I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
10
,
No
.
2
,
May
201
8
,
p
p
.
5
4
5
~5
5
3
I
SS
N:
2502
-
4752
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ee
cs
.
v
1
0
.
i2
.
p
p
545
-
5
5
3
545
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e.
co
m/jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
ijeec
s
Intellig
ent
P
a
c
k
e
t
Delivery
in
Rout
er Using
S
truc
tur
e
O
pti
m
i
zed
Neura
l Net
w
o
rk
R.
Dee
ba
la
ks
h
m
i
*
1
,
V.
L
.
J
y
o
t
hi
2
1
S
a
th
y
a
b
a
m
a
Un
iv
e
rsit
y
,
Ch
e
n
n
a
i
,
In
d
ia
2
CS
E,
Je
p
p
iaa
r
En
g
in
e
e
rin
g
Co
ll
e
g
e
,
Ch
e
n
n
a
i,
I
n
d
ia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
N
ov
2
1
,
2
0
1
7
R
ev
i
s
ed
J
an
2
9
,
2
0
1
8
A
cc
ep
ted
F
eb
1
9
,
2
0
1
8
T
h
e
In
tern
e
t
it
se
lf
i
s
a
w
o
rld
w
id
e
n
e
tw
o
rk
c
o
n
n
e
c
ti
n
g
m
il
li
o
n
s
o
f
c
o
m
p
u
ters
a
n
d
les
s
sig
n
if
ica
n
t
n
e
tw
o
rk
s.
Co
m
p
u
ters
c
o
m
m
u
n
ica
ted
b
y
ro
u
ters
.
Cru
c
ial
th
e
ro
le
o
f
a
ro
u
ter
is
t
o
o
u
r
tec
h
n
iq
u
e
o
f
c
o
m
m
u
n
ica
ti
n
g
a
n
d
c
o
m
p
u
ti
n
g
.
Ro
u
ters
a
re
s
it
u
a
ted
a
t
g
a
tew
a
y
s
,
th
e
sp
a
c
e
s
w
h
e
re
t
w
o
o
r
m
o
re
n
e
tw
o
rk
s
c
o
n
n
e
c
t,
a
n
d
a
re
th
e
d
e
c
isiv
e
d
e
v
ice
th
a
t
k
e
e
p
s
d
a
ta
f
lo
w
b
e
t
w
e
e
n
n
e
tw
o
rk
s
a
n
d
k
e
e
p
s
th
e
n
e
tw
o
rk
s
c
o
n
n
e
c
ted
to
th
e
In
ter
n
e
t.
W
h
e
n
d
a
ta
is
se
n
t
b
e
tw
e
e
n
p
lac
e
s
o
n
o
n
e
n
e
tw
o
rk
o
r
f
ro
m
o
n
e
n
e
tw
o
rk
to
a
se
c
o
n
d
n
e
tw
o
rk
th
e
d
a
ta
is
a
lwa
y
s
se
e
n
a
n
d
i
n
ten
d
e
d
f
o
r
t
o
th
e
p
ro
p
e
r
p
lac
e
b
y
th
e
r
o
u
ter.
T
h
e
ro
u
ter
c
a
rries
o
u
t
th
is
b
y
u
sin
g
h
e
a
d
e
rs
a
n
d
ro
u
ti
n
g
tab
les
to
e
sta
b
li
sh
t
h
e
b
e
st
p
a
th
f
o
r
ro
u
ti
n
g
th
e
d
a
ta
p
a
c
k
e
ts.
T
h
is
tri
m
d
o
w
n
th
e
e
ff
e
c
ti
v
e
n
e
ss
o
f
e
d
g
e
ro
u
ter
o
n
ly
w
h
e
n
th
e
p
a
th
e
n
g
a
g
e
d
,
it
w
il
l
e
n
h
a
n
c
e
d
b
y
c
l
a
ss
i
f
ica
ti
o
n
m
e
th
o
d
,
p
re
d
icta
b
le cla
ss
if
ica
ti
o
n
m
e
th
o
d
s
li
k
e
p
o
rt
b
a
se
d
,
d
e
e
p
p
a
c
k
e
t
in
sp
e
c
ti
o
n
a
n
d
sta
ti
stica
l
c
las
sif
i
c
a
ti
o
n
a
re
g
iv
e
les
s
p
re
c
isio
n
.
I
n
t
h
is
sy
ste
m
stru
c
tu
re
d
o
p
ti
m
ize
d
n
e
u
ra
l
n
e
tw
o
rk
i
s
u
se
d
f
o
r
m
o
re
p
re
c
ise
o
rg
a
n
iza
ti
o
n
.
Clas
sif
ic
a
ti
o
n
o
u
t
p
u
t
f
o
rw
a
rd
e
d
to
ro
u
ter
d
y
n
a
m
i
c
a
ll
y
f
o
r
in
tellec
t
u
a
l
p
a
c
k
e
t
d
e
li
v
e
r
y
.
T
h
e
m
e
th
o
d
w
il
l
im
p
ro
v
e
ro
u
ter
c
o
m
p
e
ten
c
e
b
y
g
re
a
ter
t
h
a
n
b
e
f
o
re
th
ro
u
g
h
p
u
t
a
n
d
d
e
c
re
a
se
d
late
n
c
y
.
K
ey
w
o
r
d
s
:
F
lo
w
c
a
teg
o
riza
ti
o
n
Ne
tw
o
rk
o
p
e
ra
ti
o
n
s
Ne
u
ra
l
n
e
tw
o
rk
a
p
p
li
c
a
ti
o
n
s
Ro
u
ti
n
g
Co
p
y
rig
h
t
©
2
0
1
8
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
All
rig
h
ts re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
R
.
Dee
b
alak
s
h
m
i
,
R
esear
ch
Sch
o
lar
,
Sath
y
ab
a
m
a
U
n
i
v
er
s
it
y
,
J
ep
p
iar
Nag
ar
,
C
h
e
n
n
a
i,
T
am
il
n
ad
u
,
I
n
d
ian
-
600119.
E
m
ail:
d
ee
p
ar
1
1
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
L
a
k
h
i
n
a
et
al.
[
1
]
ex
p
lain
ed
th
e
d
is
tr
ib
u
t
io
n
s
o
f
p
ac
k
et
f
e
atu
r
es
o
b
s
er
v
ed
in
f
lo
w
a
tr
a
ce
w
h
ic
h
r
ev
ea
ls
b
o
th
t
h
e
p
r
esen
ce
a
n
d
th
e
s
tr
u
ct
u
r
e
o
f
a
w
id
e
r
an
g
e
o
f
an
o
m
alie
s
.
Usi
n
g
en
tr
o
p
y
a
s
u
m
m
ar
izat
io
n
to
o
l
th
at
th
e
m
et
h
o
d
ex
p
r
ess
ed
th
at
th
e
an
al
y
s
is
o
f
f
ea
t
u
r
e
d
i
s
tr
ib
u
tio
n
s
lead
s
to
s
ig
n
if
ican
t
ad
v
an
ce
s
u
s
i
n
g
f
ea
tu
r
e
d
is
tr
ib
u
tio
n
s
,
an
o
m
alie
s
n
atu
r
all
y
f
all
i
n
to
d
is
tin
c
t
an
d
m
ea
n
in
g
f
u
l
cl
u
s
ter
s
.
E
s
ta
n
e
t
al.
[
2
]
in
tr
o
d
u
ce
d
s
ca
lab
le
al
g
o
r
ith
m
s
f
o
r
id
e
n
ti
f
y
in
g
t
h
e
lar
g
e
f
lo
w
s
:
s
a
m
p
l
e
an
d
h
o
ld
an
d
m
u
lti
s
ta
g
e
f
i
lter
s
,
w
h
ich
ta
k
e
a
co
n
s
ta
n
t
n
u
m
b
er
o
f
m
e
m
o
r
y
r
ef
er
en
ce
s
p
er
p
ac
k
et
an
d
u
s
e
a
s
m
all
a
m
o
u
n
t
o
f
m
e
m
o
r
y
.
T
h
e
tech
n
iq
u
e
d
escr
ib
ed
o
p
tim
izatio
n
s
s
u
ch
as
ea
r
l
y
r
e
m
o
v
al
an
d
co
n
s
er
v
ativ
e
u
p
d
ate
th
at
f
u
r
t
h
er
i
m
p
r
o
v
e
th
e
ac
cu
r
ac
y
o
f
alg
o
r
ith
m
s
,
as
m
ea
s
u
r
ed
o
n
r
ea
l
tr
af
f
ic
tr
ac
e
s
,
b
y
a
n
o
r
d
er
o
f
m
ag
n
it
u
d
e.
P
en
g
el
al.
[
3
]
ex
p
r
ess
ed
E
n
er
g
y
ef
f
icien
t
clu
s
ter
in
g
u
s
i
n
g
e
n
er
g
y
p
r
ed
ictio
n
m
ec
h
a
n
is
m
.
T
h
e
y
h
a
v
e
u
s
ed
d
if
f
er
en
t
en
er
g
y
lev
el
s
en
s
o
r
s
w
it
h
d
if
f
er
e
n
t
m
o
n
ito
r
i
n
g
o
b
j
ec
ts
.
Fo
r
clu
s
ter
h
ea
d
s
elec
tio
n
r
esid
u
al
e
n
er
g
y
o
f
n
o
d
es
is
p
r
ed
icted
in
s
tead
o
f
b
r
o
ad
ca
s
tin
g
it i
n
ev
er
y
r
o
u
n
d
,
also
th
e
y
u
s
ed
co
m
m
u
n
icatio
n
co
s
t.
Ng
u
y
e
n
et
al.
[
4
]
d
esig
n
ed
K
-
Me
a
n
s
cl
u
s
ter
in
g
al
g
o
r
ith
m
f
o
r
p
ac
k
et
s
ep
ar
atio
n
s
.
P
ac
k
e
ts
tr
ac
ed
in
co
m
p
u
ter
n
et
w
o
r
k
s
.
Fro
m
tr
ac
ed
p
ac
k
ets
f
lo
w
is
d
ef
i
n
e
d
,
an
d
f
lo
w
s
a
r
e
g
r
o
u
p
ed
i
n
to
clu
s
ter
s
.
Flo
w
s
class
i
f
ied
b
ased
o
n
k
-
cl
u
s
ter
c
en
ter
s
.
T
h
e
cl
u
s
ter
ce
n
ter
s
ar
e
d
ef
in
ed
b
y
p
ac
k
et
le
n
g
t
h
a
n
d
p
ac
k
ets ar
r
i
v
al
ti
m
e
o
f
ap
p
licatio
n
.
A
p
p
licatio
n
s
wh
ich
ar
e
h
a
v
in
g
s
a
m
e
p
ac
k
et
l
en
g
t
h
o
r
n
ea
r
b
y
p
ac
k
et
le
n
g
th
ar
e
co
m
p
ar
ed
w
it
h
ex
is
t
in
g
cl
u
s
ter
,
if
it
is
m
atc
h
ed
co
m
b
i
n
ed
to
ex
i
s
ti
n
g
cl
u
s
ter
o
t
h
er
w
is
e
n
e
w
c
lu
s
ter
g
r
o
u
p
i
f
f
o
r
m
ed
.
I
n
m
ac
h
in
e
lear
n
in
g
SV
M
it
u
s
es
s
u
p
er
v
is
ed
lear
n
i
n
g
al
g
o
r
it
h
m
a
n
d
an
al
y
ze
d
ata
u
s
ed
f
o
r
class
if
icatio
n
a
n
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4
7
5
2
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
10
,
No
.
2
,
Ma
y
2
0
1
8
:
5
4
5
–
553
546
r
eg
r
ess
io
n
.
I
f
n
o
tr
ain
ed
s
e
t
t
h
at
m
e
th
o
d
is
ca
lle
d
u
n
s
u
p
er
v
is
ed
al
g
o
r
ith
m
,
h
er
e
i
n
p
u
t
f
l
o
w
s
ar
e
g
r
o
u
p
ed
as
clu
s
ter
s
.
T
h
er
e
ar
e
tw
o
t
y
p
es o
f
SVM.
1.
L
i
n
ea
r
SVM
an
d
2.
No
n
-
li
n
ea
r
SVM.
L
inea
r
SVM
:
I
n
lin
ea
r
S
VM
,
it
h
as
li
n
ea
r
n
-
d
i
m
e
n
s
io
n
al
v
e
cto
r
an
d
if
t
h
er
e
is
a
p
o
s
s
ib
ilit
y
o
f
s
ep
ar
atin
g
n
-
1
h
y
p
er
p
la
n
e
v
ec
to
r
it
is
c
alled
a
li
n
ea
r
cla
s
s
i
f
ier
.
T
h
e
h
y
p
er
p
lan
e
is
u
s
ed
to
cla
s
s
i
f
y
d
ata
an
d
o
p
ti
m
al
h
y
p
er
p
lan
e
w
il
l
g
i
v
e
lar
g
est
s
ep
ar
ati
o
n
o
r
,
B
ad
s
elec
tio
n
o
f
h
y
p
er
p
lan
e
w
il
l
g
i
v
e
n
o
is
e
d
ata
clas
s
if
ica
tio
n
.
Dis
ta
n
ce
o
f
d
ata
co
m
p
ar
ed
w
ith
h
y
p
er
p
lan
e
s
o
m
a
x
i
m
u
m
v
al
u
e
s
tet
te
d
f
o
r
it
.
No
n
-
L
inea
r
SVM
:
I
n
n
o
n
-
l
i
n
ea
r
SVM,
t
h
e
cla
s
s
i
f
ier
i
s
cr
ea
ted
b
y
ap
p
l
y
in
g
t
h
e
k
er
n
el
tr
ick
to
m
ax
i
m
u
m
-
m
ar
g
i
n
h
y
p
er
p
lan
es.
T
h
e
y
ar
e
ef
f
ec
t
iv
e
i
n
h
ig
h
d
i
m
e
n
s
io
n
al
s
p
ac
es.
T
h
ey
ar
e
s
t
ill
ef
f
ec
tiv
e
i
n
ca
s
es
w
h
er
e
n
u
m
b
er
o
f
d
i
m
e
n
s
io
n
s
i
s
g
r
ea
t
er
th
an
t
h
e
n
u
m
b
er
o
f
s
a
m
p
le
s
.
I
t u
s
es a
s
u
b
s
et
o
f
tr
ai
n
i
n
g
p
o
in
ts
i
n
t
h
e
d
ec
is
io
n
f
u
n
ctio
n
a
n
d
s
o
i
t
is
m
e
m
o
r
y
ef
f
icien
t.
SVM
h
a
s
s
o
m
e
d
is
a
d
v
an
ta
g
es.
P
ar
ticu
lar
l
y
,
if
th
e
n
u
m
b
er
o
f
f
ea
tu
r
es
is
m
u
ch
g
r
ea
ter
th
a
n
th
e
n
u
m
b
er
o
f
s
a
m
p
les,
t
h
e
m
et
h
o
d
is
lik
el
y
to
g
iv
e
p
o
o
r
p
er
f
o
r
m
an
ce
.
T
h
e
y
d
o
n
o
t
d
ir
ec
tl
y
p
r
o
v
id
e
p
r
o
b
a
b
ilit
y
e
s
ti
m
ate
s
,
th
e
s
e
ar
e
ca
lcu
lated
u
s
in
g
an
e
x
p
en
s
i
v
e
f
i
v
e
-
f
o
ld
cr
o
s
s
-
v
alid
atio
n
.
O
NE
-
CL
A
SS
SVM
:
A
n
o
th
e
r
alter
n
ativ
e
o
f
SVM
is
o
n
e
-
class
SVM.
I
t
is
an
u
n
s
u
p
er
v
is
ed
alg
o
r
ith
m
th
a
t
lear
n
s
a
d
ec
is
io
n
f
u
n
ct
io
n
f
o
r
n
o
v
elt
y
d
etec
tio
n
t
h
at
it c
la
s
s
i
f
ies
n
e
w
d
ata
a
s
s
i
m
ilar
o
r
d
if
f
er
en
t to
th
e
tr
ain
i
n
g
s
et.
B
asicall
y
,
i
t
al
s
o
s
ep
ar
at
es
all
th
e
d
ata
p
o
in
t
s
f
r
o
m
t
h
e
o
r
ig
i
n
an
d
m
a
x
i
m
izes
t
h
e
d
is
ta
n
ce
f
r
o
m
th
i
s
h
y
p
er
p
lan
e
to
th
e
o
r
ig
in
.
T
h
e
o
p
tim
a
l
h
y
p
er
p
la
n
e
is
t
h
e
o
n
e
th
at
r
ep
r
esen
ts
t
h
e
lar
g
e
s
t
s
ep
ar
atio
n
,
o
r
m
ar
g
in
,
b
et
w
ee
n
t
h
e
t
w
o
clas
s
es.
I
f
t
h
e
n
u
m
b
er
o
f
s
a
m
p
le
s
i
s
m
o
r
e
th
an
th
e
cr
ea
tio
n
o
f
o
p
ti
m
al
b
o
u
n
d
ar
y
i
s
d
if
f
ic
u
lt
an
d
it p
er
f
o
r
m
an
ce
i
s
also
af
f
e
cted
.
I
t u
s
es o
n
l
y
u
n
s
u
p
er
v
i
s
e
d
alg
o
r
ith
m
.
He
n
ce
,
it c
a
n
n
o
t i
d
en
tify
n
e
w
tr
af
f
ic.
Mo
o
r
e
et
al.
[
5
]
d
esig
n
ed
m
ac
h
in
e
lear
n
i
n
g
Naï
v
e
B
a
y
e
s
tech
n
iq
u
e
to
ca
teg
o
r
ize
i
n
te
r
n
et
tr
af
f
ic
b
ased
o
n
ap
p
licatio
n
.
T
h
e
tr
a
f
f
ic
i
n
th
e
i
n
ter
n
et
ap
p
licatio
n
s
w
er
e
cla
s
s
i
f
ied
i
n
to
d
i
f
f
er
en
t
ca
te
g
o
r
ies,
e.
g
.
m
ail,
w
eb
s
er
v
ices,
p2p,
m
u
l
ti
m
ed
ia
an
d
g
a
m
e
s
.
T
h
e
au
th
o
r
s
u
s
ed
ac
cu
r
ac
y
a
s
a
class
i
f
icatio
n
m
etr
ic
to
ev
alu
a
te
p
er
f
o
r
m
a
n
ce
o
f
clas
s
if
ier
.
T
h
is
r
es
u
lt
d
ep
icted
th
at
n
aïv
e
b
a
y
es
tec
h
n
iq
u
es
h
a
v
e
6
5
%
ac
cu
r
ac
y
in
class
i
f
icatio
n
.
T
w
o
r
ef
i
n
e
m
e
n
ts
w
er
e
p
er
f
o
r
m
ed
f
o
r
i
m
p
r
o
v
in
g
clas
s
if
icatio
n
ac
c
u
r
ac
y
u
s
i
n
g
n
aï
v
e
b
a
y
es
k
er
n
el
es
ti
m
atio
n
a
n
d
f
as
t c
o
r
r
elatio
n
b
ased
f
ilter
m
et
h
o
d
.
I
t
g
iv
e
s
9
5
% a
s
th
e
o
v
er
all
ac
cu
r
ac
y
.
L
u
ig
i
et
al.
[
6
]
d
esig
n
ed
Se
L
eCT
tech
n
iq
u
e,
a
Sel
f
-
L
ea
r
n
i
n
g
C
las
s
i
f
ier
f
o
r
I
n
ter
n
et
T
r
af
f
ic.
I
t
u
s
e
s
u
n
s
u
p
er
v
i
s
ed
al
g
o
r
ith
m
to
a
u
t
o
m
a
ticall
y
g
r
o
u
p
s
th
e
f
lo
w
i
n
to
h
o
m
o
g
en
eo
u
s
g
r
o
u
p
b
ased
o
n
ap
p
licatio
n
.
I
t
d
o
esn
’
t
r
eq
u
ir
e
p
r
io
r
k
n
o
w
le
d
g
e
o
f
tr
ai
n
i
n
g
s
et
to
id
en
ti
f
y
t
h
e
tr
a
f
f
ic
f
lo
w
s
.
Au
to
m
ati
ca
ll
y
t
h
e
al
g
o
r
it
h
m
g
r
o
u
p
s
t
h
e
f
lo
w
s
i
n
to
h
o
m
o
g
e
n
eo
u
s
cl
u
s
ter
s
u
s
i
n
g
s
tati
s
tical
f
ea
t
u
r
es.
I
t
a
ls
o
s
i
m
p
li
f
ie
s
t
h
e
lab
el
ass
ig
n
m
en
t
b
y
as
s
i
g
n
i
n
g
lab
els
to
th
e
clu
s
ter
b
ased
o
n
ap
p
licatio
n
.
Fu
r
t
h
er
m
o
r
e,
it
u
s
e
s
s
el
f
s
ee
d
in
g
ap
p
r
o
ac
h
to
p
r
o
ce
s
s
n
ex
t
b
atch
e
s
o
f
f
lo
w
s
b
e
f
o
r
e
ass
i
g
n
in
g
lab
els
to
p
r
ev
io
u
s
clu
s
ter
.
T
h
e
au
th
o
r
ev
al
u
ated
th
e
p
er
f
o
r
m
an
ce
o
f
SeL
e
C
T
u
s
i
n
g
d
i
f
f
er
e
n
t
tr
af
f
ic
tr
ac
es
co
llected
f
r
o
m
I
SP
lo
ca
ted
in
th
e
d
if
f
er
en
t
co
n
ti
n
e
n
t
s
.
T
h
e
ex
p
er
i
m
en
t
s
s
h
o
w
ed
th
at
it
ac
h
iev
e
s
o
v
er
all
ac
cu
r
ac
y
.
T
h
e
ac
cu
r
ac
y
i
s
ac
h
ie
v
ed
is
n
ea
r
l
y
9
8
%
an
d
it
d
is
co
v
er
n
e
w
p
r
o
to
co
ls
an
d
ap
p
licatio
n
in
t
h
e
tr
af
f
ic
tr
ac
e
s
.
B
u
j
lo
w
et
al.
[
7
]
d
is
cu
s
s
ed
,
C
5
.
0
is
th
e
d
ec
is
io
n
tr
ee
b
ased
a
lg
o
r
ith
m
a
n
d
u
s
e
t
h
e
co
n
ce
p
t
o
f
m
ac
h
i
n
e
lear
n
in
g
a
lg
o
r
it
h
m
.
I
t
i
s
ea
s
ie
r
to
u
s
e
a
n
d
m
e
m
o
r
y
e
f
f
icien
t.
I
t
g
e
n
er
ates
t
h
e
d
ec
is
io
n
tr
ee
b
ased
o
n
s
et
o
f
tr
ain
i
n
g
s
et
.
T
h
e
tr
ee
s
ar
e
u
s
ed
to
class
i
f
y
th
e
s
et
o
f
te
s
t
ca
s
es.
T
h
e
c5
.
0
class
if
ier
u
s
es
co
m
m
a
n
d
lin
e
in
ter
f
ac
e
to
g
en
er
ate
th
e
r
u
le
s
f
o
r
d
ec
is
io
n
tr
ee
a
n
d
te
s
t
t
h
e
class
i
f
ier
.
T
h
e
e
x
p
er
i
m
e
n
t
w
a
s
e
x
ec
u
ted
m
a
n
y
ti
m
e
s
u
s
i
n
g
d
if
f
er
en
t
s
et
o
f
tr
a
in
i
n
g
f
lo
w
s
a
n
d
attr
ib
u
te
s
l
i
k
e
p
ac
k
et
s
ize,
p
ac
k
et
le
n
g
th
,
n
u
m
b
er
o
f
f
lo
w
s
.
T
h
e
attr
ib
u
tes
f
o
r
f
lo
w
s
ar
e
u
s
ed
to
co
n
s
tr
u
ct
test
ca
s
e
s
.
I
t
p
r
o
d
u
ce
d
9
8
%
ac
cu
r
ac
y
w
h
e
n
a
cc
u
r
ate
test
i
n
g
a
n
d
tr
ain
i
n
g
d
ata
u
s
ed
.
Kar
ag
ia
n
n
is
et
a
l.
[
8
]
in
tr
o
d
u
ce
d
a
m
o
r
e
g
en
er
al
clas
s
if
icatio
n
s
y
s
te
m
t
h
at
g
r
o
u
p
s
th
e
f
lo
w
s
u
s
i
n
g
h
o
s
t
an
d
p
o
r
t
in
f
o
r
m
atio
n
t
h
en
i
n
d
iv
id
u
al
f
lo
w
s
,
th
is
s
y
s
te
m
w
o
r
k
s
o
n
p
a
y
lo
a
d
d
etails,
f
lo
w
s
ar
e
tr
ain
ed
b
y
p
a
y
lo
ad
,
it g
i
v
es les
s
ac
cu
r
ac
y
,
i
f
p
o
r
t a
n
d
h
o
s
t i
n
f
o
r
m
at
io
n
is
u
n
a
v
ailab
le.
R
o
u
g
h
a
n
et
al.
[
9
]
class
if
ied
t
h
e
tr
af
f
ic
in
to
a
n
u
n
d
er
s
ized
n
u
m
b
er
o
f
t
y
p
es
ap
t
f
o
r
Qo
s
ap
p
licatio
n
s
.
T
h
ey
u
s
e
s
y
s
te
m
s
u
c
h
as
cl
u
s
ter
in
g
u
s
i
n
g
n
ea
r
est
n
ei
g
h
b
o
r
to
g
iv
e
t
h
e
n
ec
es
s
ar
y
ar
r
an
g
e
m
en
t.
So
m
e
d
eg
r
ee
o
f
s
m
all
s
et
o
f
f
ea
t
u
r
es
a
n
d
a
n
i
m
p
licit
ass
u
m
p
tio
n
o
f
t
h
e
ac
cu
r
ac
y
o
f
t
h
e
te
s
ti
n
g
an
d
tr
ain
i
n
g
d
ata
-
s
ets;
th
e
au
th
o
r
s
r
estrict
t
h
e
m
s
el
v
es
l
ar
g
e
s
ets
o
f
n
et
w
o
r
k
-
b
ased
ap
p
licatio
n
s
.
Af
za
l
et
al.
[
1
0
]
co
m
p
ar
ed
th
e
p
er
f
o
r
m
a
n
ce
o
f
t
w
o
v
ar
iat
i
o
n
s
o
f
b
ac
k
p
r
o
p
ag
atio
n
le
ar
n
in
g
al
g
o
r
it
h
m
(
A
d
ap
tiv
e
lear
n
i
n
g
r
ate
w
i
th
m
o
m
e
n
t
u
m
an
d
R
esil
ie
n
t)
.
B
o
th
th
e
al
g
o
r
ith
m
s
ar
e
ex
p
er
i
m
en
ted
o
n
a
v
ar
iet
y
o
f
cla
s
s
i
f
icatio
n
p
r
o
b
lem
s
i
n
o
r
d
er
to
ass
ess
t
h
e
e
f
f
ic
ien
c
y
o
f
th
e
s
e
t
w
o
lear
n
i
n
g
ap
p
r
o
ac
h
es.
E
x
p
er
i
m
en
tal
r
esu
lts
r
e
v
e
al
th
at
d
u
r
i
n
g
test
i
n
g
an
d
tr
ai
n
in
g
R
es
ilie
n
t
p
r
o
p
ag
atio
n
al
g
o
r
ith
m
o
u
tp
er
f
o
r
m
s
b
ac
k
p
r
o
p
ag
atio
n
w
it
h
A
d
ap
ti
v
e
lear
n
i
n
g
r
ate
a
n
d
m
o
m
e
n
t
u
m
.
He
m
alat
h
a
et
al.
[
1
1
]
ex
p
r
es
s
ed
th
e
p
r
iv
ac
y
-
p
r
eser
v
i
n
g
m
eth
o
d
s
w
a
s
v
u
l
n
er
ab
le
to
S
y
b
il
attac
k
s
,
w
h
er
eb
y
a
m
alicio
u
s
u
s
er
ca
n
ac
t
as
to
b
e
m
u
ltip
le
v
e
h
ic
les.
I
t
e
n
s
u
r
ed
t
h
e
a
u
t
h
en
ticit
y
o
f
t
h
e
m
e
s
s
a
g
es
p
r
o
p
ag
ated
in
V
A
NE
T
;
a
s
t
r
aig
h
t
-
f
o
r
w
ar
d
p
r
o
ce
s
s
is
u
ti
lized
p
u
b
lic
k
e
y
s
a
u
t
h
o
r
ized
b
y
a
ce
r
tific
atio
n
au
th
o
r
it
y
(
C
A
)
to
s
ig
n
t
h
e
m
e
s
s
a
g
es.
Fad
li
et
a
l.
[
1
2
]
d
ev
elo
p
ed
an
E
th
er
n
et
o
v
er
SDH
(
s
y
n
ch
r
o
n
o
u
s
d
i
g
ital
h
ier
ar
ch
y
)
a
n
d
MP
L
S
-
T
P
(
Mu
ltip
r
o
to
co
l
L
ab
el
S
w
itc
h
i
n
g
-
T
r
an
s
p
o
r
t
Pro
f
ile)
w
it
h
r
in
g
s
ec
u
r
i
t
y
to
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
I
n
tellig
en
t P
a
ck
et
Delive
r
y
in
R
o
u
ter Usi
n
g
S
tr
u
ctu
r
e
Op
timiz
ed
N
eu
r
a
l Net
w
o
r
k
(
R
.
Dee
b
a
la
ksh
mi
)
547
p
r
ed
ictab
le
n
et
w
o
r
k
f
ail
u
r
e
an
d
co
m
p
u
ted
p
er
f
o
r
m
a
n
ce
o
f
n
et
w
o
r
k
.
Am
an
et
al.
[
1
3
]
in
te
g
r
ated
co
n
tex
t
tr
an
s
p
o
r
t
an
d
m
u
lticas
t
q
u
ic
k
r
er
o
u
te,
a
n
d
d
ev
elo
p
ed
th
i
s
in
teg
r
atio
n
to
t
h
e
s
tan
d
ar
d
n
et
w
o
r
k
m
o
b
ilit
y
m
an
a
g
e
m
e
n
t.
Ha
s
h
i
m
et
al.
[
1
4
]
d
is
cu
s
s
ed
q
u
an
t
itati
v
e
a
n
al
y
s
i
s
a
n
d
it
co
m
p
u
ted
P
r
o
x
y
Mo
b
ile
I
n
ter
n
et
P
r
o
to
c
o
l V
er
s
io
n
6
(
P
M
I
P
v
6
)
m
u
lticas
t sp
ee
d
y
r
er
o
u
te
o
p
er
atio
n
s
f
o
r
d
ata
tr
af
f
ic.
R
o
u
ter
r
o
u
tes
t
h
e
p
ac
k
ets
f
r
o
m
o
n
e
n
et
w
o
r
k
to
o
th
er
n
et
wo
r
k
th
r
o
u
g
h
th
e
b
est
p
ath
.
T
o
r
o
u
te
th
e
p
ac
k
et
r
o
u
ter
m
u
s
t
k
n
o
w
th
e
f
o
llo
w
in
g
d
etail
s
tar
g
et
ad
d
r
ess
,
n
eig
h
b
o
r
r
o
u
ter
f
r
o
m
w
h
ic
h
it
ca
n
lear
n
ab
o
u
t
r
e
m
o
te
n
e
t
w
o
r
k
s
p
o
s
s
ib
le
r
o
u
t
es to
all
r
e
m
o
te
n
et
w
o
r
k
s
.
T
h
e
b
est r
o
u
te
is
ap
p
lied
to
ea
ch
r
e
m
o
te
n
et
w
o
r
k
an
d
h
o
w
to
m
ai
n
tain
an
d
v
er
i
f
y
r
o
u
tin
g
i
n
f
o
r
m
atio
n
.
T
h
e
r
o
u
t
er
lear
n
s
ab
o
u
t
r
e
m
o
te
n
et
w
o
r
k
s
f
r
o
m
n
ei
g
h
b
o
r
r
o
u
ter
o
r
f
r
o
m
a
n
ad
m
i
n
i
s
tr
at
o
r
,
th
e
r
o
u
ter
th
e
n
b
u
ild
s
a
r
o
u
ti
n
g
tab
le
th
a
t
d
escr
ib
es
h
o
w
to
f
i
n
d
th
e
r
e
m
o
te
n
et
w
o
r
k
,
i
f
a
n
et
w
o
r
k
i
s
d
ir
ec
tl
y
co
n
n
ec
ted
th
e
n
th
e
r
o
u
ter
alr
ea
d
y
k
n
o
w
s
h
o
w
to
g
et
to
i
t.
I
f
a
n
et
w
o
r
k
is
n
’
t
d
ir
ec
tl
y
co
n
n
ec
ted
to
th
e
r
o
u
t
er
,
th
e
r
o
u
ter
u
s
e
t
w
o
w
a
y
s
t
o
r
ea
ch
r
e
m
o
te
n
et
w
o
r
k
o
n
e
i
s
s
ta
tic
r
o
u
t
in
g
a
n
d
o
th
er
is
d
y
n
a
m
ic
r
o
u
tin
g
.
T
h
e
Net
w
o
r
k
tr
af
f
ic
clas
s
i
f
ic
atio
n
is
r
eq
u
ir
ed
m
ai
n
l
y
f
o
r
m
an
y
n
et
w
o
r
k
m
an
a
g
e
m
en
t
t
ask
s
.
Flo
w
p
r
io
r
itizatio
n
T
r
af
f
ic
s
h
ap
i
n
g
/p
o
licin
g
d
ia
g
n
o
s
tic
m
o
n
i
t
o
r
in
g
.
A
l
s
o
t
h
e
n
et
w
o
r
k
o
p
er
ato
r
s
ca
n
k
n
o
w
k
n
o
w
led
g
e
o
f
w
h
at
ap
p
licatio
n
s
ar
e
p
r
ese
n
t
i
n
a
n
et
w
o
r
k
,
n
et
w
o
r
k
-
o
p
er
ato
r
s
ar
e
s
u
p
er
io
r
to
p
lan
,
b
u
d
g
et
a
n
d
b
ill.
Net
w
o
r
k
-
o
p
er
ato
r
s
an
d
u
s
er
s
ab
le
to
id
e
n
ti
f
y
ab
n
o
r
m
alit
y
in
n
e
t
w
o
r
k
tr
a
f
f
ic
to
r
ed
u
ce
t
h
e
i
m
p
ac
t
o
f
m
alicio
u
s
b
e
h
av
io
r
.
B
esid
e
class
i
f
icatio
n
o
f
n
et
w
o
r
k
tr
a
f
f
ic
s
ar
e
t
h
o
s
e
t
h
at
d
e
f
i
n
e
to
u
n
d
er
s
tan
d
t
h
e
p
r
ese
n
t
an
d
v
ar
y
i
n
g
I
n
ter
n
et:
t
h
e
m
o
d
eli
n
g
o
f
tr
af
f
ic
-
m
ix
,
u
s
er
-
co
m
p
o
s
itio
n
an
d
s
o
o
n
.
I
n
n
e
u
r
al
n
e
t
w
o
r
k
class
i
f
icatio
n
d
ep
en
d
w
it
h
o
u
t
k
n
o
w
i
n
g
s
i
te
d
etail,
w
h
ich
m
ac
h
in
e
u
s
ed
,
w
i
th
o
u
t
in
ter
a
ctio
n
o
f
u
s
er
an
d
ad
m
in
i
s
tr
ato
r
an
d
w
it
h
o
u
t p
ac
k
et
p
a
y
lo
ad
in
f
o
r
m
atio
n
li
k
e
I
P
ad
d
r
ess
an
d
p
o
r
t d
etail.
T
h
is
p
ap
er
s
u
m
m
ar
ize
as
f
o
ll
o
w
s
:
Sectio
n
2
d
is
cu
s
s
ed
a
b
o
u
t
p
r
o
p
o
s
ed
m
et
h
o
d
.
Sectio
n
3
ex
p
lo
r
es
ab
o
u
t
i
m
p
le
m
e
n
ted
r
esu
lt
a
n
d
Dis
cu
s
s
io
n
.
Sectio
n
4
co
n
clu
d
es
th
e
o
v
er
all
w
o
r
k
w
it
h
f
u
t
u
r
e
p
lan
o
f
r
esear
ch
w
o
r
k
.
2.
P
RO
P
O
SE
D
SYS
T
E
M
Neu
r
al
n
et
w
o
r
k
i
s
a
co
n
ce
p
t
u
al
m
o
d
el
d
esi
g
n
ed
as
a
co
m
p
u
tatio
n
al
m
o
d
el
b
ased
o
n
h
u
m
a
n
b
r
ain
to
s
o
lv
e
ce
r
tai
n
k
i
n
d
o
f
p
r
o
b
le
m
s
,
it
’
s
a
n
et
w
o
r
k
o
f
m
a
n
y
n
e
u
r
o
n
s
a
n
d
p
er
f
o
r
m
s
in
te
llig
e
n
t
b
eh
av
io
r
,
a
n
d
n
eu
r
a
l
n
et
w
o
r
k
h
as
ab
ilit
y
to
lear
n
an
ad
ap
tiv
e
s
y
s
te
m
w
h
ic
h
m
ea
n
s
it
ca
n
ch
a
n
g
e
n
et
w
o
r
k
s
tr
u
ct
u
r
e
b
ased
o
n
in
f
o
r
m
atio
n
f
lo
w
i
n
g
th
r
o
u
g
h
it
.
A
d
a
p
tatio
n
is
d
o
n
e
b
y
ad
j
u
s
ti
n
g
w
e
ig
h
ts
o
f
n
et
w
o
r
k
.
Fig
u
r
e
1
.
Sa
m
p
le
A
r
ch
itec
tu
r
e
of
Neu
r
al
Net
w
o
r
k
I
n
Fig
u
r
e
1
cir
cle
r
ep
r
esen
ts
t
h
e
n
e
u
r
o
n
s
a
n
d
lin
e
s
i
n
d
icate
s
co
n
n
ec
tio
n
b
et
w
ee
n
t
w
o
n
e
u
r
o
n
s
also
it
is
a
p
at
h
f
o
r
f
lo
w
o
f
i
n
f
o
r
m
ati
o
n
,
ea
ch
li
n
e
h
as
w
e
i
g
h
t
it
’
s
a
n
u
m
b
er
t
h
at
co
n
tr
o
l
ac
tio
n
b
et
w
ee
n
t
w
o
n
e
u
r
o
n
s
.
I
f
t
h
e
n
e
u
r
al
n
et
w
o
r
k
g
i
v
es
g
o
o
d
o
u
tp
u
t
n
o
n
ee
d
to
ad
j
u
s
t
t
h
e
w
e
ig
h
ts
,
o
t
h
er
w
is
e
er
r
o
r
g
en
er
ated
b
y
n
et
w
o
r
k
.
So
th
e
w
eig
h
t
s
ar
e
ad
j
u
s
ted
to
g
et
g
o
o
d
o
u
tp
u
t.
A
p
er
ce
p
tr
o
n
co
n
ta
in
s
o
n
e
o
r
m
o
r
e
i
n
p
u
t
s
o
r
a
p
r
o
ce
s
s
o
r
an
d
p
r
o
d
u
ce
th
e
s
in
g
le
o
u
tp
u
t,
it
f
o
llo
w
s
t
h
e
f
ee
d
f
o
r
w
ar
d
m
o
d
el
w
h
ich
m
ea
n
s
i
n
p
u
t
s
ar
e
s
en
t
to
n
e
u
r
o
n
s
ar
e
p
r
o
ce
s
s
ed
to
g
et
o
u
tp
u
t.
P
er
ce
p
tr
o
n
alg
o
r
ith
m
Step
1
: a
ll in
p
u
t
s
ar
e
m
u
ltip
lie
d
b
y
w
ei
g
h
ts
Step
2
: su
m
a
ll th
e
w
e
ig
h
ted
i
n
p
u
t
s
Step
3
: c
o
m
p
u
te
th
e
o
u
tp
u
t b
y
s
u
m
p
ass
ed
t
h
r
o
u
g
h
a
n
ac
tiv
at
io
n
f
u
n
c
tio
n
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4
7
5
2
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
10
,
No
.
2
,
Ma
y
2
0
1
8
:
5
4
5
–
553
548
2
.
1
.
B
a
ck
P
ro
pa
g
a
t
io
n
w
it
h
Neura
l N
et
w
o
rk
(
B
P
NN)
T
h
e
B
P
NN
is
a
co
m
m
o
n
m
e
t
h
o
d
f
o
r
tr
ain
i
n
g
t
h
e
ar
tific
ial
n
eu
r
al
n
et
w
o
r
k
s
i
n
co
n
j
u
n
ctio
n
w
it
h
a
n
o
p
tim
izatio
n
m
e
th
o
d
,
s
u
ch
as
th
e
g
r
ad
ien
t
d
escen
t.
T
h
e
m
et
h
o
d
ca
lcu
late
s
t
h
e
g
r
ad
ien
t
o
f
a
lo
s
s
f
u
n
c
tio
n
w
it
h
r
esp
ec
ts
to
all
t
h
e
w
ei
g
h
ts
i
n
t
h
e
n
et
w
o
r
k
.
T
h
e
g
r
a
d
ien
t i
s
f
e
d
to
th
e
o
p
ti
m
izat
io
n
m
et
h
o
d
w
h
ic
h
i
n
tu
r
n
u
s
e
s
it
to
u
p
d
ate
th
e
w
ei
g
h
ts
,
in
a
n
at
te
m
p
t
to
m
i
n
i
m
ize
t
h
e
lo
s
s
f
u
n
ctio
n
.
B
ac
k
p
r
o
p
ag
atio
n
(
B
P
)
r
eq
u
ir
es
a
k
n
o
w
n
,
d
esire
d
o
u
tp
u
t
f
o
r
ea
ch
in
p
u
t
v
alu
e
in
o
r
d
er
to
ca
lcu
late
th
e
lo
s
s
f
u
n
ctio
n
g
r
ad
ien
t.
I
t
is
th
er
ef
o
r
e
u
s
u
all
y
co
n
s
id
er
ed
to
b
e
a
s
u
p
er
v
is
ed
lear
n
in
g
m
e
th
o
d
.
A
lt
h
o
u
g
h
,
it
is
al
s
o
u
s
ed
f
o
r
s
o
m
e
u
n
s
u
p
er
v
is
ed
n
et
w
o
r
k
.
I
t
i
s
a
g
en
er
aliza
tio
n
o
f
th
e
d
elta
r
u
le
to
m
u
lti
-
la
y
er
ed
f
ee
d
f
o
r
w
ar
d
n
et
w
o
r
k
s
,
m
ad
e
p
o
s
s
ib
l
e
b
y
u
s
i
n
g
t
h
e
ch
ai
n
r
u
le
to
iter
at
i
v
el
y
co
m
p
u
te
g
r
ad
ien
ts
f
o
r
ea
ch
la
y
er
.
T
h
e
B
P
r
eq
u
ir
es
th
at
th
e
ac
tiv
a
tio
n
f
u
n
ctio
n
u
s
ed
b
y
th
e
ar
tif
icial
n
eu
r
o
n
s
(
o
r
“
n
o
d
es”)
b
e
d
if
f
er
en
t
iab
le.
Giv
e
n
th
e
n
u
m
b
er
o
f
n
o
d
es
o
f
th
e
in
p
u
t
la
y
er
,
th
e
h
id
d
en
an
d
th
e
o
u
tp
u
t
n
et
w
o
r
k
n
,
k
,
m
,
r
esp
ec
tiv
el
y
,
t
h
e
to
tal
n
u
m
b
er
o
f
in
p
u
t
s
a
m
p
les
is
pi
x
,
w
h
ic
h
in
d
icate
s
th
at
th
e
P
s
a
m
p
le’
s
th
e
ith
in
p
u
t
v
a
lu
e,
ki
v
in
d
icate
s
th
e
ith
n
o
d
e
o
f
in
p
u
t
la
y
er
to
t
h
e
h
id
d
en
la
y
er
o
f
t
h
e
k
t
h
n
o
d
e
w
ei
g
h
t,
jk
m
ea
n
s
t
h
e
n
o
d
e
w
ei
g
h
t
f
r
o
m
h
id
d
en
la
y
er
o
f
th
e
k
to
th
e
o
u
tp
u
t
la
y
er
o
f
th
e
j
.
Fo
r
c
o
n
v
e
n
ien
ce
,
t
h
e
th
r
es
h
o
ld
is
in
clu
d
ed
co
n
n
ec
t
io
n
w
ei
g
h
ts
,
a
n
d
th
e
n
th
e
o
u
tp
u
t o
f
h
id
d
en
la
y
er
n
o
d
e
k
i
s
ex
p
r
es
s
ed
in
eq
u
atio
n
(
1
)
.
0
n
p
k
p
k
k
i
p
i
i
z
f
n
e
t
f
v
x
(
1
)
T
h
e
o
u
tp
u
t la
y
er
n
o
d
es
f
o
r
th
e
n
o
d
e
j
:
is
ex
p
r
ess
ed
in
eq
u
ati
o
n
(
2
)
.
0
n
p
j
p
j
j
k
p
k
i
y
f
n
e
t
f
w
z
(
2
)
w
h
er
e,
s
ta
n
d
ar
d
s
ig
m
o
id
f
u
n
ct
io
n
is
s
elec
ted
as i
n
ce
n
ti
v
e
f
u
n
ctio
n
ar
e
g
i
v
en
eq
u
a
tio
n
(
3
)
.
1
()
x
fx
e
(
3
)
T
h
e
d
ef
in
itio
n
o
f
g
lo
b
al
er
r
o
r
f
u
n
ctio
n
s
ca
n
b
e
ex
p
r
ess
ed
in
eq
u
atio
n
(
4
)
.
2
1
1
1
1
()
2
pp
m
p
p
j
p
j
p
p
j
E
E
t
y
(
4
)
w
h
er
e
p
E
,
is
er
r
o
r
o
f
th
e
s
a
m
p
le
p
,
pj
t
is
th
e
id
ea
l
o
u
tp
u
t.
T
h
e
ad
j
u
s
t
m
e
n
t
f
o
r
m
u
la
s
o
f
w
e
ig
h
ts
ar
e
as
f
o
llo
w
s
.
a.
T
h
e
w
e
ig
h
t
ad
j
u
s
t
m
en
t
f
o
r
m
u
la
o
f
o
u
tp
u
t
la
y
er
n
eu
r
o
n
s
w
h
ic
h
m
a
th
e
m
atica
l
e
x
p
r
ess
io
n
ar
e
g
iv
e
n
i
n
E
q
u
atio
n
(
5
)
1
(
)
.
(
1
)
.
p
j
k
p
j
p
j
p
j
p
j
p
k
p
t
y
y
y
z
(
5
)
w
h
er
e,
η
is
t
h
e
lear
n
i
n
g
r
ate,
w
h
o
s
e
g
e
n
er
al
r
an
g
e
is
0
.
1
-
0
.
3
.
b
.
T
h
e
w
ei
g
h
t a
d
j
u
s
t
m
e
n
t
f
o
r
m
u
la
o
f
ea
ch
h
id
d
e
n
la
y
er
n
e
u
r
o
n
s
ar
e
ex
p
lain
ed
i
n
eq
u
atio
n
(
6
)
11
(
1
)
p
m
k
i
p
i
j
k
p
k
p
k
p
i
pj
v
z
z
x
(
6
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
I
n
tellig
en
t P
a
ck
et
Delive
r
y
in
R
o
u
ter Usi
n
g
S
tr
u
ctu
r
e
Op
timiz
ed
N
eu
r
a
l Net
w
o
r
k
(
R
.
Dee
b
a
la
ksh
mi
)
549
T
h
e
b
asic
id
ea
o
f
th
e
B
P
alg
o
r
ith
m
is
t
h
e
lear
n
i
n
g
p
r
o
ce
s
s
an
d
ca
n
b
e
d
iv
id
ed
in
to
t
w
o
s
tag
es:
t
h
e
f
ir
s
t
s
ta
g
e
(
f
o
r
w
ar
d
p
r
o
p
ag
ati
o
n
p
r
o
ce
s
s
)
,
g
i
v
e
n
i
n
p
u
t
in
f
o
r
m
atio
n
t
h
r
o
u
g
h
a
la
y
er
b
y
la
y
er
p
r
o
ce
s
s
in
g
ea
ch
h
id
d
en
la
y
er
an
d
ca
lc
u
late
t
h
e
ac
tu
al
o
u
tp
u
t
v
alu
e
o
f
ea
c
h
u
n
it
o
f
pj
y
;
th
e
s
ec
o
n
d
s
tag
e
(
t
h
e
r
e
v
er
s
e
p
r
o
ce
s
s
)
,
if
t
h
e
o
u
tp
u
t
la
y
er
f
ai
ls
to
g
et
th
e
e
x
p
ec
ted
o
u
tp
u
t
v
al
u
e
,
th
en
la
y
er
b
y
la
y
er
r
ec
u
r
s
i
v
el
y
ca
lcu
la
tes
t
h
e
d
if
f
er
e
n
ce
o
f
er
r
o
r
b
etw
ee
n
th
e
ac
tu
al
o
u
tp
u
t
a
n
d
th
e
ex
p
ec
t
ed
o
u
tp
u
t.
T
h
e
g
r
ad
ien
t
d
escen
t
m
et
h
o
d
m
o
d
if
ie
s
th
e
w
eig
h
t
s
o
f
th
e
ki
v
,
jk
,
m
ak
i
n
g
th
e
to
tal
er
r
o
r
f
u
n
ctio
n
m
i
n
i
m
u
m
.
T
ab
le
1
d
is
p
lay
s
t
h
e
in
p
u
t
p
ar
am
eter
s
o
f
n
et
w
o
r
k
tr
af
f
ic
t
o
ev
alu
ate
t
h
e
p
r
o
p
o
s
ed
tech
n
i
q
u
es.
T
ab
le
1
Sam
p
le
I
n
p
u
t P
ar
a
m
et
er
s
o
f
Net
w
o
r
k
T
r
af
f
ic
C
l
a
ssi
f
i
c
a
t
i
o
n
Ex
a
mp
l
e
A
p
p
l
i
c
a
t
i
o
n
D
A
TA
B
A
S
E
p
o
st
g
r
e
s,
sq
l
n
e
t
o
r
a
c
l
e
,
i
n
g
r
e
s
I
N
T
ER
A
C
TI
V
E
ssh
,
k
l
o
g
i
n
,
r
l
o
g
i
n
,
t
e
l
n
e
t
M
A
I
L
i
ma
p
,
p
o
p
2
/
3
,
smt
p
S
ER
V
I
C
ES
X
1
1
,
d
n
s
,
i
d
e
n
t
,
l
d
a
p
,
n
t
p
A
TTA
C
K
I
n
t
e
r
n
e
t
w
o
r
m
a
n
d
v
i
r
u
s a
t
t
a
c
k
s
G
A
M
ES
M
i
c
r
o
so
f
t
D
i
r
e
c
t
P
l
a
y
P
2
P
K
a
Z
a
A
,
B
i
t
T
o
r
r
e
n
t
,
G
n
u
T
e
l
l
a
M
U
L
TI
M
ED
I
A
W
i
n
d
o
w
s M
e
d
i
a
P
l
a
y
e
r
,
R
e
a
l
B
U
L
K
f
t
p
WWW
www
2
.
2
.
F
lo
w
F
ea
t
ures
-
Cla
s
s
if
ica
t
io
n inp
ut
Flo
w
m
etr
ics
(
d
u
r
atio
n
,
p
ac
k
et
-
co
u
n
t,
to
tal
b
y
te
s
)
P
ac
k
et
i
n
ter
-
ar
r
i
v
al
t
i
m
e
(
m
ea
n
,
v
ar
ia
n
ce
,
1
s
t
&
3
r
d
q
u
ar
tiles
,
m
ed
ia
n
,
m
i
n
i
m
u
m
,
m
ax
i
m
u
m
.
Size
o
f
T
C
P
/I
P
co
n
tr
o
l
f
ield
s
(
m
ea
n
,
v
ar
ian
ce
,
1
s
t
&
3
r
d
q
u
ar
tiles
,
m
ed
ian
,
m
in
i
m
u
m
,
m
ax
i
m
u
m
,
)
.
T
o
tal
p
ac
k
ets
(
i
n
ea
ch
d
ir
ec
tio
n
an
d
to
tal
f
o
r
f
lo
w
)
P
a
y
lo
ad
s
ize
(
m
ea
n
,
v
ar
ian
ce
,
1
s
t
&
3
r
d
q
u
ar
tiles
,
m
ed
ia
n
,
m
in
i
m
u
m
,
m
a
x
i
m
u
m
,
.
.
.
)
.
E
f
f
ec
ti
v
e
B
an
d
w
id
t
h
R
a
n
k
ed
lis
t
is
to
p
-
ten
f
o
u
r
ier
-
tr
a
n
s
f
o
r
m
co
m
p
o
n
en
ts
o
f
p
ac
k
et
i
n
ter
-
ar
r
iv
al
ti
m
es
f
o
r
ea
ch
d
ir
ec
tio
n
.
Her
e,
T
ab
le
2
ex
p
lain
s
th
e
p
r
o
p
o
s
ed
class
if
ier
r
esu
lt o
f
p
ac
k
et
i
n
ter
-
ar
r
i
v
al
ti
m
es i
n
d
etails.
T
ab
le
2
P
r
o
p
o
s
ee
d
C
lass
i
f
ier
R
es
u
lt
of
P
ac
k
et
I
n
ter
-
A
r
r
i
v
al
T
im
es
A
l
g
o
r
i
t
h
ms
A
c
c
u
r
a
c
y
NN
-
BP
9
6
.
8
7
NN
-
B
P
w
i
t
h
O
p
t
i
mi
z
a
t
i
o
n
9
8
.
0
7
NN
-
S
t
r
u
c
t
u
r
e
o
p
t
i
mi
z
e
d
9
9
.
0
5
Fig
u
r
e
2
.
C
o
m
p
ar
is
o
n
An
al
y
s
i
s
o
f
Neu
r
al
Net
w
o
r
k
A
l
g
o
r
ith
m
s
Fig
u
r
e
2
s
h
o
w
s
th
e
co
m
p
ar
i
s
o
n
o
f
Ne
u
r
al
Net
w
o
r
k
Alg
o
r
ith
m
s
s
u
ch
a
s
NN
-
B
P
,
NN
-
B
P
w
it
h
o
p
tim
izatio
n
a
n
d
NN
-
s
tr
u
ctu
r
ed
o
p
tim
izat
io
n
an
d
it
i
s
clea
r
l
y
s
h
o
w
n
t
h
at
t
h
e
p
r
o
p
o
s
ed
m
ec
h
a
n
i
s
m
NN
-
B
P
h
as
h
ig
h
er
ac
cu
r
ac
y
o
f
9
9
.
0
5
e
s
ti
m
ated
in
ter
m
s
o
f
p
ac
k
e
t a
r
r
iv
al
ti
m
es.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4
7
5
2
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
10
,
No
.
2
,
Ma
y
2
0
1
8
:
5
4
5
–
553
550
2
.
3
.
I
m
po
rt
a
nt
f
lo
w
Ro
uting
I
n
g
e
n
er
al
r
o
u
ter
r
o
u
te
t
h
e
p
a
ck
ets
f
r
o
m
o
n
e
d
ev
ice
to
o
t
h
e
r
d
ev
ice
w
h
ic
h
ar
e
i
n
d
i
f
f
er
e
n
t
n
et
w
o
r
k
,
f
o
r
f
o
r
w
ar
d
i
n
g
p
ac
k
e
ts
it
r
e
f
er
s
t
h
e
r
o
u
ti
n
g
tab
le,
r
o
u
ti
n
g
tab
le
co
n
ta
in
s
d
etails
ab
o
u
t
n
ea
r
b
y
r
o
u
ter
s
in
f
o
r
m
atio
n
.
Fo
r
i
m
p
o
r
tan
t
f
l
o
w
r
o
u
ti
n
g
,
s
o
m
e
ad
d
itio
n
al
in
f
o
r
m
atio
n
is
u
s
ed
to
s
to
r
e
i
n
r
o
u
ti
n
g
tab
le.
I
n
r
o
u
tin
g
tab
le
I
P
ad
d
r
ess
alo
n
g
w
it
h
p
ac
k
et
co
u
n
t
f
o
r
p
ar
ticu
lar
ap
p
licatio
n
an
d
p
ac
k
et
tim
e
s
ta
m
p
is
u
p
d
ated
in
r
o
u
t
in
g
tab
le
p
er
io
d
icall
y
.
Flo
w
m
ea
s
u
r
e
m
e
n
t
s
tati
s
ti
cs
r
ep
o
r
t
p
r
o
v
id
es
p
ac
k
et
c
o
u
n
t
f
o
r
p
ar
ticu
lar
ap
p
licatio
n
.
R
o
u
ter
r
o
u
te
p
ac
k
ets in
te
lli
g
en
tl
y
u
s
i
n
g
th
i
s
clas
s
if
ica
tio
n
o
u
tp
u
t.
Fig
u
r
e
3
.
I
m
p
o
r
tan
t
Flo
w
R
o
u
tin
g
I
n
Fi
g
u
r
e
3
,
n
et
w
o
r
k
1
w
a
n
t
s
to
tr
an
s
f
er
t
h
e
p
ac
k
et
to
n
et
w
o
r
k
2
.
T
h
e
r
o
u
ter
r
o
u
te
th
e
p
ac
k
e
t
in
telli
g
e
n
tl
y
u
s
in
g
tr
af
f
ic
clas
s
if
icatio
n
o
u
tp
u
t.
I
n
r
o
u
ter
p
r
io
r
it
y
s
c
h
ed
u
li
n
g
al
g
o
r
ith
m
is
u
s
ed
to
s
ch
ed
u
le
th
e
p
ac
k
et,
b
ased
o
n
p
r
io
r
ity
p
ac
k
ets i
m
p
o
r
tan
ce
i
s
g
i
v
e
n
th
e
n
it
is
f
o
r
w
ar
d
ed
to
r
esp
ec
tiv
e
d
esti
n
atio
n
.
Fig
u
r
e
4
.
R
o
u
ti
n
g
T
ab
le
Up
d
a
t
at
io
n
R
o
u
ter
in
co
m
i
n
g
p
ac
k
ets
ar
e
m
ea
s
u
r
ed
an
d
class
i
f
ied
b
ase
d
o
n
ca
teg
o
r
izatio
n
alg
o
r
it
h
m
,
f
lo
w
ar
e
f
o
r
m
ed
as
f
lo
w
-
g
r
o
u
p
s
.
T
h
is
f
lo
w
-
g
r
o
u
p
s
i
n
f
o
r
m
atio
n
is
m
ai
n
tai
n
ed
b
y
co
u
n
ter
,
f
o
r
ea
ch
m
a
tch
i
n
g
f
lo
w
-
g
r
o
u
p
s
co
u
n
ter
i
s
i
n
cr
e
m
e
n
te
d
;
th
e
ap
p
licatio
n
p
ac
k
et
co
u
n
t
alo
n
g
w
it
h
t
h
e
p
ac
k
et
ti
m
e
s
ta
m
p
is
u
p
d
ated
in
r
o
u
tin
g
tab
le
p
er
io
d
ically
as
s
h
o
w
n
in
Fi
g
u
r
e
4
.
R
o
u
ter
w
h
ile
r
o
u
te
t
h
e
p
ac
k
et
ch
ec
k
h
o
w
to
r
ea
ch
t
h
e
d
esti
n
atio
n
ad
d
r
ess
,
an
d
al
s
o
k
n
o
w
s
t
h
e
ap
p
licatio
n
d
etail,
w
h
ic
h
p
ac
k
et
b
elo
n
g
s
to
w
h
i
ch
ap
p
licatio
n
a
n
d
h
o
w
m
a
n
y
p
ac
k
et
s
f
o
r
w
ar
d
ed
f
o
r
p
ar
ticu
lar
ap
p
licatio
n
,
b
as
ed
o
n
th
e
i
n
telli
g
e
n
t
i
n
f
o
r
m
ati
o
n
r
o
u
te
th
e
p
ac
k
et
in
telli
g
e
n
tl
y
.
I
f
s
o
m
e
ap
p
licat
io
n
p
ac
k
ets
co
u
n
ts
ar
e
f
o
r
w
ar
d
ed
m
o
r
e
an
d
o
th
er
ap
p
licati
o
n
p
ac
k
ets
ar
e
j
u
s
t
s
tar
ted
f
o
r
tr
a
n
s
m
i
s
s
io
n
,
r
o
u
te
r
f
o
llo
w
s
t
h
e
p
r
io
r
it
y
a
lg
o
r
it
h
m
f
ir
s
t p
r
io
r
it
y
is
g
i
v
e
n
f
o
r
ap
p
licatio
n
w
h
ic
h
h
a
v
e
m
o
r
e
p
ac
k
et
s
co
u
n
t,
o
th
er
ap
p
licatio
n
tr
an
s
m
i
s
s
io
n
i
s
d
e
n
ie
d
,
th
is
d
en
ia
l
o
f
s
er
v
ice
is
g
i
v
en
o
n
l
y
t
h
r
ee
ti
m
es
f
o
r
p
ar
ticu
lar
ap
p
licatio
n
u
s
i
n
g
ag
i
n
g
co
n
ce
p
t.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
I
n
tellig
en
t P
a
ck
et
Delive
r
y
in
R
o
u
ter Usi
n
g
S
tr
u
ctu
r
e
Op
timiz
ed
N
eu
r
a
l Net
w
o
r
k
(
R
.
Dee
b
a
la
ksh
mi
)
551
3.
RE
SU
L
T
S
A
ND
AN
AL
Y
SI
S
Fig
u
r
e
5
.
R
o
u
ter
p
er
f
o
r
m
an
ce
w
it
h
o
u
t in
te
lli
g
en
ce
R
o
u
ter
p
er
f
o
r
m
a
n
ce
w
it
h
o
u
t
i
n
telli
g
e
n
ce
is
s
h
o
w
n
in
Fig
u
r
e
5
.
Her
e
th
e
p
ar
am
eter
s
s
u
c
h
as
s
p
ee
d
,
laten
c
y
a
n
d
th
r
o
u
g
h
p
u
t a
r
e
p
lo
tted
f
o
r
ex
is
ti
n
g
s
y
s
te
m
w
it
h
r
esp
ec
t to
th
eir
n
et
w
o
r
k
e
f
f
icie
n
c
y
.
Fig
u
r
e
6
.
R
o
u
ter
p
er
f
o
r
m
an
ce
w
it
h
i
n
telli
g
e
n
ce
R
o
u
ter
p
er
f
o
r
m
an
ce
w
it
h
in
t
ellig
e
n
ce
i
s
s
h
o
w
n
in
Fi
g
u
r
e
6
.
Her
e
t
h
e
p
ar
a
m
e
ter
s
s
u
c
h
as
s
p
ee
d
,
laten
c
y
an
d
t
h
r
o
u
g
h
p
u
t
ar
e
p
lo
tted
f
o
r
p
r
o
p
o
s
ed
s
y
s
te
m
w
it
h
r
esp
ec
t
to
th
eir
n
et
w
o
r
k
e
f
f
icien
c
y
.
Ach
ie
v
i
n
g
h
ig
h
er
t
h
r
o
u
g
h
p
u
t
s
h
o
w
s
t
h
at
b
etter
p
er
f
o
r
m
a
n
ce
o
f
t
h
e
n
e
t
w
o
r
k
.
T
ab
le
3
ex
p
r
ess
es
t
h
e
p
r
o
p
o
s
ed
alg
o
r
ith
m
p
er
f
o
r
m
a
n
ce
r
es
u
lt i
n
p
ar
a
m
et
er
s
w
i
s
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4
7
5
2
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l.
10
,
No
.
2
,
Ma
y
2
0
1
8
:
5
4
5
–
553
552
T
ab
le
3
.
P
ar
am
eter
s
u
s
ed
f
o
r
c
o
m
p
ar
i
n
g
r
o
u
ter
p
er
f
o
r
m
a
n
ce
P
a
r
a
me
t
e
r
s
R
o
u
t
e
r
W
i
t
h
I
n
t
e
l
l
i
g
e
n
c
e
R
o
u
t
e
r
W
i
t
h
o
u
t
I
n
t
e
l
l
i
g
e
n
c
e
T
h
r
o
u
g
h
p
u
t
(
p
e
r
c
e
n
t
)
9
6
.
2
0
%
7
2
.
5
0
%
A
v
e
r
a
g
e
L
a
t
e
n
c
y
(
mi
c
r
o
se
c
o
n
d
s)
5
3
.
2
1
5
2
.
3
M
i
so
r
d
e
r
e
d
P
a
c
k
e
t
s(
P
e
r
c
e
n
t
)
0%
2
.
1
5
%
Fig
u
r
e
7.
P
er
f
o
r
m
a
n
ce
C
o
m
p
a
r
is
o
n
o
f
R
o
u
ter
W
ith
I
n
te
lli
g
e
n
ce
an
d
W
ith
o
u
t I
n
tell
ig
e
n
ce
B
ased
o
n
L
o
ad
Of
f
er
ed
Fig
u
r
e
7
d
e
m
o
n
s
tr
ates
t
h
e
p
er
f
o
r
m
an
ce
co
m
p
ar
is
o
n
o
f
r
o
u
ter
w
it
h
i
n
telli
g
e
n
ce
an
d
r
o
u
t
er
w
it
h
o
u
t
in
telli
g
e
n
ce
o
n
b
asis
o
f
lo
a
d
o
f
f
er
ed
to
th
e
n
et
w
o
r
k
.
T
h
e
th
r
o
u
g
h
p
u
t
o
f
o
f
f
er
ed
lo
ad
f
o
r
r
o
u
ter
w
i
t
h
in
telli
g
e
n
ce
is
b
etter
co
m
p
ar
ed
to
th
e
r
o
u
ter
w
it
h
o
u
t i
n
tell
ig
en
ce
.
4.
CO
NCLU
SI
O
N
I
n
r
o
u
ter
lar
g
e
v
o
lu
m
e
o
f
d
ata
ca
n
en
ter
an
d
leav
e,
th
is
h
u
g
e
a
m
o
u
n
t
o
f
d
ata
is
h
a
n
d
led
an
d
class
i
f
ied
b
y
s
tr
u
ct
u
r
e
o
p
ti
m
i
ze
d
n
eu
r
al
n
et
w
o
r
k
,
t
h
e
ac
cu
r
ac
y
o
f
p
r
o
p
o
s
ed
s
y
s
te
m
i
s
h
ig
h
er
th
e
n
tr
ad
itio
n
a
l
class
i
f
icatio
n
m
et
h
o
d
s
li
k
e
p
o
r
t
b
ased
,
d
ee
p
p
ac
k
et
i
n
s
p
ec
ti
o
n
an
d
s
tati
s
tical
clas
s
i
f
icatio
n
.
T
h
is
m
ec
h
a
n
is
m
w
il
l
i
m
p
r
o
v
e
r
o
u
ter
e
f
f
icie
n
c
y
b
y
i
n
cr
ea
s
ed
t
h
r
o
u
g
h
p
u
t
a
n
d
d
ec
r
ea
s
ed
laten
c
y
.
A
ls
o
p
r
o
p
o
s
ed
m
et
h
o
d
is
u
s
ed
b
y
t
h
e
ad
m
in
i
s
tr
ato
r
f
o
r
ta
k
in
g
g
o
o
d
co
n
tr
o
l
d
ec
is
io
n
ab
o
u
t
t
h
e
n
et
w
o
r
k
ac
ti
v
iti
es,
clas
s
i
f
ic
atio
n
o
f
ap
p
licatio
n
u
s
a
g
e
an
d
s
t
y
le,
a
n
o
m
a
l
y
d
etec
tio
n
an
d
ac
co
u
n
ti
n
g
.
RE
F
E
R
E
NC
E
S
[1
]
L
a
k
h
in
a
,
A
n
u
k
o
o
l,
M
a
rk
Cro
v
e
ll
a
,
a
n
d
Ch
r
isto
p
h
e
Dio
t
,
"
M
in
i
n
g
a
n
o
m
a
li
e
s
u
sin
g
traff
i
c
fe
a
tu
re
d
istri
b
u
ti
o
n
s
"
,
ACM
S
IGCO
M
M
Co
mp
u
ter
Co
m
mu
n
ica
t
io
n
Rev
iew
,
v
o
l.
3
5
,
n
o
.
4
,
p
p
.
2
1
7
-
2
2
8
,
2
0
0
5
.
[2
]
C.
Estan
a
n
d
G
.
V
a
rg
h
e
se
,
“
Ne
w
d
irec
ti
o
n
s
in
traff
ic
m
e
a
su
re
m
e
n
t
a
n
d
a
c
c
o
u
n
ti
n
g
:
F
o
c
u
sin
g
o
n
th
e
e
lep
h
a
n
ts,
ig
n
o
rin
g
t
h
e
m
ice
“
,
ACM
T
ra
n
sa
c
ti
o
n
s
o
n
Co
mp
u
ter
S
y
ste
ms
,
v
o
l.
2
1
,
n
o
.
3
,
p
p
.
2
7
0
-
3
1
3
,
2
0
0
3
.
[3
]
P
e
n
g
,
J.
,
L
iu
,
T
.
,
L
i,
H.,
a
n
d
G
u
o
,
B.
,
“
En
e
rg
y
-
e
ff
icie
n
t
p
re
d
ictio
n
c
lu
ste
ri
n
g
a
lg
o
rit
h
m
f
o
r
m
u
lt
il
e
v
e
l
h
e
tero
g
e
n
e
o
u
s
w
irele
ss
s
e
n
so
r
n
e
tw
o
rk
s”
,
In
ter
n
a
ti
o
n
a
l
J
o
u
rn
a
l
o
f
Distrib
u
ted
S
e
n
so
r
Ne
two
rk
s
,
v
o
l.
9
,
n
o
.
2
,
p
p
.
1
-
8
,
2
0
1
3
.
[4
]
T
.
T
.
N
g
u
y
e
n
a
n
d
G
.
A
r
m
it
a
g
e
,
“
A
su
rv
e
y
o
f
te
c
h
n
iq
u
e
s
f
o
r
In
t
e
r
n
e
t
traff
ic
c
la
ss
i
f
ica
ti
o
n
u
sin
g
m
a
c
h
in
e
lea
rn
i
n
g
”
,
EE
E
Co
mm
u
n
ica
t
io
n
s S
u
rv
e
y
s
&
T
u
to
ri
a
ls
,
v
o
l
1
0
,
n
o
.
4
,
p
5
6
-
7
6
,
2
0
0
8
.
[5
]
A
.
M
o
o
re
a
n
d
D.
Zu
e
v
,
“
In
ter
n
e
t
tra
ff
ic
c
la
ss
if
ica
t
io
n
u
sin
g
Ba
y
e
sia
n
a
n
a
lys
is
tec
h
n
iq
u
e
s”,
A
CM
S
IG
M
ET
RICS
P
e
rf
o
rm
a
n
c
e
Ev
a
lu
a
ti
o
n
Re
v
iew
,
v
o
l.
3
3
,
n
o
.
1
,
p
p
.
5
0
-
6
0
,
2
0
0
5
.
[6
]
G
ri
m
a
u
d
o
,
L
.
,
M
e
ll
ia,
M
.
,
Ba
ra
li
s,
E.
a
n
d
Ke
ra
lap
u
ra
,
R.
,
“
S
e
lec
t:
S
e
lf
-
lea
rn
in
g
c
las
si
f
ier
f
o
r
in
tern
e
t
traff
ic”
,
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Ne
two
rk
a
n
d
S
e
r
v
ice
M
a
n
a
g
e
me
n
t
,
v
o
l.
1
1
,
n
o
.
2
,
p
p
.
1
4
4
-
1
5
7
,
2
0
1
4
.
[7
]
Bu
jl
o
w
,
T
.
,
Riaz
,
T
.
a
n
d
P
e
d
e
rse
n
,
J.M
.
,
“
A
me
th
o
d
fo
r
c
la
ss
if
ica
ti
o
n
o
f
n
e
two
rk
tra
ff
ic
b
a
se
d
o
n
C5
.
0
M
a
c
h
i
n
e
L
e
a
rn
in
g
Al
g
o
rit
h
m
,
2
0
1
2
IEE
E
In
tern
a
ti
o
n
a
l
Co
n
f
e
re
n
c
e
o
n
Co
m
p
u
ti
n
g
,
Ne
tw
o
rk
in
g
a
n
d
C
o
m
m
u
n
ica
ti
o
n
s
(ICNC),
p
p
.
2
3
7
-
2
4
1
,
2
0
1
2
.
[8
]
Ka
ra
g
ian
n
is,
T
.
,
P
a
p
a
g
ian
n
a
k
i,
K.
a
n
d
F
a
lo
u
ts
o
s,
M
.
,
“
Bli
n
c
:
m
u
lt
il
e
v
e
l
traff
ic
c
las
sif
i
c
a
ti
o
n
in
th
e
d
a
rk
”
,
ACM
S
IGCO
M
M
Co
mp
u
ter
Co
mm
u
n
ic
a
ti
o
n
Rev
iew
,
v
o
l.
3
5
,
n
o
.
4
,
p
p
.
2
2
9
–
2
4
0
,
2
0
0
5
.
[9
]
Ro
u
g
h
a
n
,
M
.
,
S
e
n
,
S
.
,
S
p
a
tsc
h
e
c
k
,
O.
a
n
d
Du
ff
ield
,
N.,
“
Cla
ss
-
of
-
S
e
rv
ice
M
a
p
p
in
g
fo
r Qo
S
:
A
sta
ti
st
ica
l
sig
n
a
tu
re
-
b
a
se
d
a
p
p
ro
a
c
h
to
IP
tra
f
fi
c
c
la
ss
if
ica
ti
o
n
”
,
P
ro
c
e
e
d
in
g
s
o
f
th
e
4
th
A
CM
S
IG
COMM
c
o
n
f
e
re
n
c
e
o
n
In
ter
n
e
t
m
e
a
su
re
m
e
n
t,
T
a
o
rm
in
a
,
S
icily
,
I
taly
p
p
.
1
3
5
-
1
4
8
,
2
0
0
4
.
[1
0
]
Af
z
a
l,
S
.
,
&
W
a
n
i,
M
.
A
.
,
“
Co
m
p
a
ra
ti
v
e
S
tu
d
y
o
f
A
d
a
p
ti
v
e
L
e
a
rn
in
g
Ra
te
w
it
h
M
o
m
e
n
tu
m
a
n
d
Re
sili
e
n
t
Ba
c
k
P
r
o
p
a
g
a
ti
o
n
A
lg
o
rit
h
m
s
f
o
r
Ne
u
r
a
l
Ne
t
Clas
sif
ier
Op
ti
m
iza
ti
o
n
”
,
In
ter
n
a
t
io
n
a
l
J
o
u
r
n
a
l
o
f
Distri
b
u
ted
a
n
d
Clo
u
d
Co
mp
u
t
in
g
,
v
o
l
.
2
,
n
o
.
1
p
p
.
1
-
6,
2
0
1
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
I
n
tellig
en
t P
a
ck
et
Delive
r
y
in
R
o
u
ter Usi
n
g
S
tr
u
ctu
r
e
Op
timiz
ed
N
eu
r
a
l Net
w
o
r
k
(
R
.
Dee
b
a
la
ksh
mi
)
553
[1
1
]
T
.
He
m
a
lath
a
,
“
S
ECURE
De
tec
t
in
g
S
y
b
il
A
tt
a
c
k
w
it
h
a
S
c
a
lab
le
P
r
o
to
c
o
l”,
In
ter
n
a
ti
o
n
a
l
J
o
u
r
n
a
l
o
f
M
C
S
q
u
a
re
S
c
ien
ti
fi
c
Res
e
a
rc
h
,
v
o
l.
4
,
n
o
.
1
,
p
p
.
3
5
-
4
1
,
2
0
1
2
.
[1
2
]
F
a
d
li
S
irait.
,
“
Ca
p
a
c
it
y
I
m
p
ro
v
e
m
e
n
t
a
n
d
P
r
o
tec
ti
o
n
o
f
LT
E
Ne
tw
o
rk
o
n
Et
h
e
rn
e
t
Ba
se
d
T
e
c
h
n
iq
u
e
”
,
T
EL
KOM
NIKA
(
T
e
lec
o
mm
u
n
ica
t
io
n
C
o
mp
u
ti
n
g
E
lec
tro
n
ics
a
n
d
C
o
n
tro
l)
,
v
o
l
.
1
6
,
n
o
.
1
,
p
p
.
1
-
1
0
,
2
0
1
8
.
[1
3
]
Am
a
n
,
A
.
H.
M
.
,
Ha
sh
i
m
,
A
.
H.
A
.
,
a
n
d
Ra
m
li
,
H.
A
.
M
.
,
“
S
i
m
u
latio
n
A
n
a
l
y
sis
f
o
r
M
u
lt
ica
st
Co
n
tex
t
De
li
v
e
r
y
Ne
tw
o
rk
M
o
b
il
it
y
M
a
n
a
g
e
m
e
n
t”,
In
d
o
n
e
sia
n
J
o
u
rn
a
l
o
f
El
e
c
trica
l
En
g
i
n
e
e
rin
g
a
n
d
I
n
fo
rm
a
ti
c
s
(
IJ
E
EI)
,
v
o
l.
5
,
n
o
.
4
,
p
p
.
3
9
0
-
3
9
4
,
2
0
1
7
.
[1
4
]
Ha
sh
im
,
A
.
H.
A
.
,
Am
a
n
,
A
.
H.
M
.
,
a
n
d
Ra
m
li
,
H.
A
.
M
.
,
“
Qu
a
n
ti
tativ
e
Ev
a
lu
a
ti
o
n
f
o
r
P
M
P
Iv
6
M
u
lt
ica
st
F
a
st
Re
ro
u
te Op
e
ra
ti
o
n
s”
,
B
u
ll
e
ti
n
o
f
El
e
c
trica
l
En
g
in
e
e
rin
g
a
n
d
In
f
o
r
ma
ti
c
s
,
v
o
l.
6
,
n
o
.
4
,
p
p
.
3
7
1
-
3
7
6
,
2
0
1
7
.
Evaluation Warning : The document was created with Spire.PDF for Python.