hat) / self.test_symbol
# print("current_norm_ser:", ser)
ser_all.append(ser)
return ser_all
def test_func(self):
ser_all = []
for i in range(self.SNR_dBs.shape[0]):
ser = 0.
print('======================正在仿真SNR:%ddB================================' % (self.SNR_dBs[i]))
for j in range(self.test_symbol):
y = np.zeros((self.Nr,1),dtype=np.complex)
H_all_list = []
x_all_list = []
sigma2_all = self.Nt*self.User / (np.power(10, self.SNR_dBs[i] / 10)) # sigma2 = self.Nt / (np.power(10, snr / 10) * self.Nr) (不需要再除self.Nr)
for k in range(self.User):
########### 仅测试用################
Hr = np.random.randn(self.Nr, self.Nt)/np.sqrt(2)
Hi = np.random.randn(self.Nr, self.Nt)/np.sqrt(2)
H = Hr + 1j * Hi
# H = np.eye(2) + 1j * np.eye(2)
######################################
sigma2 = self.Nt / (np.power(10,self.SNR_dBs[i] / 10)) # sigma2 = self.Nt / (np.power(10, snr / 10) * self.Nr)
H_eff,L_send,U,P_eff = self.Waterfill2(H,1/(sigma2/self.Nt))
# H_eff, L_send, U, P_eff = self.Waterfill2(H, sigma2 / self.Nt)
H_all_list.append(H_eff)
s_index = np.random.randint(low=0, high=len(self.constellation), size=[L_send, 1])
x = self.constellation[s_index]
noise = np.sqrt(sigma2 / 2) * np.random.randn(self.Nr,1) + 1j * np.sqrt(sigma2 / 2) * np.random.randn(self.Nr,1)
y += H_eff@x + noise
x_all_list.append(x)
H_all = np.concatenate(H_all_list,axis=1)
x_all = np.concatenate(x_all_list,axis=0)
x_hat = self.MMSE(y,H_all,sigma2_all)
# x_hat = self.svd_solve(y,U,L_send,P_eff)
x_hat = np.expand_dims(x_hat,axis=1)
ser += accuracy(x_all, x_hat) / self.test_symbol
# print("current_ser:", ser)
ser_all.append(ser)
return ser_all
if __name__ == "__main__":
params = {
# 二选一
# 'dataset_dir': r'D:\Nr8Nt8batch_size500mod_nameQAM_4', # 使用固定数据集
# 'dataset_dir': "./H_data/H.mat", # 程序运行时生成数据集
'dataset_dir': None, # 程序运行时生成数据集
# ************************程序运行之前先检查下面的参数*****************
# 仿真参数
# 'constellation': np.array([0.7071, -0.7071], dtype=np.float32),
'constellation': sio.loadmat('QAM16.mat')['QAM_16'][0],
'bitmap': {0.7071 + 0.7071j:[0,0],-0.7071 + 0.7071j:[0,1],0.7071 - 0.7071j:[1,0],-0.7071 - 0.7071j:[1,1]}, # 该映射也可以自动生成
'Nt': 4, # Number of transmit antennas
'Nr': 32, # Number of receive antennas
'User': 1, # Number of Users
'M':16,
# 测试检测算法的信噪比,一般误符号率到1e-4就行
'SNR_dB_min_test':0, # Minimum SNR value in dB for simulation
'SNR_dB_max_test':9, # Maximum SNR value in dB for simulation
'SNR_step_test': 1,
'test_symbol': 1000,
#信道编译码新加字段
'rate':1/3,
'test_code':1000,
'load_dir':"Tanner_R13_K120_Z12_BG2.mat",
'Z':12
}
Mu_pre = Mu_precode(params)
ser_all = Mu_pre.test_func()
print("Water_fill", ser_all)
ser_all_norm = Mu_pre.test_func_normal()
print("MMSE_norm",ser_all_norm)