-
Notifications
You must be signed in to change notification settings - Fork 7
/
Copy pathnewSimulate.py
979 lines (918 loc) · 49.8 KB
/
newSimulate.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
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
# Proprietary Design
#import robot_ghliu
from newDDPG import actor
from newDDPG import DDPG
from newENV import BS
from newENV import plot_UE_BS_distribution_Cache
# Public Lib
from torch.autograd import Variable
import torch
from torchviz import make_dot
from torch.utils.tensorboard import SummaryWriter
#writer = SummaryWriter('runs/fashion_mnist_experiment_1')
import numpy as np
import collections
import time,copy,os,csv,random,pickle
import matplotlib
import matplotlib.pyplot as plt
import matplotlib as mpl
mpl.rcParams['agg.path.chunksize'] = 90000
from numpy.random import randn
#from random import randint
from tqdm import tqdm
from datetime import date
from datetime import datetime
today = date.today()
os.environ['CUDA_VISIBLE_DEVICES']='1'
import concurrent.futures
import multiprocessing
num_cores = multiprocessing.cpu_count()
##################### hyper parameters ####################
# Simulation Parameter
LOAD_EVN = True
RESET_CHANNEL = True
REQUEST_DUPLICATE = False
MAX_EPISODES=500
MAX_EP_STEPS=100
warmup = -1
epsilon = 0.2
#####################################
# plot style
'''
plt.rcParams.update({
"text.usetex": True,
"font.family": "sans-serif",
"font.sans-serif": ["Helvetica"]})
'''
font = {'family' : 'Verdana',
'weight' : 'normal',
'size' : 12}
matplotlib.rc('font', **font)
markerSize = 20*4**1
linestyles = [':', '-', '--', '-.']
#####################################
def plotMetric(poolEE,poolBestEE):
xScale = 100
x = range( len(poolEE[-xScale:]) )
plt.cla()
plt.plot(x,poolEE[-xScale:],'bo-',label='EE RL')
plt.plot(x,poolBestEE[-xScale:],'r^-',label='EE BF')
plt.title("Metric Visualization") # title
plt.ylabel("Bits/J") # y label
plt.xlabel("t") # x label
plt.xlim([0, xScale])
#plt.ylim([0, 40])
plt.grid()
plt.legend()
#handles, labels = plt.gca().get_legend_handles_labels()
#by_label = dict(zip(labels, handles))
#plt.legend(by_label.values(), by_label.keys())
fig = plt.gcf()
fig.show()
fig.canvas.draw()
plt.pause(0.001)
def plotHistory(env,filename,isEPS=False,loadBF=False):
# Initialization
poolEE_BF = None
poolTP_BF = None
poolPsys_BF = None
poolHR_BF = None
poolMCAP_BF = None
poolMCCPU_BF = None
poolEE_RL1act=None
poolTP_RL1act=None
poolPsys_RL1act=None
poolHR_RL1act=None
poolMCAP_RL1act=None
poolMCCPU_RL1act=None
poolLossActor1act=None
poolLossCritic1act=None
poolEE_RL2act=None
poolTP_RL2act=None
poolPsys_RL2act=None
poolHR_RL2act=None
poolMCAP_RL2act=None
poolMCCPU_RL2act=None
poolLossActor2act=None
poolLossCritic2act=None
#---------------------------------------------------------------------------------------------
# Load Brute Force Policy
if env.B==4 and env.U ==4 and env.F==5 and env.N==2 and loadBF:
with open(filename+ 'BF.pkl', 'rb') as f:
env, poolEE_BF,poolTP_BF,poolPsys_BF,poolHR_BF,poolMCAP_BF,poolMCCPU_BF,poolCL_BF,poolCA_BF = pickle.load(f)
#---------------------------------------------------------------------------------------------
# Load DDPG
file_1act = filename+ '1act' +'RL.pkl'
file_2act = filename+ '2act' +'RL.pkl'
if os.path.isfile(file_1act):
with open(file_1act,'rb') as f:
env, poolEE_RL1act,poolTP_RL1act,poolPsys_RL1act,poolHR_RL1act,poolMCAP_RL1act,poolMCCPU_RL1act,poolCL_RL,poolCA_RL\
,poolLossActor1act,poolLossCritic1act = pickle.load(f)
if os.path.isfile(file_2act):
with open(file_2act,'rb') as f:
env, poolEE_RL2act,poolTP_RL2act,poolPsys_RL2act,poolHR_RL2act,poolMCAP_RL2act,poolMCCPU_RL2act,poolCL_RL,poolCA_RL\
,poolLossActor2act,poolLossCritic2act = pickle.load(f)
# Load Benchmarks
with open(filename+'BM1.pkl','rb') as f:
env, poolEE_BM1,poolTP_BM1,poolPsys_BM1,poolHR_BM1,poolMCAP_BM1,poolMCCPU_BM1,poolCL_BM1,poolCA_BM1 = pickle.load(f)
with open(filename+'BM2.pkl','rb') as f:
env, poolEE_BM2,poolTP_BM2,poolPsys_BM2,poolHR_BM2,poolMCAP_BM2,poolMCCPU_BM2,poolCL_BM2,poolCA_BM2 = pickle.load(f)
with open(filename+'BM3.pkl','rb') as f:
env, poolEE_BM3,poolTP_BM3,poolPsys_BM3,poolHR_BM3,poolMCAP_BM3,poolMCCPU_BM3,poolCL_BM3,poolCA_BM3 = pickle.load(f)
#---------------------------------------------------------------------------------------------
# Plot EE/HR/TP/Psys/MCAP/MCCPU
#plotBrokenAxis(env,filename,topic='EE',xLabel='t',yLabel='Bits/J',yScale='linear',\
# line_BF=poolEE_BF,line_RL1act=poolEE_RL1act,line_RL2act=poolEE_RL2act,line_BM1=poolEE_BM1,line_BM2=poolEE_BM2,line_BM3=poolEE_BM3,isEPS=isEPS)
plotTopic(env,filename,topic='EE',xLabel='t',yLabel='Bits/J',yScale='log',\
line_BF=poolEE_BF,line_RL1act=poolEE_RL1act,line_RL2act=poolEE_RL2act,line_BM1=poolEE_BM1,line_BM2=poolEE_BM2,line_BM3=poolEE_BM3,isEPS=isEPS)
plotTopic(env,filename,topic='HR',xLabel='t',yLabel='Ratio',yScale='linear',\
line_BF=poolHR_BF,line_RL1act=poolHR_RL1act,line_RL2act=poolHR_RL2act,line_BM1=poolHR_BM1,line_BM2=poolHR_BM2,line_BM3=poolHR_BM3,isEPS=isEPS)
#plotBrokenAxis(env,filename,topic='TP',xLabel='t',yLabel='Bits/s',yScale='linear',\
# line_BF=poolTP_BF,line_RL1act=poolTP_RL1act,line_RL2act=poolTP_RL2act,line_BM1=poolTP_BM1,line_BM2=poolTP_BM2,line_BM3=poolTP_BM3,isEPS=isEPS)
plotTopic(env,filename,topic='TP',xLabel='t',yLabel='Bits/s',yScale='linear',\
line_BF=poolTP_BF,line_RL1act=poolTP_RL1act,line_RL2act=poolTP_RL2act,line_BM1=poolTP_BM1,line_BM2=poolTP_BM2,line_BM3=poolTP_BM3,isEPS=isEPS)
#plotBrokenAxis(env,filename,topic='Psys',xLabel='t',yLabel='W',yScale='linear',\
# line_BF=poolPsys_BF,line_RL1act=poolPsys_RL1act,line_RL2act=poolPsys_RL2act,line_BM1=poolPsys_BM1,line_BM2=poolPsys_BM2,line_BM3=poolPsys_BM3,isEPS=isEPS)
plotTopic(env,filename,topic='Psys',xLabel='t',yLabel='W',yScale='log',\
line_BF=poolPsys_BF,line_RL1act=poolPsys_RL1act,line_RL2act=poolPsys_RL2act,line_BM1=poolPsys_BM1,line_BM2=poolPsys_BM2,line_BM3=poolPsys_BM3,isEPS=isEPS)
plotTopic(env,filename,topic='MCAP',xLabel='t',yLabel='Counts',yScale='linear',\
line_BF=poolMCAP_BF,line_RL1act=poolMCAP_RL1act,line_RL2act=poolMCAP_RL2act,line_BM1=poolMCAP_BM1,line_BM2=poolMCAP_BM2,line_BM3=poolMCAP_BM3,isEPS=isEPS)
plotTopic(env,filename,topic='MCCPU',xLabel='t',yLabel='Counts',yScale='linear',\
line_BF=poolMCCPU_BF,line_RL1act=poolMCCPU_RL1act,line_RL2act=poolMCCPU_RL2act,line_BM1=poolMCCPU_BM1,line_BM2=poolMCCPU_BM2,line_BM3=poolMCCPU_BM3,isEPS=isEPS)
#---------------------------------------------------------------------------------------------
if poolLossActor1act != None or poolLossActor2act != None:
# plot RL: poolLossCritic/poolLossActor
plt.clf()
if os.path.isfile(file_1act):
plt.plot(range(len(poolLossCritic1act)),poolLossCritic1act,'b-',label='Loss of critic 1act')
plt.plot(range(len(poolLossActor1act)),poolLossActor1act,'c-',label='Loss of actor 1act')
if os.path.isfile(file_2act):
plt.plot(range(len(poolLossCritic2act)),poolLossCritic2act,'r-',label='Loss of critic 2act')
plt.plot(range(len(poolLossActor2act)),poolLossActor2act,'m-',label='Loss of actor 2act')
plt.grid()
plt.legend()
#plt.tight_layout()
plt.title('Critic and Actor Loss\n' + env.TopologyName) # title
plt.ylabel("Q") # y label
plt.xlabel("t") # x label
plt.savefig(filename + 'Loss.png', format='png')
'''
fig = plt.gcf()
fig.savefig(filename + 'Loss.png', format='png',dpi=120)
if isEPS:
fig.savefig(filename + 'Loss.eps', format='eps',dpi=120)
'''
#---------------------------------------------------------------------------------------------
def plotTopic(env,filename,topic,xLabel,yLabel,yScale,line_BF=None,line_RL1act=None,line_RL2act=None,line_BM1=None,line_BM2=None,line_BM3=None,isEPS=False):
plt.clf()
nXpt=len(line_BM1[0])
#---------------------------------------------------------------------------------------------
# plot Brute Force
if line_BF!=None:
plt.plot(range(nXpt),line_BF,'k:',label='Brute Force',linewidth=4)
finalValue = "{:.2f}".format(line_BF[-1])
#plt.annotate(finalValue, (nXpt,line_BF[-1]),textcoords="offset points",xytext=(20,10),ha='center',color='k')
#---------------------------------------------------------------------------------------------
# plot DDPG 1act
if line_RL1act != None:
plt.plot(range(nXpt),line_RL1act,'b-',label='Proposed',linewidth=3)
finalValue = "{:.2f}".format(line_RL1act[-1])
#plt.annotate(finalValue, (nXpt,line_RL1act[-1]),textcoords="offset points",xytext=(20,-10),ha='center',color='b')
#---------------------------------------------------------------------------------------------
# plot DDPG 2act
if line_RL2act != None:
plt.plot(range(nXpt),line_RL2act,'r-',label='Proposed 2act',linewidth=3)
finalValue = "{:.2f}".format(line_RL2act[-1])
#plt.annotate(finalValue, (nXpt,line_RL2act[-1]),textcoords="offset points",xytext=(20,-10),ha='center',color='r')
#---------------------------------------------------------------------------------------------
if line_BM1!=None and line_BM2!=None and line_BM3!=None:
for l in [1,env.L]:
#for l in range(1,env.L+1):
# plot BM1
#plt.plot(range(nXpt),line_BM1[l],color='g',linestyle=linestyles[l],label='BM1('+r'$l=$'+str(l)+')')
plt.plot(range(nXpt),line_BM1[l],color='g',linestyle=linestyles[l],label='BM1(l='+str(l)+')',linewidth=2)
finalValue = "{:.2f}".format(line_BM1[l][-1])
#plt.annotate(finalValue, (nXpt,line_BM1[-1]),textcoords="offset points",xytext=(20,10),ha='center',color='g')
# plot BM2
plt.plot(range(nXpt),line_BM2[l],color='y',linestyle=linestyles[l],label='BM2(l='+str(l)+')',linewidth=1.2)
finalValue = "{:.2f}".format(line_BM2[l][-1])
#plt.annotate(finalValue, (nXpt,line_BM2[-1]),textcoords="offset points",xytext=(20,10),ha='center',color='y')
# plot BM3
plt.plot(range(nXpt),line_BM3[l],color='c',linestyle=linestyles[l],label='BM3(l='+str(l)+')',linewidth=1)
finalValue = "{:.2f}".format(line_BM3[l][-1])
#plt.annotate(finalValue, (nXpt,line_BM3[-1]),textcoords="offset points",xytext=(20,10),ha='center',color='y')
#---------------------------------------------------------------------------------------------
'''
if 'Training' in filename:
phaseName = 'Training Phase'
elif 'Evaluation' in filename:
phaseName = 'Evaluation Phase'
elif 'Preview' in filename:
phaseName = 'Preview Phase'
plt.title(phaseName+': Energy Efficiency (EE)\n Topology:'+env.TopologyCode) # title
'''
plt.grid()
plt.legend()
plt.xlim(0,nXpt-1)
#plt.tight_layout()
plt.autoscale()
plt.xlabel(xLabel,fontsize=12) # x label
#plt.xlabel(r'x') # x label
plt.ylabel(yLabel,fontsize=12,loc='top') # y label
plt.yscale(yScale)
plt.legend(loc = 'lower left', fontsize=10)
plt.savefig(filename + topic +'.png', format='png',dpi=600,bbox_inches='tight')
if isEPS:
plt.savefig(filename + topic +'.eps', format='eps',dpi=600,bbox_inches='tight')
def plotBrokenAxis(env,filename,topic,xLabel,yLabel,yScale,line_BF=None,line_RL1act=None,line_RL2act=None,line_BM1=None,line_BM2=None,line_BM3=None,isEPS=False):
plt.clf()
nXpt=len(line_BM1[0])
f, (ax, ax2) = plt.subplots(2, 1, sharex=True)
# plot the same data on both axes
# plot Brute Force
if line_BF!=None:
ax.plot(range(nXpt),line_BF,'k:',label='Brute Force',linewidth=4)
ax2.plot(range(nXpt),line_BF,'k:',label='Brute Force',linewidth=4)
#---------------------------------------------------------------------------------------------
# plot DDPG 1act
if line_RL1act != None:
ax.plot(range(nXpt),line_RL1act,'b-',label='Proposed',linewidth=3)
ax2.plot(range(nXpt),line_RL1act,'b-',label='Proposed',linewidth=3)
#---------------------------------------------------------------------------------------------
# plot DDPG 2act
if line_RL2act != None:
ax.plot(range(nXpt),line_RL2act,'r-',label='Proposed 2act',linewidth=3)
ax2.plot(range(nXpt),line_RL2act,'r-',label='Proposed 2act',linewidth=3)
#---------------------------------------------------------------------------------------------
if line_BM1!=None and line_BM2!=None and line_BM3!=None:
for l in [1,env.L]:
#for l in range(1,env.L+1):
# plot BM1
#plt.plot(range(nXpt),line_BM1[l],color='g',linestyle=linestyles[l],label='BM1('+r'$l=$'+str(l)+')')
ax.plot(range(nXpt),line_BM1[l],color='g',linestyle=linestyles[l],label='BM1(l='+str(l)+')',linewidth=2)
ax2.plot(range(nXpt),line_BM1[l],color='g',linestyle=linestyles[l],label='BM1(l='+str(l)+')',linewidth=2)
# plot BM2
ax.plot(range(nXpt),line_BM2[l],color='y',linestyle=linestyles[l],label='BM2(l='+str(l)+')')
ax2.plot(range(nXpt),line_BM2[l],color='y',linestyle=linestyles[l],label='BM2(l='+str(l)+')')
# plot BM3
ax.plot(range(nXpt),line_BM3[l],color='c',linestyle=linestyles[l],label='BM3(l='+str(l)+')')
ax2.plot(range(nXpt),line_BM3[l],color='c',linestyle=linestyles[l],label='BM3(l='+str(l)+')')
# zoom-in / limit the view to different portions of the data
# Scenario1: EE
#ax.set_ylim(195, 215)
#ax2.set_ylim(0, 5)
# Scenario1: TP
ax.set_ylim(5.8, 6.1)
ax2.set_ylim(0, 4.5)
# Scenario1: Psys
#ax.set_ylim(1, 1.55)
#ax2.set_ylim(0.01, 0.03)
# hide the spines between ax and ax2
ax.spines['bottom'].set_visible(False)
ax2.spines['top'].set_visible(False)
ax.xaxis.tick_top()
ax.tick_params(labeltop=False) # don't put tick labels at the top
ax2.xaxis.tick_bottom()
# This looks pretty good, and was fairly painless, but you can get that
# cut-out diagonal lines look with just a bit more work. The important
# thing to know here is that in axes coordinates, which are always
# between 0-1, spine endpoints are at these locations (0,0), (0,1),
# (1,0), and (1,1). Thus, we just need to put the diagonals in the
# appropriate corners of each of our axes, and so long as we use the
# right transform and disable clipping.
d = .015 # how big to make the diagonal lines in axes coordinates
# arguments to pass to plot, just so we don't keep repeating them
kwargs = dict(transform=ax.transAxes, color='k', clip_on=False)
ax.plot(( -d, +d), (-d, +d), **kwargs) # top-left diagonal
ax.plot((1 - d, 1 + d), (-d, +d), **kwargs) # top-right diagonal
kwargs.update(transform=ax2.transAxes) # switch to the bottom axes
ax2.plot((-d, +d), (1 - d, 1 + d), **kwargs) # bottom-left diagonal
ax2.plot((1 - d, 1 + d), (1 - d, 1 + d), **kwargs) # bottom-right diagonal
#plt.show()
ax.grid()
ax2.grid()
#plt.xlim(0,nXpt-1)
#plt.tight_layout()
#plt.autoscale()
plt.xlabel(xLabel,fontsize=12) # x label
#plt.ylabel(yLabel,fontsize=12) # y label
ax.set_ylabel(yLabel, loc='top',fontsize=12)
plt.yscale(yScale)
plt.legend(loc = 'lower left', fontsize=10)
plt.savefig(filename + topic +'.png', format='png',dpi=120,bbox_inches='tight')
if isEPS:
plt.savefig(filename + topic +'.eps', format='eps',dpi=120,bbox_inches='tight')
print('try broken axis')
def trainModel(env,actMode,changeReq,changeChannel,loadActor,randSEED=0):
# new ACT
#modelPath = 'D:\\/Model/' + env.TopologyName+'/'
modelPath = 'data/'+env.TopologyCode+'/Model/'
if actMode == '2act':
ddpg_cl = DDPG(obs_dim = env.dimObs, act_dim = env.dimActCL,memMaxSize=25000)
ddpg_ca = DDPG(obs_dim = env.dimObs, act_dim = env.dimActCA,memMaxSize=25000)
if(loadActor):
ddpg_cl.loadModel(modelPath = modelPath, modelName= '['+ str(randSEED) +']'+ actMode+'_cl')
ddpg_ca.loadModel(modelPath = modelPath, modelName= '['+ str(randSEED) +']'+ actMode+'_ca')
elif actMode == '1act':
ddpg_s = DDPG(obs_dim = env.dimObs, act_dim = env.dimAct,memMaxSize=25000)###
if(loadActor):
ddpg_s.loadModel(modelPath = modelPath, modelName= '['+ str(randSEED) +']' + actMode)
'''
print('Actor Network')
print(ddpg_s.actor)
print('Critic Network')
print(ddpg_s.critic)
'''
#---------------------------------------------------------------------------------------------
mu = 0
noiseSigma = 1 # control exploration
countChangeReq = 0
countChangeChannel = 0
nItr = MAX_EPISODES*MAX_EP_STEPS
# RL
poolEE_RL = [0]*nItr
poolTP_RL = [0]*nItr
poolPsys_RL = [0]*nItr
poolHR_RL = [0]*nItr
poolMCAP_RL = [0]*nItr # MCAP = miss file count at APs
poolMCCPU_RL= [0]*nItr # MCCPU = miss file count at CPU
poolCL_RL = [0]*nItr
poolCA_RL = [0]*nItr
poolLossActor = [0]*nItr
poolLossCritic= [0]*nItr
poolVarLossCritic = []
poolVarEE = []
# BM1 Initialization [SNR-based]
poolEE_BM1 = [[0]*nItr for i in range(env.L+1)]
poolTP_BM1 = [[0]*nItr for i in range(env.L+1)]
poolPsys_BM1=[[0]*nItr for i in range(env.L+1)]
poolHR_BM1 = [[0]*nItr for i in range(env.L+1)]
poolMCAP_BM1=[[0]*nItr for i in range(env.L+1)]
poolMCCPU_BM1=[[0]*nItr for i in range(env.L+1)]
poolCL_BM1 = [[0]*nItr for i in range(env.L+1)]
poolCA_BM1 = [[0]*nItr for i in range(env.L+1)]
# BM2 Initialization [SNR-based]
poolEE_BM2 = [[0]*nItr for i in range(env.L+1)]
poolTP_BM2 = [[0]*nItr for i in range(env.L+1)]
poolPsys_BM2=[[0]*nItr for i in range(env.L+1)]
poolHR_BM2 = [[0]*nItr for i in range(env.L+1)]
poolMCAP_BM2=[[0]*nItr for i in range(env.L+1)]
poolMCCPU_BM2=[[0]*nItr for i in range(env.L+1)]
poolCL_BM2 = [[0]*nItr for i in range(env.L+1)]
poolCA_BM2 = [[0]*nItr for i in range(env.L+1)]
# BM3 Initialization [File-based]
poolEE_BM3 = [[0]*nItr for i in range(env.L+1)]
poolTP_BM3 = [[0]*nItr for i in range(env.L+1)]
poolPsys_BM3=[[0]*nItr for i in range(env.L+1)]
poolHR_BM3 = [[0]*nItr for i in range(env.L+1)]
poolMCAP_BM3=[[0]*nItr for i in range(env.L+1)]
poolMCCPU_BM3=[[0]*nItr for i in range(env.L+1)]
poolCL_BM3 = [[0]*nItr for i in range(env.L+1)]
poolCA_BM3 = [[0]*nItr for i in range(env.L+1)]
for epsilon in tqdm(range(MAX_EPISODES),bar_format='{l_bar}{bar:10}{r_bar}{bar:-10b}'):
ep_reward = 0
obs = env.reset()# Get initial state
for step in range(MAX_EP_STEPS):
ep = epsilon*MAX_EP_STEPS+step
#Epsilon-Greedy Algorithm
# https://medium.com/analytics-vidhya/the-epsilon-greedy-algorithm-for-reinforcement-learning-5fe6f96dc870
'''
if np.random.rand() > epsilon:
noise = np.zeros(env.dimActCL)
a_cl = ddpg_cl.action(obs,noise)# choose action [ env.U*env.B x 1 ]
a_ca = opt_caching_state.flatten()
action = np.concatenate((a_cl, a_ca), axis=0)
else:
action = env.action_space.sample()
action[-env.dimActCA:] = opt_caching_state.flatten()
'''
if actMode == '2act':
noise = np.random.normal(mu, noiseSigma,size=env.dimActCL)
a_cl = ddpg_cl.action(obs,noise)# choose action [ env.U*env.B x 1 ]
#a_ca = opt_caching_state.flatten()
noise = np.random.normal(mu, noiseSigma,size=env.dimActCA)
a_ca = ddpg_ca.action(obs,noise)# choose action [ env.U*env.B x 1 ]
action = np.concatenate((a_cl, a_ca), axis=0)
elif actMode == '1act':
noise = np.random.normal(mu, noiseSigma,size=env.dimAct)
action = ddpg_s.action(obs,noise)
# take action to ENV
obs2, reward, done, info = env.step(action)
EE_RL = reward
TP_RL = sum(env.Throughput)
Psys_RL = env.P_sys/1000
HR_RL = env.HR
CL_Policy_UE_RL,CA_Policy_BS_RL = env.action2Policy(action)
poolEE_RL[ep] = EE_RL
poolTP_RL[ep] = TP_RL
poolPsys_RL[ep] = Psys_RL
poolHR_RL[ep] = HR_RL
poolMCAP_RL[ep] = env.missCounterAP
poolMCCPU_RL[ep]= env.missCounterCPU
poolCL_RL[ep] = CL_Policy_UE_RL
poolCA_RL[ep] = CA_Policy_BS_RL
#===========================================================================
# Experience Injection
#if ep < 1000:
# EE_BM1, SNR_CL_Policy_UE_BM1, POP_CA_Policy_BS_BM1,bestL = env.getBestEE_snrCL_popCA(cacheMode='pref')
# RL Add Memory
if actMode == '2act':
ddpg_cl.addMemory([obs,a_cl,reward,obs2])
ddpg_ca.addMemory([obs,a_ca,-env.P_sys,obs2])# remind the reward of CA agent
elif actMode == '1act':
ddpg_s.addMemory([obs,action,reward,obs2])
#===========================================================================
# RL Upadate
if actMode == '2act':
if len(ddpg_cl.memory) > ddpg_cl.batch_size:
lossActor, lossCritic = ddpg_cl.train()
poolLossActor[ep] = lossActor.item()
poolLossCritic[ep] = lossCritic.item()
lossActor, lossCritic = ddpg_ca.train()
elif actMode == '1act':
if len(ddpg_s.memory) > ddpg_s.batch_size:
lossActor, lossCritic = ddpg_s.train()
poolLossActor[ep] = lossActor.item()
poolLossCritic[ep] = lossCritic.item()
obs = obs2
ep_reward += reward
#---------------------------------------------------------------------------------------------
# iteration update
if ep % 30 == 0:
noiseSigma*=0.995
if (ep % 1000) == 0:
#print(list(poolLossCritic[-100:]))
poolVarLossCritic.append( np.var(poolLossCritic[-1000:]) )
poolVarEE.append( np.var(poolEE_RL[-1000:]) )
if (poolVarLossCritic[-1] < 10) and (poolVarEE[-1] < 10) and changeReq:
env.resetReq()
print('**Change Request: ',env.Req)
countChangeReq+=1
noiseSigma = 1 # reset explore
fixInterval=1000
if (ep % fixInterval) == 0 and ep>0: # Mectric Snapshot
for l in range(1,env.L+1):
# Benchmark 1
EE_BM1, CL_Policy_UE_BM1, CA_Policy_BS_BM1 = env.getPolicy_BM1(cacheMode='pref',nLink=l)
EE_BM1 = env.calEE(CL_Policy_UE_BM1,CA_Policy_BS_BM1)
TP_BM1 = sum(env.Throughput)
Psys_BM1 = env.P_sys/1000 # mW->W
HR_BM1 = env.calHR(CL_Policy_UE_BM1,CA_Policy_BS_BM1)
poolEE_BM1[l][ep:ep+fixInterval] =[EE_BM1]*fixInterval
poolTP_BM1[l][ep:ep+fixInterval] =[TP_BM1]*fixInterval
poolPsys_BM1[l][ep:ep+fixInterval] =[Psys_BM1]*fixInterval
poolHR_BM1[l][ep:ep+fixInterval] =[HR_BM1]*fixInterval
poolMCAP_BM1[l][ep:ep+fixInterval] =[env.missCounterAP]*fixInterval
poolMCCPU_BM1[l][ep:ep+fixInterval] =[env.missCounterCPU]*fixInterval
poolCL_BM1[l][ep:ep+fixInterval] =[CL_Policy_UE_BM1]*fixInterval
poolCA_BM1[l][ep:ep+fixInterval] =[CA_Policy_BS_BM1]*fixInterval
#print('EE_BM1'+'_L'+str(l),'=', EE_BM1)
# Benchmark 2
EE_BM2, CL_Policy_UE_BM2, CA_Policy_BS_BM2 = env.getPolicy_BM2(nLink=l)
EE_BM2 = env.calEE(CL_Policy_UE_BM2,CA_Policy_BS_BM2)
TP_BM2 = sum(env.Throughput)
Psys_BM2 = env.P_sys/1000 # mW->W
HR_BM2 = env.calHR(CL_Policy_UE_BM2,CA_Policy_BS_BM2)
poolEE_BM2[l][ep:ep+fixInterval] =[EE_BM2]*fixInterval
poolTP_BM2[l][ep:ep+fixInterval] =[TP_BM2]*fixInterval
poolPsys_BM2[l][ep:ep+fixInterval] =[Psys_BM2]*fixInterval
poolHR_BM2[l][ep:ep+fixInterval] =[HR_BM2]*fixInterval
poolMCAP_BM2[l][ep:ep+fixInterval] =[env.missCounterAP]*fixInterval
poolMCCPU_BM2[l][ep:ep+fixInterval] =[env.missCounterCPU]*fixInterval
poolCL_BM2[l][ep:ep+fixInterval] =[CL_Policy_UE_BM2]*fixInterval
poolCA_BM2[l][ep:ep+fixInterval] =[CA_Policy_BS_BM2]*fixInterval
#print('EE_BM2'+'_L'+str(l),'=', EE_BM2)
# Benchmark 3
EE_BM3, CL_Policy_UE_BM3, CA_Policy_BS_BM3 = env.getPolicy_BM3(nLink=l)
EE_BM3 = env.calEE(CL_Policy_UE_BM3,CA_Policy_BS_BM3)
TP_BM3 = sum(env.Throughput)
Psys_BM3 = env.P_sys/1000 # mW->W
HR_BM3 = env.calHR(CL_Policy_UE_BM3,CA_Policy_BS_BM3)
poolEE_BM3[l][ep:ep+fixInterval] =[EE_BM3]*fixInterval
poolTP_BM3[l][ep:ep+fixInterval] =[TP_BM3]*fixInterval
poolPsys_BM3[l][ep:ep+fixInterval] =[Psys_BM3]*fixInterval
poolHR_BM3[l][ep:ep+fixInterval] =[HR_BM3]*fixInterval
poolMCAP_BM3[l][ep:ep+fixInterval] =[env.missCounterAP]*fixInterval
poolMCCPU_BM3[l][ep:ep+fixInterval] =[env.missCounterCPU]*fixInterval
poolCL_BM3[l][ep:ep+fixInterval] =[CL_Policy_UE_BM3]*fixInterval
poolCA_BM3[l][ep:ep+fixInterval] =[CA_Policy_BS_BM3]*fixInterval
#print('EE_BM3'+'_L'+str(l),'=', EE_BM3)
# Preview / Save Line
filename = 'data/'+env.TopologyCode+'/Preview/'+'['+ str(randSEED) +']'+ env.TopologyName +str(MAX_EPISODES*MAX_EP_STEPS)+'_Train_'
with open(filename+ actMode +'RL.pkl', 'wb') as f:
pickle.dump([env, poolEE_RL,poolTP_RL,poolPsys_RL,poolHR_RL,poolMCAP_RL,poolMCCPU_RL,None,None,poolLossActor,poolLossCritic], f)
with open(filename+'BM1.pkl', 'wb') as f:
pickle.dump([env, poolEE_BM1,poolTP_BM1,poolPsys_BM1,poolHR_BM1,poolMCAP_BM1,poolMCCPU_BM1,None,None], f)
with open(filename+'BM2.pkl', 'wb') as f:
pickle.dump([env, poolEE_BM2,poolTP_BM2,poolPsys_BM2,poolHR_BM2,poolMCAP_BM2,poolMCCPU_BM2,None,None], f)
with open(filename+'BM3.pkl', 'wb') as f:
pickle.dump([env, poolEE_BM3,poolTP_BM3,poolPsys_BM3,poolHR_BM3,poolMCAP_BM3,poolMCCPU_BM3,None,None], f)
plotHistory(env,filename,isEPS=False,loadBF=False)
'''
if poolEE_RL[-1]>EE_BM1:
print('poolEE_RL win!',poolEE_RL[-1], 'EE_BM1 loss QQ', EE_BM1)
elif (ep % 30000) == 0:
noiseSigma = 1 # reset explore
'''
if changeChannel:
env.timeVariantChannel()
countChangeChannel+=1
#if ep_reward>28500:
# print('\nEpisode:{} Reward:{} Explore:{}'.format(ep,ep_reward,noiseSigma))
#---------------------------------------------------------------------------------------------
# Save Model
if actMode == '2act':
ddpg_cl.saveModel(modelPath = modelPath,modelName= '['+ str(randSEED) +']'+ actMode+'_cl')
ddpg_ca.saveModel(modelPath = modelPath,modelName= '['+ str(randSEED) +']'+ actMode+'_ca')
elif actMode == '1act':
ddpg_s.saveModel(modelPath = modelPath,modelName= '['+ str(randSEED) +']' + actMode)
# Save Line
filename = 'data/'+env.TopologyCode+'/TrainingPhase/'+'['+ str(randSEED) +']'+ env.TopologyName +str(MAX_EPISODES*MAX_EP_STEPS)+'_Train_'
with open(filename+ actMode +'RL.pkl', 'wb') as f:
pickle.dump([env, poolEE_RL,poolTP_RL,poolPsys_RL,poolHR_RL,poolMCAP_RL,poolMCCPU_RL,poolCL_RL,poolCA_RL,poolLossActor,poolLossCritic], f)
with open(filename+'BM1.pkl', 'wb') as f:
pickle.dump([env, poolEE_BM1,poolTP_BM1,poolPsys_BM1,poolHR_BM1,poolMCAP_BM1,poolMCCPU_BM1,poolCL_BM1,poolCA_BM1], f)
with open(filename+'BM2.pkl', 'wb') as f:
pickle.dump([env, poolEE_BM2,poolTP_BM2,poolPsys_BM2,poolHR_BM2,poolMCAP_BM2,poolMCCPU_BM2,poolCL_BM2,poolCA_BM2], f)
with open(filename+'BM3.pkl', 'wb') as f:
pickle.dump([env, poolEE_BM3,poolTP_BM3,poolPsys_BM3,poolHR_BM3,poolMCAP_BM3,poolMCCPU_BM3,poolCL_BM3,poolCA_BM3], f)
# clean preview
dir = 'data/'+env.TopologyCode+'/Preview/'
for f in os.listdir(dir):
os.remove(os.path.join(dir, f))
if min(poolEE_RL)>= max(poolEE_BM1[1]):
return True
else:
return False
def evaluateModel(env,actMode, nItr=100, randSEED=0,isBF=False,loadCT=False):
# load channel Trajectory
filename = 'data/'+env.TopologyCode+'/Topology/['+str(env.SEED)+']Topology_'+ env.TopologyName
with open(filename+'CT.pkl','rb') as f:
channelTrajectory = pickle.load(f)
# new ACT
modelPath = 'data/'+env.TopologyCode+'/Model/'
if actMode == '2act':
ddpg_cl = DDPG(obs_dim = env.dimObs, act_dim = env.dimActCL,memMaxSize=20000)
ddpg_ca = DDPG(obs_dim = env.dimObs, act_dim = env.dimActCA,memMaxSize=20000)
# load Model
ddpg_cl.loadModel(modelPath = modelPath, modelName= '['+ str(randSEED) +']' + actMode+'_cl')
ddpg_ca.loadModel(modelPath = modelPath, modelName= '['+ str(randSEED) +']' + actMode+'_ca')
elif actMode == '1act':
ddpg_s = DDPG(obs_dim = env.dimObs, act_dim = env.dimAct,memMaxSize=20000)###
# load Model
ddpg_s.loadModel(modelPath = modelPath, modelName= '['+ str(randSEED) +']'+ actMode)
# BF
if env.B==4 and env.U ==4 and env.F==5 and env.N==2:
poolEE_BF = [0]*nItr
poolTP_BF = [0]*nItr
poolPsys_BF = [0]*nItr
poolHR_BF = [0]*nItr
poolMCAP_BF = [0]*nItr # MCAP = miss file count at APs
poolMCCPU_BF= [0]*nItr # MCCPU = miss file count at CPU
poolCL_BF = [0]*nItr
poolCA_BF = [0]*nItr
# RL
poolEE_RL = [0]*nItr
poolTP_RL = [0]*nItr
poolPsys_RL = [0]*nItr
poolHR_RL = [0]*nItr
poolMCAP_RL = [0]*nItr # MCAP = miss file count at APs
poolMCCPU_RL= [0]*nItr # MCCPU = miss file count at CPU
poolCL_RL = [0]*nItr
poolCA_RL = [0]*nItr
poolLossActor = None
poolLossCritic = None
# BM1 Initialization [SNR-based]
poolEE_BM1 = [[0]*nItr for i in range(env.L+1)]
poolTP_BM1 = [[0]*nItr for i in range(env.L+1)]
poolPsys_BM1=[[0]*nItr for i in range(env.L+1)]
poolHR_BM1 = [[0]*nItr for i in range(env.L+1)]
poolMCAP_BM1=[[0]*nItr for i in range(env.L+1)]
poolMCCPU_BM1=[[0]*nItr for i in range(env.L+1)]
poolCL_BM1 = [[0]*nItr for i in range(env.L+1)]
poolCA_BM1 = [[0]*nItr for i in range(env.L+1)]
# BM2 Initialization [SNR-based]
poolEE_BM2 = [[0]*nItr for i in range(env.L+1)]
poolTP_BM2 = [[0]*nItr for i in range(env.L+1)]
poolPsys_BM2=[[0]*nItr for i in range(env.L+1)]
poolHR_BM2 = [[0]*nItr for i in range(env.L+1)]
poolMCAP_BM2=[[0]*nItr for i in range(env.L+1)]
poolMCCPU_BM2=[[0]*nItr for i in range(env.L+1)]
poolCL_BM2 = [[0]*nItr for i in range(env.L+1)]
poolCA_BM2 = [[0]*nItr for i in range(env.L+1)]
# BM3 Initialization [File-based]
poolEE_BM3 = [[0]*nItr for i in range(env.L+1)]
poolTP_BM3 = [[0]*nItr for i in range(env.L+1)]
poolPsys_BM3=[[0]*nItr for i in range(env.L+1)]
poolHR_BM3 = [[0]*nItr for i in range(env.L+1)]
poolMCAP_BM3=[[0]*nItr for i in range(env.L+1)]
poolMCCPU_BM3=[[0]*nItr for i in range(env.L+1)]
poolCL_BM3 = [[0]*nItr for i in range(env.L+1)]
poolCA_BM3 = [[0]*nItr for i in range(env.L+1)]
for ep in tqdm(range(nItr),bar_format='{l_bar}{bar:10}{r_bar}{bar:-10b}'):
# iterate each L
#===============================
for l in range(1,env.L+1):
# Benchmark 1
EE_BM1, CL_Policy_UE_BM1, CA_Policy_BS_BM1 = env.getPolicy_BM1(cacheMode='pref',nLink=l)
EE_BM1 = env.calEE(CL_Policy_UE_BM1,CA_Policy_BS_BM1)
TP_BM1 = sum(env.Throughput)
Psys_BM1 = env.P_sys/1000 # mW->W
HR_BM1 = env.calHR(CL_Policy_UE_BM1,CA_Policy_BS_BM1)
poolEE_BM1[l][ep] =EE_BM1
poolTP_BM1[l][ep] =TP_BM1
poolPsys_BM1[l][ep] =Psys_BM1
poolHR_BM1[l][ep] =HR_BM1
poolMCAP_BM1[l][ep] =env.missCounterAP
poolMCCPU_BM1[l][ep] =env.missCounterCPU
poolCL_BM1[l][ep] =CL_Policy_UE_BM1
poolCA_BM1[l][ep] =CA_Policy_BS_BM1
#print('EE_BM1'+'_L'+str(l),'=', EE_BM1)
# Benchmark 2
EE_BM2, CL_Policy_UE_BM2, CA_Policy_BS_BM2 = env.getPolicy_BM2(nLink=l)
EE_BM2 = env.calEE(CL_Policy_UE_BM2,CA_Policy_BS_BM2)
TP_BM2 = sum(env.Throughput)
Psys_BM2 = env.P_sys/1000 # mW->W
HR_BM2 = env.calHR(CL_Policy_UE_BM2,CA_Policy_BS_BM2)
poolEE_BM2[l][ep] =EE_BM2
poolTP_BM2[l][ep] =TP_BM2
poolPsys_BM2[l][ep] =Psys_BM2
poolHR_BM2[l][ep] =HR_BM2
poolMCAP_BM2[l][ep] =env.missCounterAP
poolMCCPU_BM2[l][ep] =env.missCounterCPU
poolCL_BM2[l][ep] =CL_Policy_UE_BM2
poolCA_BM2[l][ep] =CA_Policy_BS_BM2
#print('EE_BM2'+'_L'+str(l),'=', EE_BM2)
# Benchmark 3
EE_BM3, CL_Policy_UE_BM3, CA_Policy_BS_BM3 = env.getPolicy_BM3(nLink=l)
EE_BM3 = env.calEE(CL_Policy_UE_BM3,CA_Policy_BS_BM3)
TP_BM3 = sum(env.Throughput)
Psys_BM3 = env.P_sys/1000 # mW->W
HR_BM3 = env.calHR(CL_Policy_UE_BM3,CA_Policy_BS_BM3)
poolEE_BM3[l][ep] =EE_BM3
poolTP_BM3[l][ep] =TP_BM3
poolPsys_BM3[l][ep] =Psys_BM3
poolHR_BM3[l][ep] =HR_BM3
poolMCAP_BM3[l][ep] =env.missCounterAP
poolMCCPU_BM3[l][ep] =env.missCounterCPU
poolCL_BM3[l][ep] =CL_Policy_UE_BM3
poolCA_BM3[l][ep] =CA_Policy_BS_BM3
#print('EE_BM3'+'_L'+str(l),'=', EE_BM3)
#===============================
# DDPG
if actMode == '2act':
EE_RL, CL_Policy_UE_RL, CA_Policy_BS_RL = getEE_RL(env,actMode = actMode,ddpg_cl=ddpg_cl,ddpg_ca=ddpg_ca)
elif actMode == '1act':
EE_RL, CL_Policy_UE_RL, CA_Policy_BS_RL = getEE_RL(env,actMode = actMode,ddpg_s=ddpg_s)
TP_RL = sum(env.Throughput)
Psys_RL = env.P_sys/1000
HR_RL = env.calHR(CL_Policy_UE_RL,CA_Policy_BS_RL)
poolEE_RL[ep] = EE_RL
poolTP_RL[ep] = TP_RL
poolPsys_RL[ep] = Psys_RL
poolHR_RL[ep] = HR_RL
poolMCAP_RL[ep] = env.missCounterAP
poolMCCPU_RL[ep]= env.missCounterCPU
poolCL_RL[ep] = CL_Policy_UE_RL
poolCA_RL[ep] = CA_Policy_BS_RL
# BF
if env.B==4 and env.U ==4 and env.F==5 and env.N==2 and isBF:
print('Calculate BF itr index:',ep)
EE_BF, CL_Policy_UE_BF, CA_Policy_BS_BF = env.getOptEE_BF(isSave=True)
EE_BF = env.calEE(CL_Policy_UE_BF,CA_Policy_BS_BF)
TP_BF = sum(env.Throughput)
Psys_BF = env.P_sys/1000 # mW->W
HR_BF = env.calHR(CL_Policy_UE_BF,CA_Policy_BS_BF)
poolEE_BF[ep] = EE_BF
poolTP_BF[ep] = TP_BF
poolPsys_BF[ep] = Psys_BF
poolHR_BF[ep] = HR_BF
poolMCAP_BF[ep] = env.missCounterAP
poolMCCPU_BF[ep] = env.missCounterCPU
poolCL_BF[ep] = CL_Policy_UE_BF
poolCA_BF[ep] = CA_Policy_BS_BF
# Change Environment
if loadCT:
env.g = channelTrajectory[ep]
else:
env.timeVariantChannel()
#env.resetReq()
'''
# Sample CL/CA Policy Visualization
sampled = int(nItr/2)
filename = 'data/'+env.TopologyCode+'/EVSampledPolicy/'+'['+ str(randSEED) +']'+ env.TopologyName +'_EVSampledPolicy_'
# BF PV
if env.B==4 and env.U ==4 and env.F==5 and env.N==2:
plot_UE_BS_distribution_Cache(env, poolCL_BF[sampled], poolCA_BF[sampled], poolEE_BF[sampled],filename+'BF',isEPS=True)
# RL PV
plot_UE_BS_distribution_Cache(env, poolCL_RL[sampled], poolCA_RL[sampled], poolEE_RL[sampled],filename+actMode+'RL',isEPS=False)
# BM PV
for l in range(1,env.L+1):
plot_UE_BS_distribution_Cache(env, poolCL_BM1[l][sampled], poolCA_BM1[l][sampled], poolEE_BM1[l][sampled],filename+'BM1_L'+str(l),isEPS=False)
plot_UE_BS_distribution_Cache(env, poolCL_BM2[l][sampled], poolCA_BM2[l][sampled], poolEE_BM2[l][sampled],filename+'BM2_L'+str(l),isEPS=False)
plot_UE_BS_distribution_Cache(env, poolCL_BM3[l][sampled], poolCA_BM3[l][sampled], poolEE_BM3[l][sampled],filename+'BM3_L'+str(l),isEPS=False)
'''
# Save Line
filename = 'data/'+env.TopologyCode+'/EvaluationPhase/'+'['+ str(randSEED) +']'+ env.TopologyName +'_Evaluation_'
if env.B==4 and env.U ==4 and env.F==5 and env.N==2 and isBF:
with open(filename+ 'BF.pkl', 'wb') as f:
pickle.dump([env, poolEE_BF,poolTP_BF,poolPsys_BF,poolHR_BF,poolMCAP_BF,poolMCCPU_BF,poolCL_BF,poolCA_BF], f)
with open(filename+ actMode +'RL.pkl', 'wb') as f:
pickle.dump([env, poolEE_RL,poolTP_RL,poolPsys_RL,poolHR_RL,poolMCAP_RL,poolMCCPU_RL,poolCL_RL,poolCA_RL,poolLossActor,poolLossCritic], f)
with open(filename+'BM1.pkl', 'wb') as f:
pickle.dump([env, poolEE_BM1,poolTP_BM1,poolPsys_BM1,poolHR_BM1,poolMCAP_BM1,poolMCCPU_BM1,poolCL_BM1,poolCA_BM1], f)
with open(filename+'BM2.pkl', 'wb') as f:
pickle.dump([env, poolEE_BM2,poolTP_BM2,poolPsys_BM2,poolHR_BM2,poolMCAP_BM2,poolMCCPU_BM2,poolCL_BM2,poolCA_BM2], f)
with open(filename+'BM3.pkl', 'wb') as f:
pickle.dump([env, poolEE_BM3,poolTP_BM3,poolPsys_BM3,poolHR_BM3,poolMCAP_BM3,poolMCCPU_BM3,poolCL_BM3,poolCA_BM3], f)
# Calculate Loss Count
lossCount = 0
for i in range(len(poolEE_RL)):
if poolEE_RL[i] < poolEE_BM1[1][i]:
lossCount+=1
return lossCount
def getEE_RL(env,actMode,ddpg_s=None,ddpg_cl=None,ddpg_ca=None):
EE_RL = 0
RL_CLPolicy_UE=[]
RL_CAPolicy_BS=[]
obs = env.reset()# Get initial state
for i in range(1000):
if actMode == '2act':
noise = np.random.normal(0, 0,size=env.dimActCL)
a_cl = ddpg_cl.action(obs,noise)# choose action [ env.U*env.B x 1 ]
#a_ca = opt_caching_state.flatten()
noise = np.random.normal(0, 0,size=env.dimActCA)
a_ca = ddpg_ca.action(obs,noise)# choose action [ env.U*env.B x 1 ]
action = np.concatenate((a_cl, a_ca), axis=0)
elif actMode == '1act':
noise = np.random.normal(0, 0,size=env.dimAct)
action = ddpg_s.action(obs,noise)
obs, reward, done, info = env.step(action)
if reward>EE_RL:
EE_RL = reward
RL_CLPolicy_UE, RL_CAPolicy_BS = env.action2Policy(action)
#print('rlBestCLPolicy_UE:',rlBestCLPolicy_UE)
#print('rlBestCAPolicy_BS:',rlBestCAPolicy_BS)
#print('rlBestEE:',rlBestEE)
EE_RL = env.calEE(RL_CLPolicy_UE,RL_CAPolicy_BS)
return EE_RL, RL_CLPolicy_UE,RL_CAPolicy_BS
if __name__ == '__main__':
actMode = '1act'
envSeed = 0
nItr = 100
# Good case: 4.4.5.2 [31] / 10.5.20.2 [3,6]
#for randSEED in range(1,30):
for randSEED in [31]:
print('randSEED:',randSEED)
print('Execution Timestamp:',datetime.now())
##################### hyper parameters ####################
# Random Seed
np.random.seed(randSEED)
torch.manual_seed(randSEED)
torch.cuda.manual_seed_all(randSEED)
allWin = True
#------------------------------------------------------------------------
'''
# Training Phase
# new ENV
env = BS(nBS=4,nUE=4,nMaxLink=2,nFile=5,nMaxCache=2,loadENV = True,SEED=0,obsIdx=2)
#env = BS(nBS=10,nUE=5,nMaxLink=3,nFile=20,nMaxCache=2,loadENV=True,SEED=envSeed,obsIdx=1)
allWin = trainModel(env,actMode=actMode,changeReq=False, changeChannel=True, loadActor = False,randSEED=randSEED)
filename = 'data/'+env.TopologyCode+'/TrainingPhase/'+'['+ str(randSEED) +']'+ env.TopologyName +str(MAX_EPISODES*MAX_EP_STEPS)+'_Train_'
plotHistory(env,filename,isEPS=False,loadBF=False)
#------------------------------------------------------------------------
# Evaluation Phase
# new ENV
env = BS(nBS=4,nUE=4,nMaxLink=2,nFile=5,nMaxCache=2,loadENV = True,SEED=0,obsIdx=2)
#env = BS(nBS=10,nUE=5,nMaxLink=3,nFile=20,nMaxCache=2,loadENV=True,SEED=envSeed,obsIdx=1)
if allWin:
lossCount = evaluateModel(env,actMode=actMode, nItr=nItr,randSEED=randSEED,loadCT=True,isBF=True)
# Plot Performance
filename = 'data/'+env.TopologyCode+'/EvaluationPhase/'+'['+ str(randSEED) +']'+ env.TopologyName +'_Evaluation_'
plotHistory(env,filename,isEPS=False,loadBF=False)
'''
#==============================================================================================
'''
# plot Evaluation Final for 4.4.5.2
envSeed=0
randSEED=31
nItr = 10
sampled = int(nItr/2)
# new ENV
env = BS(nBS=4,nUE=4,nMaxLink=2,nFile=5,nMaxCache=2,loadENV=True,SEED=envSeed,obsIdx=1)
filename = 'data/'+env.TopologyCode+'/[envSeed '+str(envSeed)+'][randSeed '+str(randSEED)+'][OBS1]/EvaluationPhase/'+'['+ str(randSEED) +']'+ env.TopologyName +'_Evaluation_'
plotHistory(env,filename,isEPS=True,loadBF=True)
#------------------------------------------------------------------------
# Plot Sampled CL/CA Policy Visualization
# Load Brute Force Policy
with open(filename+ 'BF.pkl', 'rb') as f:
env, poolEE_BF,poolTP_BF,poolPsys_BF,poolHR_BF,poolMCAP_BF,poolMCCPU_BF,poolCL_BF,poolCA_BF = pickle.load(f)
# Load RL Policies
with open(filename+'BF.pkl','rb') as f:
env, poolEE_BM3,poolTP_BM3,poolPsys_BM3,poolHR_BM3,poolMCAP_BM3,poolMCCPU_BM3,poolCL_BM3,poolCA_BM3 = pickle.load(f)
with open(filename+ actMode +'RL.pkl','rb') as f:
env, poolEE_RL,poolTP_RL,poolPsys_RL,poolHR_RL,poolMCAP_RL,poolMCCPU_RL,poolCL_RL,poolCA_RL\
,poolLossActor,poolLossCritic = pickle.load(f)
# Load Benchmarks Policies
with open(filename+'BM1.pkl','rb') as f:
env, poolEE_BM1,poolTP_BM1,poolPsys_BM1,poolHR_BM1,poolMCAP_BM1,poolMCCPU_BM1,poolCL_BM1,poolCA_BM1 = pickle.load(f)
with open(filename+'BM2.pkl','rb') as f:
env, poolEE_BM2,poolTP_BM2,poolPsys_BM2,poolHR_BM2,poolMCAP_BM2,poolMCCPU_BM2,poolCL_BM2,poolCA_BM2 = pickle.load(f)
with open(filename+'BM3.pkl','rb') as f:
env, poolEE_BM3,poolTP_BM3,poolPsys_BM3,poolHR_BM3,poolMCAP_BM3,poolMCCPU_BM3,poolCL_BM3,poolCA_BM3 = pickle.load(f)
#
filename = 'data/'+env.TopologyCode+'/[envSeed '+str(envSeed)+'][randSeed '+str(randSEED)+'][OBS1]/EVSampledPolicy/'+'['+ str(randSEED) +']'+ env.TopologyName +'_EVSampledPolicy_'
# BF PV
plot_UE_BS_distribution_Cache(env, poolCL_BF[sampled], poolCA_BF[sampled], poolEE_BF[sampled],filename+'BF',isDetail=True)
plot_UE_BS_distribution_Cache(env, poolCL_BF[sampled], poolCA_BF[sampled], poolEE_BF[sampled],filename+'BF',isEPS=True)
# RL PV
plot_UE_BS_distribution_Cache(env, poolCL_RL[sampled], poolCA_RL[sampled], poolEE_RL[sampled],filename+'Proposed',isDetail=True)
plot_UE_BS_distribution_Cache(env, poolCL_RL[sampled], poolCA_RL[sampled], poolEE_RL[sampled],filename+'Proposed',isEPS=True)
# BM PV
for l in [1,env.L]:
plot_UE_BS_distribution_Cache(env, poolCL_BM1[l][sampled], poolCA_BM1[l][sampled], poolEE_BM1[l][sampled],filename+'BM1(l='+str(l)+')',isDetail=True)
plot_UE_BS_distribution_Cache(env, poolCL_BM2[l][sampled], poolCA_BM2[l][sampled], poolEE_BM2[l][sampled],filename+'BM2(l='+str(l)+')',isDetail=True)
plot_UE_BS_distribution_Cache(env, poolCL_BM3[l][sampled], poolCA_BM3[l][sampled], poolEE_BM3[l][sampled],filename+'BM3(l='+str(l)+')',isDetail=True)
plot_UE_BS_distribution_Cache(env, poolCL_BM1[l][sampled], poolCA_BM1[l][sampled], poolEE_BM1[l][sampled],filename+'BM1(l='+str(l)+')',isEPS=True)
plot_UE_BS_distribution_Cache(env, poolCL_BM2[l][sampled], poolCA_BM2[l][sampled], poolEE_BM2[l][sampled],filename+'BM2(l='+str(l)+')',isEPS=True)
plot_UE_BS_distribution_Cache(env, poolCL_BM3[l][sampled], poolCA_BM3[l][sampled], poolEE_BM3[l][sampled],filename+'BM3(l='+str(l)+')',isEPS=True)
'''
#==============================================================================================
# plot Evaluation Final for 10.5.20.2
envSeed = 0
randSEED = 3
nItr = 100
sampled = int(nItr/2)
# new ENV
env = BS(nBS=10,nUE=5,nMaxLink=3,nFile=20,nMaxCache=2,loadENV=True,SEED=envSeed,obsIdx=1)
#------------------------------------------------------------------------
# Plot Performance
filename = 'data/'+env.TopologyCode+'/[envSeed '+str(envSeed)+'][randSeed '+str(randSEED)+'][OBS1]/EvaluationPhase/'+'['+ str(randSEED) +']'+ env.TopologyName +'_Evaluation_'
#plotHistory(env,filename,isEPS=True,loadBF=False)
#------------------------------------------------------------------------
# Plot Sampled CL/CA Policy Visualization
# Load RL Policies
with open(filename+ actMode +'RL.pkl','rb') as f:
env, poolEE_RL,poolTP_RL,poolPsys_RL,poolHR_RL,poolMCAP_RL,poolMCCPU_RL,poolCL_RL,poolCA_RL\
,poolLossActor,poolLossCritic = pickle.load(f)
# Load Benchmarks Policies
with open(filename+'BM1.pkl','rb') as f:
env, poolEE_BM1,poolTP_BM1,poolPsys_BM1,poolHR_BM1,poolMCAP_BM1,poolMCCPU_BM1,poolCL_BM1,poolCA_BM1 = pickle.load(f)
with open(filename+'BM2.pkl','rb') as f:
env, poolEE_BM2,poolTP_BM2,poolPsys_BM2,poolHR_BM2,poolMCAP_BM2,poolMCCPU_BM2,poolCL_BM2,poolCA_BM2 = pickle.load(f)
with open(filename+'BM3.pkl','rb') as f:
env, poolEE_BM3,poolTP_BM3,poolPsys_BM3,poolHR_BM3,poolMCAP_BM3,poolMCCPU_BM3,poolCL_BM3,poolCA_BM3 = pickle.load(f)
#
filename = 'data/'+env.TopologyCode+'/[envSeed '+str(envSeed)+'][randSeed '+str(randSEED)+'][OBS1]/EVSampledPolicy/'+'['+ str(randSEED) +']'+ env.TopologyName +'_EVSampledPolicy_'
'''
for sampled in range(0,100,10):
# RL PV
plot_UE_BS_distribution_Cache(env, poolCL_RL[sampled], poolCA_RL[sampled], poolEE_RL[sampled],filename+'Proposed',isEPS=True)
plot_UE_BS_distribution_Cache(env, poolCL_RL[sampled], poolCA_RL[sampled], poolEE_RL[sampled],filename+'Proposed',isDetail=True)
# BM PV
for l in [1,env.L]:
plot_UE_BS_distribution_Cache(env, poolCL_BM1[l][sampled], poolCA_BM1[l][sampled], poolEE_BM1[l][sampled],filename+'BM1(l='+str(l)+')',isEPS=True)
plot_UE_BS_distribution_Cache(env, poolCL_BM2[l][sampled], poolCA_BM2[l][sampled], poolEE_BM2[l][sampled],filename+'BM2(l='+str(l)+')',isEPS=True)
plot_UE_BS_distribution_Cache(env, poolCL_BM3[l][sampled], poolCA_BM3[l][sampled], poolEE_BM3[l][sampled],filename+'BM3(l='+str(l)+')',isEPS=True)
plot_UE_BS_distribution_Cache(env, poolCL_BM1[l][sampled], poolCA_BM1[l][sampled], poolEE_BM1[l][sampled],filename+'BM1(l='+str(l)+')',isDetail=True)
plot_UE_BS_distribution_Cache(env, poolCL_BM2[l][sampled], poolCA_BM2[l][sampled], poolEE_BM2[l][sampled],filename+'BM2(l='+str(l)+')',isDetail=True)
plot_UE_BS_distribution_Cache(env, poolCL_BM3[l][sampled], poolCA_BM3[l][sampled], poolEE_BM3[l][sampled],filename+'BM3(l='+str(l)+')',isDetail=True)
'''
for sampled in range(1,100):
for k in range(len(poolCL_RL[sampled])):
if (poolCL_RL[sampled][k] == poolCL_RL[sampled-1][k]).all():
print('the same')
else:
print('diff occur at '+sampled)
for sampled in range(1,100):
for k in range(len(poolCL_BM1[1][sampled])):
if (poolCL_BM1[1][sampled][k] == poolCL_BM1[1][sampled-1][k]).all():
print('the same')
else:
print('diff occur at '+sampled)
for sampled in range(1,100):
for k in range(len(poolCL_BM1[3][sampled])):
if (poolCL_BM1[3][sampled][k] == poolCL_BM1[3][sampled-1][k]).all():
print('the same')
else:
print('diff occur at '+sampled)
#==============================================================================================
# multi-instance training
'''
nJob=2
#with concurrent.futures.ProcessPoolExecutor(max_workers= (num_cores-2) ) as executor:
with concurrent.futures.ProcessPoolExecutor(max_workers= nJob ) as executor:
futures = []
for i in range(nJob):
future = executor.submit(trainModel,env,actMode=actMode,changeReq=False, changeChannel=True, loadActor = False,randSEED=i)
futures.append(future)
for future in tqdm(concurrent.futures.as_completed(futures),total=len(futures),bar_format='{l_bar}{bar:10}{r_bar}{bar:-10b}'):
#print(future.result())
randSEED = future.result()
print('\n randSEED: ',randSEED,' is Completed')
for randSEED in range(nJob):
filename = 'data/'+env.TopologyCode+'/TrainingPhase/'+'['+ str(randSEED) +']'+ env.TopologyName +str(MAX_EPISODES*MAX_EP_STEPS)+'_Train_'
plotHistory(filename,isPlotEE=True,isPlotTP=True,isPlotPsys=True,isPlotHR=True,isEPS=False)
'''