-
Notifications
You must be signed in to change notification settings - Fork 0
/
extractor.py
462 lines (397 loc) · 14.4 KB
/
extractor.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
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
import numpy as np
import math
import os
import argparse
import logging
import configparser
import const
import glob
import multiprocessing as mp
import pandas as pd
import pickle
logger = logging.getLogger('kf')
def init_logger():
logger.setLevel(logging.DEBUG)
# create console handler and set level to debug
ch = logging.StreamHandler()
# create formatter
formatter = logging.Formatter(const.LOG_FORMAT)
# add formatter to ch
ch.setFormatter(formatter)
# add ch to logger
logger.addHandler(ch)
return logger
### Parameters ###
# -1 is IN, 1 is OUT
#file format: "direction time size"
"""Feeder functions"""
def neighborhood(iterable):
iterator = iter(iterable)
prev = (0)
item = iterator.__next__() # throws StopIteration if empty.
for next in iterator:
yield (prev,item,next)
prev = item
item = next
yield (prev,item,None)
def chunkIt(seq, num):
avg = len(seq) / float(num)
out = []
last = 0.0
while last < len(seq):
out.append(seq[int(last):int(last + avg)])
last += avg
return out
"""Non-feeder functions"""
def get_pkt_list(trace_data):
first_line = trace_data[0]
first_line = first_line.split()
first_time = float(first_line[0])
dta = []
for line in trace_data:
a = line
b = a.split()
if float(b[1]) > 0:
#dta.append(((float(b[0])- first_time), abs(int(b[2])), 1))
dta.append(((float(b[0])- first_time), 1))
else:
#dta.append(((float(b[1]) - first_time), abs(int(b[2])), -1))
dta.append(((float(b[0]) - first_time), -1))
return dta
def In_Out(list_data):
In = []
Out = []
for p in list_data:
if p[1] == -1:
In.append(p)
if p[1] == 1:
Out.append(p)
return In, Out
############### TIME FEATURES #####################
def inter_pkt_time(list_data):
times = [x[0] for x in list_data]
temp = []
for elem,next_elem in zip(times, times[1:]+[times[0]]):
temp.append(next_elem-elem)
return temp[:-1]
def interarrival_times(list_data):
In, Out = In_Out(list_data)
IN = inter_pkt_time(In)
OUT = inter_pkt_time(Out)
TOTAL = inter_pkt_time(list_data)
return IN, OUT, TOTAL
def interarrival_maxminmeansd_stats(list_data):
interstats = []
In, Out, Total = interarrival_times(list_data)
if In and Out:
avg_in = sum(In)/float(len(In))
avg_out = sum(Out)/float(len(Out))
avg_total = sum(Total)/float(len(Total))
interstats.append((max(In), max(Out), max(Total), avg_in, avg_out, avg_total, np.std(In), np.std(Out), np.std(Total), np.percentile(In, 75), np.percentile(Out, 75), np.percentile(Total, 75)))
elif Out and not In:
avg_out = sum(Out)/float(len(Out))
avg_total = sum(Total)/float(len(Total))
interstats.append((0, max(Out), max(Total), 0, avg_out, avg_total, 0, np.std(Out), np.std(Total), 0, np.percentile(Out, 75), np.percentile(Total, 75)))
elif In and not Out:
avg_in = sum(In)/float(len(In))
avg_total = sum(Total)/float(len(Total))
interstats.append((max(In), 0, max(Total), avg_in, 0, avg_total, np.std(In), 0, np.std(Total), np.percentile(In, 75), 0, np.percentile(Total, 75)))
else:
interstats.extend(([0]*15))
return interstats
def time_percentile_stats(trace_data):
Total = get_pkt_list(trace_data)
In, Out = In_Out(Total)
In1 = [x[0] for x in In]
Out1 = [x[0] for x in Out]
Total1 = [x[0] for x in Total]
STATS = []
if In1:
STATS.append(np.percentile(In1, 25)) # return 25th percentile
STATS.append(np.percentile(In1, 50))
STATS.append(np.percentile(In1, 75))
STATS.append(np.percentile(In1, 100))
if not In1:
STATS.extend(([0]*4))
if Out1:
STATS.append(np.percentile(Out1, 25)) # return 25th percentile
STATS.append(np.percentile(Out1, 50))
STATS.append(np.percentile(Out1, 75))
STATS.append(np.percentile(Out1, 100))
if not Out1:
STATS.extend(([0]*4))
if Total1:
STATS.append(np.percentile(Total1, 25)) # return 25th percentile
STATS.append(np.percentile(Total1, 50))
STATS.append(np.percentile(Total1, 75))
STATS.append(np.percentile(Total1, 100))
if not Total1:
STATS.extend(([0]*4))
return STATS
def number_pkt_stats(trace_data):
Total = get_pkt_list(trace_data)
In, Out = In_Out(Total)
return len(In), len(Out), len(Total)
def first_and_last_30_pkts_stats(trace_data):
Total = get_pkt_list(trace_data)
first30 = Total[:30]
last30 = Total[-30:]
first30in = []
first30out = []
for p in first30:
if p[1] == -1:
first30in.append(p)
if p[1] == 1:
first30out.append(p)
last30in = []
last30out = []
for p in last30:
if p[1] == -1:
last30in.append(p)
if p[1] == 1:
last30out.append(p)
stats= []
stats.append(len(first30in))
stats.append(len(first30out))
stats.append(len(last30in))
stats.append(len(last30out))
return stats
#concentration of outgoing packets in chunks of 20 packets
def pkt_concentration_stats(trace_data):
Total = get_pkt_list(trace_data)
chunks= [Total[x:x+20] for x in range(0, len(Total), 20)]
concentrations = []
for item in chunks:
c = 0
for p in item:
if p[1] == 1:
c+=1
concentrations.append(c)
return np.std(concentrations), sum(concentrations)/float(len(concentrations)), np.percentile(concentrations, 50), min(concentrations), max(concentrations), concentrations
#Average number packets sent and received per second
def number_per_sec(trace_data):
Total = get_pkt_list(trace_data)
last_time = Total[-1][0]
last_second = math.ceil(last_time)
temp = []
l = []
for i in range(1, int(last_second)+1):
c = 0
for p in Total:
if p[0] <= i:
c+=1
temp.append(c)
for prev,item,next in neighborhood(temp):
x = item - prev
l.append(x)
avg_number_per_sec = sum(l)/float(len(l))
return avg_number_per_sec, np.std(l), np.percentile(l, 50), min(l), max(l), l
#Variant of packet ordering features
def avg_pkt_ordering_stats(trace_data):
Total = get_pkt_list(trace_data)
c1 = 0
c2 = 0
temp1 = []
temp2 = []
for p in Total:
if p[1] == 1:
temp1.append(c1)
c1+=1
if p[1] == -1:
temp2.append(c2)
c2+=1
avg_in = sum(temp1)/float(len(temp1))
avg_out = sum(temp2)/float(len(temp2))
return avg_in, avg_out, np.std(temp1), np.std(temp2)
def perc_inc_out(trace_data):
Total = get_pkt_list(trace_data)
In, Out = In_Out(Total)
percentage_in = len(In)/float(len(Total))
percentage_out = len(Out)/float(len(Total))
return percentage_in, percentage_out
############### SIZE FEATURES #####################
#def total_size(list_data):
# return sum([x[1] for x in list_data])
#def in_out_size(list_data):
# In, Out = In_Out(list_data)
# size_in = sum([x[1] for x in In])
# size_out = sum([x[1] for x in Out])
# return size_in, size_out
#def average_total_pkt_size(list_data):
# return np.mean([x[1] for x in list_data])
#def average_in_out_pkt_size(list_data):
# In, Out = In_Out(list_data)
# average_size_in = np.mean([x[1] for x in In])
# average_size_out = np.mean([x[1] for x in Out])
# return average_size_in, average_size_out
#def variance_total_pkt_size(list_data):
# return np.var([x[1] for x in list_data])
#def variance_in_out_pkt_size(list_data):
# In, Out = In_Out(list_data)
# var_size_in = np.var([x[1] for x in In])
# var_size_out = np.var([x[1] for x in Out])
# return var_size_in, var_size_out
#def std_total_pkt_size(list_data):
# return np.std([x[1] for x in list_data])
#def std_in_out_pkt_size(list_data):
# In, Out = In_Out(list_data)
# std_size_in = np.std([x[1] for x in In])
# std_size_out = np.std([x[1] for x in Out])
# return std_size_in, std_size_out
#def max_in_out_pkt_size(list_data):
# In, Out = In_Out(list_data)
# max_size_in = max([x[1] for x in In])
# max_size_out = max([x[1] for x in Out])
# return max_size_in, max_size_out
#def unique_pkt_lengths(list_data):
# pass
############### FEATURE FUNCTION #####################
#If size information available add them in to function below
def TOTAL_FEATURES(trace_data, max_size=175):
list_data = get_pkt_list(trace_data)
ALL_FEATURES = []
# ------TIME--------
intertimestats = [x for x in interarrival_maxminmeansd_stats(list_data)[0]]
timestats = time_percentile_stats(trace_data)
number_pkts = list(number_pkt_stats(trace_data))
thirtypkts = first_and_last_30_pkts_stats(trace_data)
stdconc, avgconc, medconc, minconc, maxconc, conc = pkt_concentration_stats(trace_data)
avg_per_sec, std_per_sec, med_per_sec, min_per_sec, max_per_sec, per_sec = number_per_sec(trace_data)
avg_order_in, avg_order_out, std_order_in, std_order_out = avg_pkt_ordering_stats(trace_data)
perc_in, perc_out = perc_inc_out(trace_data)
altconc = []
alt_per_sec = []
altconc = [sum(x) for x in chunkIt(conc, 70)]
alt_per_sec = [sum(x) for x in chunkIt(per_sec, 20)]
if len(altconc) == 70:
altconc.append(0)
if len(alt_per_sec) == 20:
alt_per_sec.append(0)
# ------SIZE--------
#tot_size = total_size(list_data)
#in_size, out_size = in_out_size(list_data)
#avg_total_size = average_total_pkt_size(list_data)
#avg_size_in, avg_size_out = average_in_out_pkt_size(list_data)
#var_total_size = variance_total_pkt_size(list_data)
#var_size_in, var_size_out = variance_in_out_pkt_size(list_data)
#std_total_size = std_total_pkt_size(list_data)
#std_size_in, std_size_out = std_in_out_pkt_size(list_data)
#max_size_in, max_size_out = max_in_out_pkt_size(list_data)
# TIME Features
ALL_FEATURES.extend(intertimestats)
ALL_FEATURES.extend(timestats)
ALL_FEATURES.extend(number_pkts)
ALL_FEATURES.extend(thirtypkts)
ALL_FEATURES.append(stdconc)
ALL_FEATURES.append(avgconc)
ALL_FEATURES.append(avg_per_sec)
ALL_FEATURES.append(std_per_sec)
ALL_FEATURES.append(avg_order_in)
ALL_FEATURES.append(avg_order_out)
ALL_FEATURES.append(std_order_in)
ALL_FEATURES.append(std_order_out)
ALL_FEATURES.append(medconc)
ALL_FEATURES.append(med_per_sec)
ALL_FEATURES.append(min_per_sec)
ALL_FEATURES.append(max_per_sec)
ALL_FEATURES.append(maxconc)
ALL_FEATURES.append(perc_in)
ALL_FEATURES.append(perc_out)
ALL_FEATURES.extend(altconc)
ALL_FEATURES.extend(alt_per_sec)
ALL_FEATURES.append(sum(altconc))
ALL_FEATURES.append(sum(alt_per_sec))
ALL_FEATURES.append(sum(intertimestats))
ALL_FEATURES.append(sum(timestats))
ALL_FEATURES.append(sum(number_pkts))
#SIZE FEATURES
#ALL_FEATURES.append(tot_size)
#ALL_FEATURES.append(in_size)
#ALL_FEATURES.append(out_size)
#ALL_FEATURES.append(avg_total_size)
#ALL_FEATURES.append(avg_size_in)
#ALL_FEATURES.append(avg_size_out)
#ALL_FEATURES.append(var_total_size)
#ALL_FEATURES.append(var_size_in)
#ALL_FEATURES.append(var_size_out)
#ALL_FEATURES.append(std_total_size)
#ALL_FEATURES.append(std_size_in)
#ALL_FEATURES.append(std_size_out)
#ALL_FEATURES.append(max_size_in)
#ALL_FEATURES.append(max_size_out)
# This is optional, since all other features are of equal size this gives the first n features
# of this particular feature subset, some may be padded with 0's if too short.
ALL_FEATURES.extend(conc)
ALL_FEATURES.extend(per_sec)
while len(ALL_FEATURES) < max_size:
ALL_FEATURES.append(0)
features = ALL_FEATURES[:max_size]
return features
def read_conf(file):
cf = configparser.ConfigParser()
cf.read(file)
return dict(cf['default'])
def parallel(flist,n_jobs = 20):
pool = mp.Pool(n_jobs)
data_dict = pool.map(extractfeature, flist)
return data_dict
def extractfeature(f):
global MON_SITE_NUM
fname = f.split('/')[-1].split(".")[0]
# logger.info('Processing %s...'%fname)
try:
with open(f,'r') as f:
tcp_dump = f.readlines()
if len(tcp_dump) < 50:
return None
feature = TOTAL_FEATURES(tcp_dump)
if '-' in fname:
label = fname.split('-')
label = (int(label[0]), int(label[1]))
else:
label = (MON_SITE_NUM, int(fname))
return (feature,label)
except:
return None
if __name__== '__main__':
global MON_SITE_NUM
'''initialize logger'''
logger = init_logger()
'''read config file'''
cf = read_conf(const.confdir)
MON_SITE_NUM = int(cf['monitored_site_num'])
MON_INST_NUM = int(cf['monitored_inst_num'])
if cf['open_world'] == '1':
UNMON_SITE_NUM = int(cf['unmonitored_site_num'])
else:
UNMON_SITE_NUM = 0
print(MON_SITE_NUM)
print(MON_INST_NUM)
print(UNMON_SITE_NUM)
'''read in arg'''
parser = argparse.ArgumentParser(description='k-FP feature extraction')
parser.add_argument('traces_path',
metavar='<traces path>',
help='Path to the directory with the traffic traces to be simulated.')
parser.add_argument('-format',
metavar='<file suffix>',
default = "", )
args = parser.parse_args()
data_dict = {'feature':[],'label':[]}
flist = []
for i in range(MON_SITE_NUM):
for j in range(MON_INST_NUM):
print(os.path.join(args.traces_path, str(i) + "-" + str(j)+ args.format))
if os.path.exists( os.path.join(args.traces_path, str(i) + "-" + str(j)+ args.format) ):
flist.append(os.path.join(args.traces_path, str(i) + "-" + str(j)+ args.format) )
for i in range(UNMON_SITE_NUM):
if os.path.exists( os.path.join(args.traces_path, str(i)+ args.format) ):
flist.append( os.path.join(args.traces_path, str(i)+ args.format) )
print(len(flist))
res = parallel(flist,n_jobs = 20)
raw_data_dict = [i for i in res if i]
print("After removal: {}".format(len(raw_data_dict)))
data_dict['feature'], data_dict['label'] = zip(*raw_data_dict)
outputdir = const.outputdir+args.traces_path.rstrip('/').split('/')[-1]
pickle.dump(data_dict, open(outputdir, 'wb'))
# logger.info('Save to %s.npy'%outputdir)