-
Notifications
You must be signed in to change notification settings - Fork 5
/
Client.py
428 lines (383 loc) · 14.6 KB
/
Client.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
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
import warnings
warnings.filterwarnings("ignore")
import ssl
ssl._create_default_https_context = ssl._create_unverified_context
import socket
import threading
import dill
import tensorflow as tf
from tqdm import tqdm
import time
import json
import numpy as np
from collections import namedtuple
#import math
from Model import AlexNet
from tensorflow.keras.utils import to_categorical
import csv
import sys
data_size = None
client_vars = None
ID = None
client_state = 0
global_vars = None
input_shape = None
num_classes = None
learning_rate = None
client_dataset_train_a = None
client_dataset_train_b = None
client_dataset_test_a = None
client_dataset_test_b = None
client_epoch = None
batch_size = None
if_to_start = 0
if_to_stop = 0
next_epoch_var = 0
next_epoch_nowepoch = 0
test_batchsize = None
drop_rate = None
now_epoch= 0
##########网络部署###########
Client_HOST=sys.argv[1]
Client_PORT=int(sys.argv[2])
Client_PORT_TCP = int(sys.argv[3])
Server_HOST=sys.argv[4]
Server_PORT=int(sys.argv[5])
Server_PORT_TCP =int(sys.argv[6])
s= socket.socket(socket.AF_INET,socket.SOCK_DGRAM)
s.bind((Client_HOST,Client_PORT))
tcp_tunnel_s=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
tcp_tunnel_s.bind((Client_HOST, Client_PORT_TCP))
tcp_tunnel_s.listen(5)
symbol=str(time.time())+"_ID="+str(ID)+"_HOST="+Client_HOST+"_PORT="+str(Client_PORT)+".csv"
def sendmsg(mysocket,senddata,HOST,PORT):
senddata = dill.dumps(senddata)
mysocket.sendto(senddata, (HOST, PORT))
print(HOST, PORT)
return 0
def recvmsg(mysocket):
while 1:
try:
clientdata, addr = mysocket.recvfrom(4096)
clientdata = dill.loads(clientdata)
print("get:",clientdata,"from:",addr)
deal_recv(clientdata, addr)
except:
pass
return 0
def tcp_tunnel_send(data,HOST,PORT):
global now_epoch, symbol
while 1:
try:
tcp_start_time=time.time()
senddata = dill.dumps(data)
print(sys.getsizeof(senddata))
tcp_tunnel_ss=socket.socket(socket.AF_INET, socket.SOCK_STREAM)
tcp_tunnel_ss.connect((HOST, PORT))
tcp_tunnel_ss.sendall(senddata)
print("TCP 传输完成")
tcp_end_time=time.time()
tcp_tunnel_ss.close()
f = open(symbol,"a",encoding="UTF-8",newline="")
csvwriter=csv.writer(f)
csvwriter.writerow([now_epoch, "", "", time.time(), tcp_end_time-tcp_start_time, "tcp_send"])
f.close()
return 0
except:
print("TCP传输失败,等待2s后重新尝试传输"+str(HOST)+"//"+str(PORT))
time.sleep(2)
return 0
def tcp_tunnel_recv(tcp_tunnel_s):
global now_epoch, symbol
while 1:
try:
tcp_start_time_recv=time.time()
sock, addr = tcp_tunnel_s.accept()
total_data = b''
data = sock.recv(1024)
total_data += data
num = len(data)
# 如果没有数据了,读出来的data长度为0,len(data)==0
with tqdm(total=np.ceil(18500)) as bar:
while len(data) > 0:
data = sock.recv(1024)
num += len(data)
total_data += data
bar.update(1)
clientdata=total_data
print("获取TCP传输,来自" + str(addr))
clientdata=dill.loads(clientdata)
tcp_end_time_recv=time.time()
f = open(symbol,"a",encoding="UTF-8",newline="")
csvwriter=csv.writer(f)
csvwriter.writerow([now_epoch, "", "", time.time(), tcp_end_time_recv - tcp_start_time_recv, "tcp_recv"])
f.close()
deal_recv(clientdata, addr)
except:
pass
return 0
thread_recvmsg=threading.Thread(target=recvmsg, args=(s,))
thread_recvmsg.start()
thread_recvmsg_tcp=threading.Thread(target=tcp_tunnel_recv, args=(tcp_tunnel_s,))
thread_recvmsg_tcp.start()
def deal_recv(clientdata, addr):
if clientdata["msg"]=="set_global_vars":
set_global_vars(clientdata)
if clientdata["msg"]=="client_init_data":
client_init_data(clientdata)
if clientdata["msg"]=="client_dataset":
client_dataset(clientdata)
if clientdata["msg"]=="distribute_model":
distribute_model(clientdata)
if clientdata["msg"]=="detect_client":
detect_client(clientdata)
if clientdata["msg"]=="start":
client_start(clientdata)
if clientdata["msg"]=="server_now_epoch":
set_now_epoch(clientdata)
if clientdata["msg"]=="stop":
client_stop(clientdata)
return 0
################网络部署########################
def set_global_vars(clientdata):
global global_vars, next_epoch_var, learning_rate
global_vars=clientdata["global_vars"]
learning_rate=clientdata["learning_rate"]
next_epoch_var=1
print("执行:set_global_vars,已设置:global_vars")
return 0
def client_init_data(clientdata):
global input_shape, num_classes, learning_rate
input_shape=clientdata["input_shape"]
num_classes=clientdata["num_classes"]
learning_rate=clientdata["learning_rate"]
print("执行:client_init_data,已设置:input_shape num_classes learning_rate")
return 0
def client_dataset(clientdata):
global client_dataset_train_a, client_dataset_train_b, client_dataset_test_a, client_dataset_test_b, client_state
client_dataset_train_a=clientdata["client_dataset_train_a"]
client_dataset_train_b=clientdata["client_dataset_train_b"]
client_dataset_test_a=clientdata["client_dataset_test_a"]
client_dataset_test_b=clientdata["client_dataset_test_b"]
print("执行:client_dataset,已设置:dataset")
if client_state==0:
client_state=1
return 0
if client_state==1:
return 0
if client_state==2:
client_state=3
return 0
return 0
def distribute_model(clientdata):
global client_epoch, batch_size, test_batchsize, drop_rate
client_epoch=clientdata["client_epoch"]
batch_size=clientdata["batchsize"]
test_batchsize = clientdata["test_batchsize"]
drop_rate = clientdata["drop_rate"]
print("执行:distribute_model,已设置:client_epoch batchsize")
check_client_state()
return 0
def detect_client(clientdata):
global ID, client_state, s, Server_HOST, Server_PORT
ID=clientdata["ID"]
print("已被Server发现,本Client的ID为"+str(ID))
detect_client_recv={
"msg": "detect_client_recv",
"ID": ID,
"client_state": client_state
}
print("已响应Server的呼唤...当前client_state="+str(client_state))
sendmsg(s, detect_client_recv, Server_HOST, Server_PORT)
return 0
def client_start(clientdata):
global if_to_start
if_to_start = 1
print("服务器已下达开始命令")
return 0
def client_stop(clientdata):
global if_to_stop
if_to_stop = 1
print("服务器已通知,训练结束")
return 0
def set_now_epoch(clientdata):
global now_epoch, next_epoch_nowepoch
now_epoch=clientdata["server_now_epoch"]
print("大循环已进行到第"+str(now_epoch)+"轮")
next_epoch_nowepoch=1
return 0
def check_client_state():
global client_state, ID, input_shape, num_classes, learning_rate, client_epoch, batch_size, test_batchsize, drop_rate
if ID is None:
print("ID is None")
if input_shape is None:
print("input_shape is None")
if num_classes is None:
print("num_classes is None")
if learning_rate is None:
print("learning_rate is None")
if client_epoch is None:
print("client_epoch is None")
if batch_size is None:
print("batch_size is None")
if test_batchsize is None:
print("test_batchsize is None")
if drop_rate is None:
print("drop_rate is None")
if (ID is not None) and (input_shape is not None) and (num_classes is not None) and (learning_rate is not None) and (client_epoch is not None) and (batch_size is not None) and (test_batchsize is not None) and (drop_rate is not None):
if client_state==0:
client_state=2
return 0
if client_state==1:
client_state=3
return 0
if client_state==2:
return 0
return 0
def get_client_vars(mysocket, epoch):
global Server_HOST, Server_PORT, ID, data_size, client_vars
msg={
"msg":"get_client_vars",
"datasize": data_size,
"client_vars": client_vars,
"epoch": epoch,
"ID": ID
}
tcp_tunnel_send(msg,Server_HOST,Server_PORT_TCP)
def test_client(mysocket, epoch, acc, loss, time_slot):
global ID, Server_PORT, Server_HOST, test_batchsize, client_vars
msg={
"msg":"test_client",
"acc":acc,
"loss":loss,
"ID": ID,
"epoch": epoch,
"test_batchsize": test_batchsize,
"time_slot": time_slot
}
sendmsg(mysocket,msg,Server_HOST,Server_PORT)
FedModel = namedtuple('FedModel', 'X Y DROP_RATE train_op loss_op acc_op')
class Clients:
def __init__(self, input_shape, num_classes, learning_rate, client_dataset_train_a, client_dataset_train_b, client_dataset_test_a, client_dataset_test_b):
global data_size
self.graph = tf.Graph()
self.sess = tf.Session(graph=self.graph)
# Call the create function to build the computational graph of AlexNet
net = AlexNet(input_shape, num_classes, learning_rate, self.graph)
self.model = FedModel(*net)
# initialize
with self.graph.as_default():
self.sess.run(tf.global_variables_initializer())
# Load Cifar-10 dataset
# NOTE: len(self.dataset.train) == clients_num
(x_train, y_train), (x_test, y_test) = tf.keras.datasets.cifar10.load_data()
y_train = to_categorical(y_train, 10)
y_test = to_categorical(y_test, 10)
x_train = x_train.astype('float32') / 255
x_test = x_test.astype('float32') / 255
self.dataset_x_train = x_train[client_dataset_train_a:client_dataset_train_b]
self.dataset_y_train = y_train[client_dataset_train_a:client_dataset_train_b]
self.dataset_x_test = x_test[client_dataset_test_a:client_dataset_test_b]
self.dataset_y_test = y_test[client_dataset_test_a:client_dataset_test_b]
self.dataset_train_size = client_dataset_train_b - client_dataset_train_a + 1
self.dataset_test_size = client_dataset_test_b - client_dataset_test_a + 1
data_size = self.dataset_train_size
def run_test(self, test_batchsize):
with self.graph.as_default():
if (test_batchsize<1) or (test_batchsize>self.dataset_test_size):
test_batchsize=self.dataset_test_size
batch_x = self.dataset_x_test[0:test_batchsize]
batch_y = self.dataset_y_test[0:test_batchsize]
feed_dict = {
self.model.X: batch_x,
self.model.Y: batch_y,
self.model.DROP_RATE: 0
}
return self.sess.run([self.model.acc_op, self.model.loss_op],
feed_dict=feed_dict)
def train_epoch(self, batch_size=32, dropout_rate=0.5):
"""
Train one client with its own data for one epoch
cid: Client id
"""
datasize = self.dataset_train_size
temp = 0
print("start_train_epoch")################################################################################
with self.graph.as_default():
with tqdm(total=np.ceil(datasize/batch_size)) as bar:
while temp < (datasize-batch_size):
batch_x = self.dataset_x_train[temp:temp+batch_size]
batch_y = self.dataset_y_train[temp:temp+batch_size]
feed_dict = {
self.model.X: batch_x,
self.model.Y: batch_y,
self.model.DROP_RATE: dropout_rate
}
self.sess.run(self.model.train_op, feed_dict=feed_dict)
temp=temp+batch_size
bar.update(1)
return datasize
def get_client_vars_class(self):
""" Return all of the variables list """
with self.graph.as_default():
client_vars = self.sess.run(tf.trainable_variables())
return client_vars
def set_global_vars_class(self, global_vars):
""" Assign all of the variables with global vars """
with self.graph.as_default():
all_vars = tf.trainable_variables()
for variable, value in zip(all_vars, global_vars):
variable.load(value, self.sess)
print("已运行")
while if_to_start == 0:
time.sleep(1)
f = open(symbol,"w",encoding="UTF-8",newline="")
csvwriter=csv.writer(f)
csvwriter.writerow(["epoch","acc","loss","local_time","time_slot","operation"])
f.close()
Client = Clients(input_shape, num_classes, learning_rate, client_dataset_train_a, client_dataset_train_b, client_dataset_test_a, client_dataset_test_b)
while if_to_stop==0:
while next_epoch_nowepoch==0 or next_epoch_var==0:
print(".",end = "")
if if_to_stop==1:
break
time.sleep(1)
if if_to_stop==1:
break
next_epoch_nowepoch=0
next_epoch_var=0
if global_vars is not None:
Client.set_global_vars_class(global_vars)
print("开始训练")
train_start_time=time.time()
for ii_client_epoch in tqdm(range(0,client_epoch), desc='Client_Epoch'):
Client.train_epoch(batch_size,drop_rate)
train_end_time=time.time()
print("当前轮训练完毕,回传数据到服务器,epoch="+str(now_epoch))
acc, loss = Client.run_test(test_batchsize)
print("epoch="+str(now_epoch)+" acc="+str(acc)+" loss="+str(loss))
client_vars = Client.get_client_vars_class()
test_client(s, now_epoch, acc, loss, train_end_time-train_start_time)
get_client_vars(s, now_epoch)
print("等待下一轮训练指令",end = "")
f = open(symbol,"a",encoding="UTF-8",newline="")
csvwriter=csv.writer(f)
csvwriter.writerow([now_epoch, acc, loss, time.time(), train_end_time-train_start_time, "test_message"])
f.close()
print("")
print("全部训练完成!")
print("等待最终模型分发!")
while next_epoch_var==0:
print(".",end="")
time.sleep(1)
if global_vars is not None:
Client.set_global_vars_class(global_vars)
final_acc, final_loss=Client.run_test(Client.dataset_test_size)
print("Final!!! "+" acc="+str(final_acc)+" loss="+str(final_loss))
test_client(s, -1, final_acc, final_loss,-1)
f = open(symbol,"a",encoding="UTF-8",newline="")
csvwriter=csv.writer(f)
csvwriter.writerow([-1, final_acc, final_loss,time.time(),"","final_test"])
f.close()
tcp_tunnel_s.close()
print("程序执行完毕")