I
nte
rna
t
io
na
l J
o
urna
l o
f
I
nfo
rm
a
t
ics a
nd
Co
m
m
un
ica
t
io
n T
ec
hn
o
lo
g
y
(
I
J
-
I
CT
)
Vo
l.
15
,
No
.
4
,
Dec
em
b
er
20
26
,
p
p
.
1
7
4
6
~1
7
5
8
I
SS
N:
2252
-
8
7
7
6
,
DOI
:
1
0
.
1
1
5
9
1
/iji
ct
.
v15
i
4
.
pp
1
7
4
6
-
1
7
5
8
1746
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ij
ict.
ia
esco
r
e.
co
m
Dev
elo
ping
a
ma
chine
lea
rning
-
ba
sed
a
lg
o
rithm f
o
r
secure
ro
uting
in auto
no
mo
us v
ehicles
Sh
wet
a
ns
h P
riy
a
da
rsh
i
,
Alo
k
Sriv
a
s
t
a
v
a
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
o
n
i
c
s a
n
d
C
o
mm
u
n
i
c
a
t
i
o
n
,
S
c
h
o
o
l
o
f
En
g
i
n
e
e
r
i
n
g
S
c
i
e
n
c
e
s
,
G
D
G
o
e
n
k
a
U
n
i
v
e
r
s
i
t
y
,
G
u
r
u
g
r
a
m,
H
a
r
y
a
n
a
,
I
n
d
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
1
2
,
2
0
2
5
R
ev
is
ed
Ma
y
2
3
,
2
0
2
6
Acc
ep
ted
J
u
l 5
,
2
0
2
6
Ov
e
r
th
e
las
t
se
v
e
ra
l
y
e
a
rs,
we
h
a
v
e
n
o
ti
c
e
d
t
h
a
t
t
h
e
a
d
m
in
istratio
n
a
n
d
c
a
r
c
o
m
p
a
n
ies
h
a
v
e
b
e
e
n
re
fra
in
i
n
g
fro
m
in
v
e
stin
g
c
a
p
it
a
l
in
c
o
n
n
e
c
ted
a
u
to
n
o
m
o
u
s
c
a
rs.
A
c
h
a
i
n
o
f
in
terc
o
n
n
e
c
ted
se
n
s
o
rs,
a
c
tu
a
to
rs,
a
n
d
d
e
v
ice
s
—
c
o
m
b
in
i
n
g
li
d
a
r,
G
P
S
,
ra
d
a
r,
in
-
sy
ste
m
m
a
c
h
in
e
s,
c
a
m
e
ra
s,
a
n
d
m
o
re
—
a
ll
o
ws
a
u
to
n
o
m
o
u
s
c
a
rs
to
lea
rn
h
o
w
to
o
p
e
ra
te
in
re
a
l
-
ti
m
e
sc
e
n
a
rio
s.
Ho
we
v
e
r,
d
a
ta
fr
o
m
re
c
e
n
t
c
y
b
e
ra
tt
a
c
k
s
a
n
d
m
ish
a
p
s
su
g
g
e
sts
t
h
a
t
m
a
li
c
io
u
s
a
tt
a
c
k
e
rs
c
a
n
a
c
c
e
ss
th
is
se
n
so
r
d
a
ta
o
r
e
v
e
n
t
h
e
m
a
i
n
c
o
n
tro
l
sy
ste
m
th
ro
u
g
h
h
a
c
k
in
g
,
lea
d
in
g
to
a
v
a
riety
o
f
l
o
ss
e
s.
As
ti
m
e
p
a
ss
e
s
b
y
,
p
e
o
p
le
wi
ll
b
e
les
s
in
tere
ste
d
i
n
c
o
n
n
e
c
ted
a
u
t
o
n
o
m
o
u
s
v
e
h
icle
s
(CAV
s)
,
a
n
d
v
a
rio
u
s
su
r
v
e
y
s
sa
y
t
h
a
t
i
t
will
b
e
m
a
in
ly
b
e
c
a
u
se
o
f
th
e
a
b
se
n
c
e
o
f
a
p
ro
p
e
r
se
c
u
rit
y
i
n
fra
stru
c
tu
re
.
A
stu
d
y
o
f
c
y
b
e
ra
tt
a
c
k
s
o
n
C
AV
s
a
n
d
stra
teg
ies
to
e
l
imin
a
te
t
h
e
m
is
d
isc
u
ss
e
d
in
th
is
stu
d
y
.
I
n
t
h
is
st
u
d
y
,
we
h
a
v
e
p
ro
p
o
se
d
a
h
y
b
rid
m
o
d
e
l
so
l
u
ti
o
n
t
h
a
t
u
se
s
c
o
n
v
o
lu
ti
o
n
a
l
n
e
u
r
a
l
n
e
two
r
k
(
CNN
)
,
lo
n
g
s
h
o
rt
-
term
m
e
m
o
ry
(
LS
TM
)
,
a
n
d
g
a
ted
re
c
u
rre
n
t
u
n
it
(
G
RU
)
,
a
n
d
we
a
re
o
p
ti
m
izi
n
g
t
h
e
Hy
p
e
rp
a
ra
m
e
ters
u
sin
g
t
h
e
d
wa
rf
m
o
n
g
o
o
se
o
p
ti
m
iza
ti
o
n
a
l
g
o
ri
th
m
(DMOA
)
.
Th
e
p
r
o
p
o
se
d
m
o
d
e
l
is
a
lso
tes
ted
i
n
Co
m
p
a
riso
n
wit
h
o
t
h
e
r
e
x
isti
n
g
m
o
d
e
ls
,
a
n
d
we
a
re
a
b
le
to
d
e
tec
t
Do
S
,
fu
z
z
y
,
sp
o
o
f
in
g
t
h
e
d
ri
v
in
g
g
e
a
r
,
a
n
d
s
p
o
o
fin
g
t
h
e
RP
M
g
a
u
g
e
a
tt
a
c
k
s
with
a
n
a
c
c
u
ra
c
y
o
f
m
o
re
t
h
a
n
9
8
%
.
K
ey
w
o
r
d
s
:
Au
to
n
o
m
o
u
s
ca
r
s
C
o
n
n
ec
ted
v
eh
icles
C
y
b
er
-
cr
im
es
Ma
licio
u
s
-
attac
k
s
R
o
u
te
d
iv
er
s
io
n
T
h
is i
s
a
n
o
p
e
n
a
c
c
e
ss
a
rticle
u
n
d
e
r th
e
CC B
Y
-
SA
li
c
e
n
se
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Sh
weta
n
s
h
Priy
ad
ar
s
h
i
Dep
ar
tm
en
t o
f
E
lectr
o
n
ics an
d
C
o
m
m
u
n
icatio
n
,
Sch
o
o
l o
f
E
n
g
in
ee
r
in
g
Scien
ce
s
GD
Go
en
k
a
Un
iv
er
s
ity
Gu
r
u
g
r
am
,
Har
y
a
n
a,
I
n
d
ia
E
m
ail: sh
weta
n
s
h
0
8
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
As
s
ec
u
r
ity
v
u
ln
er
ab
ilit
ies
in
co
n
n
ec
ted
au
t
o
n
o
m
o
u
s
v
eh
i
cles
(
C
AV
s
)
ar
e
a
m
ajo
r
co
n
ce
r
n
f
o
r
h
u
m
an
life
an
d
in
f
r
astru
ctu
r
e
s
ec
u
r
ity
,
it
is
alwa
y
s
s
u
g
g
ested
to
h
av
e
a
r
o
b
u
s
t
ar
ch
itectu
r
e
in
ter
m
s
o
f
cy
b
er
s
ec
u
r
ity
f
o
r
C
AVs.
Ma
n
y
Au
to
n
o
m
o
u
s
ca
r
m
an
u
f
a
ctu
r
er
s
h
av
e
im
p
ac
tf
u
l
id
ea
s
th
at
f
o
cu
s
o
n
th
e
s
ec
u
r
ity
ar
ea
,
an
d
th
ey
h
a
v
e
m
ad
e
h
u
g
e
s
tr
id
es.
Ho
wev
er
,
th
e
p
r
o
p
o
s
ed
s
o
lu
tio
n
m
et
h
o
d
s
a
r
e
n
o
t
ab
le
to
m
ee
t
th
e
r
eq
u
ir
e
m
en
ts
,
an
d
t
h
er
e
a
r
e
s
till
s
ev
er
al
s
ec
u
r
ity
v
u
ln
e
r
ab
ilit
ies
th
at
p
er
s
is
t
ac
r
o
s
s
t
h
e
in
d
u
s
tr
y
.
I
n
t
h
is
r
esear
ch
,
L
ev
el
5
o
f
C
AVs
ar
e
ev
alu
ated
.
W
ith
th
is
r
ec
o
m
m
en
d
ed
s
o
lu
tio
n
,
it
is
ex
p
ec
te
d
to
p
r
o
v
id
e
a
s
af
e
an
d
s
ec
u
r
e
en
v
ir
o
n
m
en
t
f
o
r
Au
to
n
o
m
o
u
s
ca
r
s
an
d
ca
r
u
s
er
s
.
T
h
is
s
o
lu
tio
n
s
h
o
u
ld
b
e
ab
le
to
h
el
p
in
p
r
o
ce
s
s
in
g
th
e
o
p
e
r
atio
n
o
f
d
ataset
s
eg
r
eg
atio
n
to
b
e
u
s
ed
f
o
r
e
m
er
g
en
c
y
(
SOS),
m
ap
p
i
n
g
,
a
n
d
m
an
o
e
u
v
r
e
s
ce
n
ar
io
s
b
y
li
n
k
in
g
to
t
h
e
d
atab
ases
co
n
n
ec
ted
i
n
th
e
C
AV
ar
ch
itectu
r
e.
T
h
e
s
o
lu
tio
n
av
ailab
le
o
v
er
th
e
g
lo
b
e
u
s
es
co
n
v
o
lu
tio
n
al
n
e
u
r
al
n
etwo
r
k
s
(
C
NN
)
,
r
ec
u
r
r
en
t
n
eu
r
al
n
etwo
r
k
s
(
R
NN
)
,
o
r
ar
tific
ial
n
eu
r
al
n
etwo
r
k
(
ANN
)
m
o
d
el
f
o
r
tr
ai
n
in
g
th
ei
r
C
AV
s
y
s
tem
s
,
wh
ich
ar
e
h
ar
d
ly
ab
le
t
o
ac
h
ie
v
e
an
ac
cu
r
ac
y
o
f
8
5
%
[
1
]
-
[
6
]
.
B
u
t,
th
e
m
o
d
el
wh
ich
we
ar
e
p
r
o
p
o
s
in
g
is
a
h
y
b
r
id
m
o
d
el
o
f
R
NN
an
d
C
NN
with
h
y
p
er
p
a
r
am
eter
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Dev
elo
p
in
g
a
ma
ch
in
e
le
a
r
n
in
g
-
b
a
s
ed
a
lg
o
r
ith
m
fo
r
s
ec
u
r
ed
r
o
u
te
in
… (
S
h
w
eta
n
s
h
P
r
iya
d
a
r
s
h
i
)
1747
tu
n
in
g
an
d
is
ex
p
ec
ted
to
h
a
v
e
9
9
%
ac
cu
r
ac
y
.
T
h
e
s
o
lu
tio
n
ex
p
ec
te
d
in
th
is
p
a
p
er
is
b
ased
o
n
m
ac
h
in
e
L
ea
r
n
in
g
,
wh
ich
p
r
o
v
id
es a
r
e
al
-
tim
e
in
ter
co
n
n
ec
tio
n
lay
er
[
7
]
.
2.
L
I
T
E
R
AT
U
RE
R
E
VI
E
W
Fo
r
th
is
Ar
ticle,
we
h
av
e
u
s
ed
1
5
0
+
m
a
n
u
s
cr
ip
ts
f
o
r
o
u
r
r
ef
er
en
ce
s
tu
d
y
,
a
n
d
af
ter
ca
r
ef
u
l
,
in
-
d
ep
t
h
s
tu
d
y
,
we
f
o
u
n
d
ar
o
u
n
d
5
0
m
an
u
s
cr
ip
ts
alig
n
in
g
clo
s
ely
to
o
u
r
r
esear
ch
ar
ea
.
T
h
e
cy
b
e
r
s
ec
u
r
ity
is
s
u
e,
d
ata
s
ec
u
r
ity
ch
allen
g
es
,
an
d
also
th
e
cy
b
er
attac
k
s
in
Au
to
n
o
m
o
u
s
ca
r
s
ar
e
a
v
er
y
v
ast
ar
ea
,
an
d
o
u
r
r
esear
c
h
is
v
er
y
m
u
ch
r
estricte
d
to
s
ec
u
r
i
n
g
th
e
r
o
u
te
o
f
c
o
n
n
ec
ted
au
to
n
o
m
o
u
s
ca
r
s
.
Ou
r
r
esear
ch
ar
e
a
also
d
r
aws
a
th
in
lin
e
b
etwe
en
p
r
e
v
io
u
s
r
esear
c
h
an
d
o
u
r
r
esear
ch
as
s
ec
u
r
in
g
th
e
r
o
u
te
o
f
f
u
lly
au
to
n
o
m
o
u
s
ca
r
s
en
co
m
p
ass
es
all
th
e
m
ajo
r
is
s
u
es
in
Au
to
n
o
m
o
u
s
ca
r
s
.
I
f
th
er
e
is
a
n
y
m
alf
u
n
ctio
n
in
th
e
d
ata
co
m
in
g
f
r
o
m
th
e
clo
u
d
,
if
th
e
ADAS
o
f
AV
i
s
attac
k
ed
,
o
r
if
th
e
s
en
s
o
r
s
,
r
ad
ar
s
,
GPS
o
r
an
y
ca
r
m
o
u
n
ted
eq
u
ip
m
en
t
’
s
wh
ich
co
llects
d
ata
f
r
o
m
e
x
ter
n
al
s
o
u
r
ce
s
ar
e
b
ein
g
co
m
p
r
o
m
is
ed
,
th
e
n
it
is
v
er
y
lik
ely
th
at
th
e
r
o
u
te
o
f
Au
t
o
n
o
m
o
u
s
ca
r
s
ca
n
b
e
co
m
p
r
o
m
is
ed
.
T
h
e
r
e
ar
e
ce
r
tai
n
p
o
i
n
ts
o
f
ac
ce
s
s
th
at
cy
b
e
r
c
r
im
in
als
ex
p
lo
it
t
o
g
et
ac
ce
s
s
to
C
AVs
’
s
y
s
tem
s
an
d
th
at
n
ee
d
r
o
b
u
s
t
s
ec
u
r
ity
in
o
r
d
er
t
o
av
o
id
t
h
ese
attac
k
s
,
wh
ich
we
f
o
u
n
d
m
is
s
in
g
i
n
th
e
p
r
ev
i
o
u
s
r
esear
ch
p
ap
er
s
.
I
n
o
u
r
r
esear
ch
,
we
h
av
e
tr
ied
to
co
v
er
th
o
s
e
ar
ea
s
an
d
also
p
r
o
v
id
e
a
latest
an
d
v
er
y
r
o
b
u
s
t
m
ec
h
an
is
m
wh
ich
n
o
t
o
n
ly
s
af
eg
u
ar
d
s
th
e
r
o
u
te
o
f
Au
t
o
n
o
m
o
u
s
ca
r
s
b
u
t
also
p
r
o
v
i
d
es
a
s
af
e
d
r
iv
i
n
g
en
v
ir
o
n
m
en
t f
o
r
ca
r
s
an
d
ca
r
o
wn
er
s
,
in
wh
ich
ca
r
d
r
iv
er
s
'
s
a
f
ety
is
also
n
o
t c
o
m
p
r
o
m
is
ed
.
T
h
e
Pre
v
io
u
s
d
ee
p
lear
n
in
g
m
o
d
els
i.e
.
C
NN
,
A
NN
,
o
r
tr
ad
itio
n
al
R
NN
f
ailed
in
o
n
e
way
o
r
an
o
th
e
r
,
an
d
in
th
is
r
esear
ch
,
we
h
av
e
tr
ied
to
a
d
d
r
ess
all
th
e
e
x
is
tin
g
lo
o
p
h
o
les
in
th
ese
m
o
d
els.
I
n
s
o
lu
tio
n
to
th
e
g
ap
s
a
n
d
lo
o
p
h
o
les
in
th
ese
m
o
d
els,
we
h
a
v
e
p
r
o
p
o
s
ed
a
h
y
b
r
id
m
o
d
e
l
th
at
co
m
p
r
is
es
o
f
C
NN+
lo
n
g
s
h
o
r
t
-
ter
m
m
em
o
r
y
(
L
STM
)
+
g
ated
r
ec
u
r
r
en
t
u
n
it
(
GR
U)
.
T
h
is
h
e
lp
s
in
ad
ju
s
tin
g
a
n
d
f
ix
in
g
ev
en
th
e
s
m
allest
f
ailu
r
es
o
f
an
y
in
d
iv
i
d
u
al
m
o
d
el.
T
h
is
m
o
d
el
is
f
o
llo
wed
with
a
n
aim
to
ac
h
iev
e
h
ig
h
er
ac
cu
r
ac
y
an
d
p
r
o
v
id
e
a
r
o
b
u
s
t
an
d
s
ec
u
r
e
m
o
d
el
;
af
ter
th
at
we
h
av
e
tr
ied
o
p
tim
izatio
n
with
d
war
f
m
o
n
g
o
o
s
e
o
p
tim
izatio
n
alg
o
r
it
hm
(
DM
O
A
)
i.e
.
,
h
y
p
er
p
ar
am
eter
s
(
n
eu
r
o
n
s
in
ea
ch
la
y
er
)
a
r
e
o
p
t
im
ized
u
s
in
g
th
e
DM
OA
.
3.
RE
S
E
ARCH
M
E
T
H
O
D
3
.
1
.
Wo
r
k
ing
o
f
a
uto
no
m
o
u
s
v
ehicles
C
AVs
,
al
s
o
k
n
o
wn
as
in
d
ep
e
n
d
en
t
ca
r
s
i.e
.
it
d
ep
en
d
s
u
p
o
n
th
e
lev
el
o
f
Au
to
m
atio
n
th
ese
ca
r
s
h
av
e
as
s
h
o
wn
T
ab
le
1
,
p
er
f
o
r
m
ac
tio
n
s
u
s
in
g
a
co
m
b
in
atio
n
o
f
s
u
p
er
-
ad
v
an
ce
d
f
ea
tu
r
es
th
at
allo
w
th
em
to
u
n
d
er
s
tan
d
th
eir
s
u
r
r
o
u
n
d
i
n
g
s
,
m
ak
e
d
ec
is
io
n
s
,
a
n
d
c
o
n
tr
o
l
th
e
v
eh
icle
with
o
u
t
h
u
m
an
a
s
s
is
tan
ce
[
8
]
-
[
1
0
]
.
So
,
in
o
r
d
er
to
u
n
d
e
r
s
tan
d
co
m
p
lete
f
u
n
ctio
n
in
g
o
f
Au
to
n
o
m
o
u
s
v
eh
icles
,
it
ca
n
b
e
s
eg
r
e
g
ated
o
n
a
r
o
le
b
asis
:
a)
Sen
s
o
r
s
:
T
h
ese
s
en
s
o
r
s
h
elp
Au
to
n
o
m
o
u
s
v
eh
icles
with
v
i
s
io
n
an
d
g
iv
e
th
em
th
e
ab
ilit
y
to
p
er
f
o
r
m
ac
tio
n
s
.
C
AVs
ar
e
well
eq
u
ip
p
ed
with
s
m
ar
t
s
en
s
o
r
s
to
d
et
ec
t
th
e
s
u
r
r
o
u
n
d
in
g
s
.
Au
to
n
o
m
o
u
s
Veh
icles
ar
e
eq
u
ip
p
ed
with
L
id
a
r
,
R
ad
a
r
,
C
am
er
as,
an
d
Ultr
aso
n
ic
s
en
s
o
r
s
.
b)
L
o
ca
lizatio
n
:
I
n
o
r
d
e
r
to
n
av
i
g
ate
p
r
o
p
er
ly
,
C
AVs
n
ee
d
ex
a
ct
lo
ca
tio
n
s
o
n
a
r
eg
u
lar
b
asis
.
Au
to
n
o
m
o
u
s
v
eh
icles
ar
e
eq
u
ip
p
ed
with
GPS,
Hig
h
-
d
ef
in
itio
n
m
ap
s
,
an
d
s
en
s
o
r
f
u
s
io
n
,
wh
ich
co
m
b
i
n
es
d
ata
f
r
o
m
s
en
s
o
r
s
an
d
m
ap
s
.
c)
Per
ce
p
tio
n
:
T
h
is
is
a
b
o
u
t
wh
a
t
is
h
ap
p
e
n
in
g
in
a
n
d
a
r
o
u
n
d
th
e
v
eh
icle.
I
n
o
r
d
er
t
o
k
ee
p
tr
ac
k
o
f
ca
r
s
,
o
b
jects,
s
tr
ee
t
walk
er
s
,
cy
clis
ts
,
r
o
ad
s
ig
n
s
,
an
d
tr
af
f
ic
s
ig
n
als
.
It
u
s
es
AI
an
d
m
ac
h
in
e
v
is
io
n
t
o
u
n
d
er
s
tan
d
t
h
e
r
aw
d
ata
f
r
o
m
s
en
s
o
r
s
.
d)
Plan
n
in
g
:
T
h
is
is
all
ab
o
u
t
h
o
w
au
to
n
o
m
o
u
s
ca
r
s
p
lan
t
h
eir
wo
r
k
,
wh
at
alg
o
r
ith
m
is
b
ein
g
u
s
ed
to
n
av
ig
ate
f
r
o
m
o
n
e
p
lace
t
o
an
o
th
e
r
,
m
a
k
in
g
r
ea
l
-
life
d
ec
is
io
n
s
,
p
lan
n
in
g
c
o
m
p
let
e
r
o
u
te
,
an
d
ap
p
r
o
ac
h
in
g
th
e
d
esti
n
atio
n
.
e)
C
o
n
tr
o
l:
T
h
is
s
eg
m
en
t
is
a
ll
ab
o
u
t
th
e
m
ajo
r
co
n
tr
o
l
o
f
h
o
w
AVs
d
ea
l
with
s
teer
in
g
,
b
r
ak
es,
an
d
ac
ce
ler
atio
n
.
T
h
e
c
o
m
p
lete
s
y
s
tem
m
an
ag
es a
n
d
a
d
ju
s
ts
its
el
f
to
k
ee
p
th
e
v
e
h
icle
o
n
c
o
u
r
s
e
s
af
ely
.
f)
Ar
tific
ial
in
tellig
en
ce
:
I
t w
o
r
k
s
as th
e
b
r
ain
o
f
C
AVs.
T
h
e
ca
r
'
s
s
y
s
tem
is
we
ll tr
ain
ed
wit
h
h
u
g
e
d
atasets
i.e
.
v
ar
io
u
s
ca
s
e
s
ce
n
ar
io
s
an
d
h
o
w
to
h
an
d
le
th
em
.
I
t
ca
n
p
r
ed
ict
b
eh
av
io
u
r
,
r
ec
o
g
n
ize
p
atter
n
s
,
an
d
ac
t
ac
co
r
d
in
g
to
v
a
r
io
u
s
s
itu
atio
n
s
.
3
.
2
.
T
y
pes
o
f
a
uto
no
mo
us
v
ehicles
W
h
en
th
e
v
eh
icles
s
tar
ted
co
m
in
g
to
m
a
r
k
et
at
th
at
tim
e
n
o
t
ev
en
0
.
1
%
au
to
m
atio
n
was
th
er
e
an
d
with
tim
e
th
e
au
to
m
atio
n
lev
el
in
cr
ea
s
ed
;
th
is
ca
n
b
e
s
ee
n
i
n
T
ab
le
1
.
3
.
3
.
Wo
r
k
ing
o
f
deep
lea
rning
m
o
dels
C
o
n
n
ec
ted
Au
to
n
o
m
o
u
s
ca
r
s
u
s
e
v
ar
io
u
s
al
g
o
r
ith
m
s
t
o
f
u
n
ctio
n
b
ased
o
n
AI
an
d
m
ac
h
in
e
lear
n
in
g
,
s
u
ch
as
cr
y
p
to
g
r
ap
h
ic
alg
o
r
ith
m
s
an
d
s
et
o
f
p
r
o
to
co
ls
as
a
co
m
b
in
atio
n
to
s
af
eg
u
ar
d
th
e
ca
r
s
f
r
o
m
cy
b
e
r
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
4
,
Dec
em
b
er
20
26
:
1
7
4
6
-
1
7
5
8
1748
th
r
ea
ts
[
1
1
]
.
T
h
ese
alg
o
r
ith
m
s
ar
e
th
e
b
ac
k
b
o
n
e
o
n
wh
ich
t
h
e
co
m
p
lete
AVs
s
y
s
tem
wo
r
k
s
u
ch
as
s
ec
u
r
in
g
th
e
co
n
n
ec
tio
n
s
an
d
c
o
m
m
u
n
icatio
n
s
,
co
n
tr
o
llin
g
th
e
ac
ce
s
s
,
an
d
s
af
eg
u
ar
d
in
g
th
e
d
ata,
as sh
o
wn
in
Fig
u
r
e
1
.
a)
E
n
cr
y
p
tio
n
alg
o
r
ith
m
:
E
n
cr
y
p
tio
n
is
m
ain
ly
d
iv
id
ed
in
to
two
ty
p
es,
i.e
.
s
y
m
m
etr
ic
an
d
Asy
m
m
etr
ic
alg
o
r
ith
m
.
Sy
m
m
etr
ic
alg
o
r
it
h
m
s
ar
e
u
s
ed
f
o
r
lar
g
e
a
m
o
u
n
ts
o
f
d
ata
an
d
in
th
at
ca
r
s
u
s
e
ad
v
an
ce
d
en
cr
y
p
tio
n
s
tan
d
ar
d
(
AE
S)
f
o
r
s
to
r
in
g
d
ata
an
d
s
af
eg
u
a
r
d
in
g
co
m
m
u
n
icatio
n
[
1
2
]
-
[
1
5
]
.
Mo
r
e
o
v
er
,
Asy
m
m
etr
ic
alg
o
r
ith
m
en
c
r
y
p
ted
k
e
y
ex
c
h
an
g
e,
au
t
h
en
tic
atio
n
,
s
ec
u
r
e
lo
g
in
.
I
t
u
s
ed
R
SA
f
o
r
d
ig
ital
s
ig
n
s
an
d
in
ter
ch
an
g
in
g
th
e
k
ey
s
an
d
ellip
tic
cu
r
v
e
c
r
y
p
t
o
g
r
ap
h
y
(
E
C
C
)
is
u
s
ed
f
o
r
ef
f
icien
t
em
b
e
d
d
e
d
s
y
s
tem
in
v
eh
icles.
b)
Hash
in
g
alg
o
r
ith
m
:
I
t
is
u
s
ed
f
o
r
d
ata
i
n
teg
r
atio
n
b
y
u
s
in
g
SHA
-
2
5
6
/
SHA
-
3
to
v
e
r
if
y
s
o
f
twar
e’
s
an
d
s
ec
u
r
in
g
b
lo
c
k
ch
ain
l
o
g
s
.
c)
Secu
r
e
co
m
m
u
n
icatio
n
p
r
o
to
c
o
ls
:
I
ts
u
s
es
tr
an
s
p
o
r
t
lay
e
r
s
ec
u
r
ity
(
T
L
S)
f
o
r
d
ata
en
cr
y
p
tio
n
b
etwe
en
V2
X,
d
atag
r
am
T
L
S
(
DT
L
S)
u
s
ed
f
o
r
UDP
b
ased
C
o
m
m
u
n
icatio
n
,
I
Ps
ec
u
s
ed
f
o
r
p
r
o
tectin
g
d
at
a
p
ac
k
ets.
d)
Me
s
s
ag
e
au
th
en
ticatio
n
co
d
es
(
MA
C
s
)
:
is
u
s
ed
in
o
r
d
er
to
v
er
if
y
an
d
p
r
o
tect
m
ess
ag
es
w
ith
th
e
h
elp
o
f
h
ash
-
b
ased
m
ess
ag
e
au
th
e
n
ticatio
n
co
d
e
(
HM
AC
)
.
e)
I
n
tr
u
s
io
n
d
etec
tio
n
alg
o
r
ith
m
s
:
I
t
is
u
s
ed
to
d
etec
t
s
u
s
p
icio
u
s
b
eh
av
io
u
r
an
d
a
n
y
in
v
asio
n
attem
p
ts
o
n
ca
r
.
T
h
er
e
a
r
e
two
ty
p
es o
f
I
D
S u
s
ed
n
o
r
m
ally
m
ac
h
in
e
lear
n
in
g
-
b
ased
an
d
r
u
le
-
b
ased
I
DS.
f)
Dig
ital sig
n
atu
r
es: is
a
s
u
p
p
lem
en
tar
y
way
o
f
p
r
o
tectio
n
u
s
e
d
f
o
r
v
er
if
y
in
g
th
e
m
ess
ag
e
an
d
d
ata
co
m
in
g
f
r
o
m
v
ar
i
o
u
s
s
o
u
r
ce
s
.
I
t
u
s
es
ellip
tic
cu
r
v
e
d
ig
ital
s
ig
n
at
u
r
e
alg
o
r
ith
m
(
E
C
DSA)
an
d
R
SA
Dig
ital
Sig
n
atu
r
es.
g)
B
lo
ck
ch
ain
:
T
h
is
is
n
o
t
m
an
d
ato
r
ily
u
s
ed
b
u
t
it
is
u
s
ed
f
o
r
s
ec
u
r
in
g
th
e
lo
g
g
in
g
o
f
d
ata
to
av
o
id
tam
p
er
in
g
.
T
ab
le
1
.
L
e
v
el
o
f
a
u
to
m
atio
n
C
a
t
e
g
o
r
y
H
u
ma
n
a
ssi
s
t
a
n
c
e
Le
v
e
l
o
f
a
u
t
o
ma
t
i
o
n
Tr
a
d
i
t
i
o
n
a
l
Y
e
s
0
–
M
a
n
u
a
l
C
o
n
t
r
o
l
S
mart
Y
e
s/
N
o
1
–
D
r
i
v
e
r
R
e
q
u
i
r
e
d
S
mart
Y
e
s/
N
o
2
–
P
a
r
t
i
a
l
S
u
p
p
o
r
t
R
e
q
u
i
r
e
d
A
u
t
o
n
o
m
o
u
s
Y
e
s/
N
o
3
–
C
o
n
d
i
t
i
o
n
a
l
A
ss
i
st
a
n
c
e
A
u
t
o
n
o
m
o
u
s
Y
e
s/
N
o
4
–
H
i
g
h
l
y
A
u
t
o
ma
t
e
d
A
u
t
o
n
o
m
o
u
s
No
5
–
F
u
l
l
y
A
u
t
o
m
a
t
e
d
Fig
u
r
e
1
.
W
o
r
k
i
n
g
o
f
ML
alg
o
r
ith
m
s
[
1
3
]
3
.
4
.
E
x
is
t
ing
mo
dels
T
h
er
e
ar
e
m
o
d
els
o
n
wh
ich
Au
to
n
o
m
o
u
s
ca
r
s
wo
r
k
i.e
.
h
o
w
th
ese
alg
o
r
ith
m
s
f
u
n
ctio
n
.
So
,
o
v
er
all,
th
is
is
p
er
f
o
r
m
ed
b
y
m
ac
h
in
e
l
ea
r
n
in
g
wh
ic
h
co
m
es u
n
d
er
th
e
AI
s
eg
m
en
t o
f
C
AVs [
1
6
]
-
[
2
5
]
.
Her
e
in
s
tead
o
f
d
o
in
g
p
r
o
g
r
am
m
in
g
,
s
y
s
tem
is
m
ad
e
to
lear
n
an
d
e
n
h
an
c
e
its
elf
with
th
e
h
elp
o
f
h
u
g
e
d
atasets
.
T
h
ese
tr
ain
in
g
s
in
v
o
lv
e
alg
o
r
ith
m
s
th
at
ca
n
ev
alu
ate
d
ata,
r
ec
o
g
n
i
ze
p
atter
n
s
,
an
d
p
r
ed
ict
co
m
i
n
g
s
itu
atio
n
s
b
ased
o
n
av
ailab
le
p
atter
n
s
.
Dee
p
l
e
ar
n
in
g
p
la
y
s
a
v
er
y
v
ital r
o
le
i
n
wo
r
k
in
g
o
f
a
u
t
o
n
o
m
o
u
s
v
eh
icles su
ch
a
s
s
ee
in
g
,
an
aly
s
in
g
,
an
d
r
ea
ctin
g
.
T
h
ese
m
o
d
els ar
e
ey
es,
b
r
ain
,
h
an
d
s
f
o
r
a
u
t
o
n
o
m
o
u
s
v
eh
icles Fig
u
r
e
2.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Dev
elo
p
in
g
a
ma
ch
in
e
le
a
r
n
in
g
-
b
a
s
ed
a
lg
o
r
ith
m
fo
r
s
ec
u
r
ed
r
o
u
te
in
… (
S
h
w
eta
n
s
h
P
r
iya
d
a
r
s
h
i
)
1749
a)
C
NNs
:
I
t
is
u
s
ed
f
o
r
a
n
ticip
atio
n
,
d
etec
tin
g
o
b
jects,
r
ec
o
g
n
i
zin
g
tr
af
f
ic
s
ig
n
als
an
d
s
ig
n
s
,
d
etec
tin
g
t
h
e
m
ar
k
in
g
o
f
lan
es.
I
t
u
s
es
s
em
an
tic
s
eg
m
en
tat
io
n
to
d
is
tin
g
u
is
h
b
etwe
en
walk
a
b
le
ar
ea
an
d
d
r
i
v
in
g
ar
ea
,
lan
e
r
ec
o
g
n
itio
n
,
etc.
I
t
also
u
s
es
s
in
g
le
s
h
o
t
d
etec
to
r
(
SS
D
)
f
o
r
f
aster
tr
ac
k
in
g
o
f
o
b
ject
an
d
ac
tin
g
as
p
er
th
e
r
e
q
u
ir
em
e
n
t.
b)
R
NNs
an
d
L
STM
s
:
T
h
ese
m
o
d
els
ar
e
u
s
ed
f
o
r
d
etec
tin
g
t
h
e
b
eh
a
v
io
u
r
o
f
ca
r
s
,
a
n
d
p
e
d
estrian
s
.
Her
e
L
STM
is
u
s
ed
f
o
r
p
r
e
d
ictin
g
t
h
e
b
eh
a
v
io
u
r
s
o
f
o
th
e
r
ca
r
s
,
an
d
p
ed
estrian
s
.
c)
T
r
an
s
f
o
r
m
e
r
m
o
d
els
:
T
h
is
m
o
d
el
is
in
tr
o
d
u
ce
d
f
o
r
s
elf
-
atten
tio
n
,
an
d
ev
alu
atio
n
o
f
s
ce
n
es
an
d
d
ec
id
in
g
th
e
ap
p
r
o
p
r
iate
ac
tio
n
.
T
h
is
m
o
d
el
is
also
r
esp
o
n
s
ib
le
f
o
r
Sen
s
o
r
f
u
s
io
n
(
e.
g
.
,
co
m
b
i
n
in
g
L
iDAR
+
ca
m
er
as).
Fig
u
r
e
2
.
Mo
d
els
o
f
d
ee
p
lear
n
in
g
u
s
ed
i
n
C
AV
[
2
0
]
3
.
5
.
T
y
pes
o
f
a
t
t
a
c
k
s
a
dd
re
s
s
ed
in o
ur
m
o
del
I
n
th
e
p
r
o
p
o
s
ed
m
o
d
el
we
ar
e
f
o
cu
s
in
g
o
n
4
m
ajo
r
cy
b
er
-
attac
k
s
,
w
h
ich
h
a
v
e
d
ev
astatin
g
c
o
n
s
e
q
u
e
n
c
e
s
o
n
a
u
t
o
n
o
m
o
u
s
v
e
h
i
c
l
e
s
s
u
c
h
as
l
o
s
s
o
f
v
e
h
icl
e
,
l
o
s
s
o
f
li
f
e
,
l
o
s
s
o
f
I
n
f
r
as
t
r
u
c
t
u
r
e
,
l
a
r
g
e
s
c
a
l
e
t
r
a
f
f
i
c
I
n
c
i
d
e
n
ts
w
h
i
c
h
l
e
a
d
t
o
lo
s
s
o
f
l
i
f
e
f
o
r
o
t
h
e
r
r
o
a
d
u
s
e
r
s
to
o
[
2
5
]
-
[
3
5
]
.
A
t
t
a
c
k
s
i
n
c
l
u
d
e
d
i
n
o
u
r
m
o
d
e
l
a
r
e
:
a)
Do
S
attac
k
:
Do
S
s
tan
d
s
f
o
r
d
en
ial
-
of
-
Ser
v
ice
attac
k
b
y
f
lo
o
d
in
g
th
e
a
u
to
n
o
m
o
u
s
v
eh
icles
co
m
m
u
n
icatio
n
s
y
s
tem
,
s
o
th
a
t
th
e
ca
r
s
n
etwo
r
k
r
eso
u
r
ce
s
a
r
e
u
n
a
v
ailab
le
to
p
r
o
v
id
e
th
e
s
er
v
ice
.
T
h
ese
m
alici
o
u
s
tr
af
f
ic
m
a
k
es
th
e
ac
tu
al
d
ata
in
f
l
o
w
tr
o
u
b
leso
m
e
a
n
d
th
is
lead
s
to
f
ailin
g
o
f
s
en
s
o
r
s
an
d
s
af
ety
d
ev
ices in
au
to
n
o
m
o
u
s
v
eh
icl
es.
b)
Fu
zz
y
attac
k
s
:
Fu
zz
y
attac
k
s
ar
e
way
s
to
ex
p
o
s
e
th
e
v
u
ln
e
r
ab
ilit
ies
o
f
au
to
n
o
m
o
u
s
ca
r
s
b
y
in
s
er
tin
g
th
e
r
an
d
o
m
d
ata
p
ac
k
ets
in
to
th
e
v
eh
icle'
s
co
n
tr
o
ller
ar
ea
n
etwo
r
k
wh
ich
m
ak
es
th
e
b
eh
av
io
u
r
o
f
v
eh
icl
e
o
u
ts
id
e
th
e
ex
p
ec
tatio
n
s
o
f
ca
r
o
wn
er
o
r
ca
n
b
e
ca
lled
as v
er
y
u
n
p
r
ed
ictab
le
f
o
r
ca
r
u
s
er
s
.
c)
Sp
o
o
f
in
g
th
e
d
r
i
v
in
g
g
e
ar
:
T
h
is
is
o
n
e
o
f
th
e
d
a
n
g
er
o
u
s
attac
k
s
in
au
to
n
o
m
o
u
s
v
e
h
icle
wh
er
e
attac
k
er
s
s
en
d
f
alse
s
tatu
s
o
f
v
eh
icles
c
u
r
r
en
t
g
ea
r
s
tatu
s
to
th
e
co
n
t
r
o
ller
ar
ea
n
etwo
r
k
b
u
s
.
Fo
r
ex
am
p
le:
-
p
ar
k
,
r
ev
er
s
al,
n
eu
tr
al,
an
d
d
r
iv
e
.
t
h
is
lead
s
to
u
n
s
af
e
ac
tio
n
s
b
y
v
eh
icle
wh
ic
h
will
b
e
co
n
t
r
ad
ictab
le
t
o
v
eh
icle
an
d
e
n
v
ir
o
n
m
en
t’
s
cu
r
r
en
t d
em
an
d
.
d)
Sp
o
o
f
in
g
th
e
R
PM
g
au
g
e
:
Sp
o
o
f
in
g
th
e
r
ev
o
l
u
tio
n
p
e
r
m
in
u
t
e
(
R
PM)
g
au
g
e
in
an
au
to
n
o
m
o
u
s
v
eh
icle
is
a
cy
b
er
-
attac
k
p
r
ac
tice
b
y
atta
ck
er
s
to
in
ject
a
f
alse
o
r
m
alici
o
u
s
m
ess
ag
e
ab
o
u
t th
e
R
PM
o
r
th
e
en
g
in
e
’
s
s
p
ee
d
to
t
h
e
co
n
tr
o
ller
ar
ea
n
etwo
r
k
-
bus
.
T
h
is
ac
t
is
d
o
n
e
with
an
aim
to
t
r
ick
th
e
ac
ti
o
n
o
f
c
o
n
tr
o
l
s
y
s
tem
o
f
v
eh
icle
wh
ic
h
lead
s
to
m
is
u
n
d
e
r
s
tan
d
in
g
at
elec
tr
o
n
ic
u
n
it
c
o
n
tr
o
l
(
E
C
U)
L
ev
e
l.
Her
e,
wh
en
E
C
U
b
eliev
es
th
at
th
e
v
eh
icle
o
r
en
g
in
e
is
r
u
n
n
in
g
at
d
if
f
er
en
t
s
p
ee
d
th
an
th
e
ac
tu
al
s
p
ee
d
wh
ich
lead
s
to
ac
tio
n
s
lik
e
ch
an
g
in
g
th
e
g
e
ar
o
r
s
o
m
e
u
n
p
r
ed
ictab
le
ac
tio
n
th
at
ca
n
b
e
h
az
ar
d
o
u
s
.
3
.
6
.
P
r
o
po
s
ed
m
o
del
E
v
en
af
ter
u
s
in
g
d
ee
p
lear
n
i
n
g
tech
n
iq
u
es
an
d
in
tr
o
d
u
cin
g
m
o
d
els
lik
e
C
NN,
R
N
N,
an
d
ANN.
we
ar
e
u
n
ab
le
to
ac
h
iev
e
ac
cu
r
ac
y
in
ter
m
s
o
f
cy
b
er
-
attac
k
s
.
T
h
is
is
a
p
o
ten
tial
th
r
ea
t
th
at
is
ca
u
s
in
g
lo
s
s
o
f
d
ata,
lo
s
s
o
f
liv
es,
an
d
lo
s
s
o
f
in
f
r
astru
ctu
r
e.
T
o
o
v
e
r
c
o
m
e
th
is
,
we
ar
e
in
tr
o
d
u
cin
g
a
m
o
d
el
th
at
will
b
e
h
av
in
g
h
ig
h
er
ac
cu
r
ac
y
th
a
n
th
e
ex
is
tin
g
m
o
d
els
[
3
5
]
-
[
4
0
]
.
I
f
we
tr
y
to
u
n
d
er
s
tan
d
th
e
ac
cu
r
ac
y
b
y
d
ata
C
NN
-
8
5
%
a
cc
u
r
ac
y
,
R
NN
-
8
5
.
6
%
a
cc
u
r
ac
y
,
ANN
-
8
0
%
ac
cu
r
ac
y
,
w
h
ich
h
av
e
b
ee
n
h
ig
h
lig
h
ted
in
p
r
ev
io
u
s
s
tu
d
ies
as
well
[
4
1
]
.
W
e
ar
e
in
tr
o
d
u
cin
g
a
m
o
d
el
th
at
will
b
e
a
co
m
b
in
atio
n
o
f
an
a
d
v
a
n
ce
d
R
NN
an
d
C
NN
m
o
d
el
an
d
th
is
will
in
cr
ea
s
e
th
e
a
cc
u
r
ac
y
p
er
ce
n
tag
e
Fig
u
r
e
3
.
Mo
r
e
o
v
er
,
in
th
is
h
y
b
r
i
d
m
o
d
el
we
will
b
e
in
tr
o
d
u
cin
g
co
n
ce
p
t
o
f
h
y
p
e
r
p
ar
am
eter
t
u
n
in
g
o
n
d
ataset
o
f
C
NN
an
d
a
d
v
an
ce
d
R
NN.
So
,
th
is
wil
l
g
iv
e
u
s
a
h
ig
h
er
a
cc
u
r
ac
y
,
wh
ic
h
is
ex
p
ec
ted
t
o
b
e
9
5
% a
n
d
ab
o
v
e.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
4
,
Dec
em
b
er
20
26
:
1
7
4
6
-
1
7
5
8
1750
Fig
u
r
e
3
.
R
ep
r
esen
tatio
n
o
f
h
y
b
r
id
m
o
d
el
3
.
7
.
F
l
o
w
-
cha
rt
o
f
pro
po
s
ed
m
o
del
T
h
is
s
ec
tio
n
co
n
s
is
ts
o
f
a
f
lo
wch
ar
t
th
at
e
x
p
lain
s
th
e
en
tir
e
wo
r
k
f
l
o
w
o
f
o
u
r
m
o
d
el
an
d
h
o
w
th
e
r
esu
lts
ar
e
ac
h
iev
ed
Fig
u
r
e
4
.
Star
tin
g
f
r
o
m
d
ata
lo
a
d
in
g
till
class
if
icat
io
n
o
f
attac
k
s
is
ex
p
lain
ed
i
n
th
is
f
lo
wch
ar
t.
Fig
u
r
e
4
.
W
o
r
k
f
lo
w
o
f
p
r
o
p
o
s
ed
m
o
d
el
3
.
8
.
M
o
del
a
rc
hite
ct
ure
I
n
th
e
y
ea
r
o
f
2
0
2
5
,
s
till
Au
to
n
o
m
o
u
s
ca
r
is
a
n
em
e
r
g
in
g
to
p
ic
a
n
d
w
h
en
we
s
ay
o
r
t
er
m
it
as
E
m
er
g
in
g
it
m
ea
n
s
th
at
lo
t
c
an
b
e
ch
a
n
g
ed
a
n
d
lo
t
ca
n
b
e
im
p
r
o
v
e
d
.
T
h
e
m
o
u
n
ted
eq
u
ip
m
en
t’
s
i.e
.
th
e
ca
m
er
as,
th
e
s
en
s
o
r
s
th
ey
s
tr
u
g
g
le
in
h
ea
v
y
f
o
g
an
d
in
h
illy
ar
ea
s
.
T
h
is
eq
u
ip
m
en
t f
ails
o
r
s
tr
u
g
g
le
in
d
etec
ted
tr
af
f
ic
in
cid
en
ts
,
p
ed
estrian
a
ctiv
ities
[
4
2
]
-
[
4
5
]
.
I
f
we
talk
ab
o
u
t
th
e
h
ig
h
ly
eq
u
ip
p
ed
a
n
d
ad
v
a
n
ce
d
d
ee
p
lear
n
in
g
m
o
d
els
i.e
.
C
NN,
A
NN,
R
NN
s
tr
u
g
g
les
o
r
f
ail
wh
en
th
e
s
ce
n
ar
io
d
if
f
er
s
f
r
o
m
th
e
tr
ain
in
g
d
ata.
T
h
e
s
e
Au
to
n
o
m
o
u
s
ca
r
s
co
m
p
letely
r
ely
o
n
i
n
ter
n
et
wh
ic
h
also
cr
ea
tes
th
r
ea
t
o
f
C
y
b
e
r
-
attac
k
s
[
4
6
]
.
T
h
is
lead
s
to
f
ea
r
am
o
n
g
p
eo
p
le
wh
o
ar
e
s
ca
r
ed
o
f
g
i
v
in
g
f
u
ll c
o
n
t
r
o
l to
m
ac
h
i
n
es.
W
e
ar
e
p
r
o
p
o
s
in
g
a
m
o
d
el
b
y
ad
d
r
ess
in
g
all
th
e
r
esear
ch
g
ap
s
in
th
e
p
r
ev
io
u
s
r
esear
ch
p
ap
er
s
.
Ou
r
m
o
d
el
C
o
m
p
r
is
es
o
f
C
NN+
L
STM
+G
R
U
m
o
d
el
as
s
h
o
wn
i
n
Fig
u
r
e
5
,
in
th
is
L
STM
an
d
GR
U
ar
e
th
e
ty
p
es
o
f
latest
R
NN
m
o
d
els
wh
ich
h
elp
s
in
o
v
er
co
m
in
g
all
th
e
ch
a
llen
g
es
f
ac
ed
b
y
tr
ad
itio
n
al
R
NN.
W
e
h
av
e
u
s
ed
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Dev
elo
p
in
g
a
ma
ch
in
e
le
a
r
n
in
g
-
b
a
s
ed
a
lg
o
r
ith
m
fo
r
s
ec
u
r
ed
r
o
u
te
in
… (
S
h
w
eta
n
s
h
P
r
iya
d
a
r
s
h
i
)
1751
th
ese
m
o
d
els
to
o
v
er
c
o
m
e
th
e
ch
alle
n
g
es
f
ac
e
d
b
y
th
e
o
t
h
er
m
o
d
el
i.e
.
C
NN
f
o
r
f
ea
t
u
r
e
ex
tr
ac
tio
n
with
L
STM
an
d
GR
U
f
o
r
s
eq
u
en
c
e
lear
n
in
g
.
T
h
e
R
esu
lt
attain
e
d
b
y
th
is
m
o
d
el
was
8
5
.
4
7
%
wh
ich
was
n
ea
r
to
ac
cu
r
ac
y
attain
ed
b
y
C
NN
m
o
d
el.
T
h
en
,
in
o
r
d
er
t
o
ac
h
iev
e
b
etter
ac
cu
r
ac
y
,
we
h
a
v
e
d
o
n
e
o
p
tim
izatio
n
u
s
in
g
DM
OA
.
T
h
e
Neu
r
o
n
s
in
ea
ch
lay
er
ar
e
o
p
tim
ized
b
y
u
s
in
g
th
e
DM
OA
.
A
d
ictio
n
ar
y
an
d
C
SV
lo
g
s
ar
e
u
s
ed
to
tr
ac
k
th
e
p
r
e
v
io
u
s
ly
ev
al
u
a
ted
ar
ch
itectu
r
es,
wh
ich
p
r
ev
en
ts
f
r
o
m
r
ed
u
n
d
a
n
t
co
m
p
u
ta
tio
n
s
.
T
h
en
we
ar
e
p
er
f
o
r
m
in
g
m
o
d
el
tr
ain
in
g
b
y
tr
ain
in
g
E
ac
h
ca
n
d
id
ate
m
o
d
el
f
o
r
5
ep
o
ch
s
an
d
u
s
in
g
ea
r
ly
s
to
p
p
in
g
wh
ich
p
r
ev
en
ts
o
v
er
f
itti
n
g
.
Po
s
t
th
at
p
er
f
o
r
m
an
ce
ev
alu
atio
n
is
d
o
n
e
b
y
m
ea
s
u
r
in
g
th
e
Acc
u
r
ac
y
o
n
th
e
test
s
et
to
g
u
id
e
th
e
o
p
tim
izatio
n
p
r
o
ce
s
s
.
Af
ter
o
p
tim
izatio
n
,
th
e
b
est
-
p
er
f
o
r
m
in
g
m
o
d
el
i
s
r
elo
ad
e
d
f
o
r
ev
al
u
atio
n
.
3
.
9
.
E
x
perim
ent
a
l pro
ce
du
r
e
I
n
o
u
r
R
esear
ch
,
we
h
av
e
p
e
r
f
o
r
m
e
d
m
u
ltip
le
s
tep
s
s
tar
tin
g
f
r
o
m
d
ata
co
llectio
n
,
d
ata
ev
alu
atio
n
,
r
em
o
v
al
o
f
n
u
ll
v
alu
es,
p
er
f
o
r
m
in
g
o
v
er
f
itti
n
g
an
d
u
n
d
er
f
itti
n
g
,
co
r
r
elatio
n
an
al
y
s
is
.
An
d
p
o
s
t
th
is
we
ev
alu
ate
all
th
e
ex
is
tin
g
m
o
d
els
an
d
ch
ec
k
f
o
r
th
e
ac
c
u
r
a
cy
.
On
ce
we
ar
e
awa
r
e
o
f
th
e
lo
o
p
h
o
les
in
th
e
ex
is
tin
g
m
o
d
els
th
en
we
h
av
e
p
er
f
o
r
m
ed
t
h
e
p
r
o
ce
s
s
o
n
o
u
r
m
o
d
el
an
d
th
en
ap
p
lied
h
y
p
e
r
p
ar
am
eter
tu
n
in
g
u
s
in
g
DM
OA
.
I
f
we
s
tar
t
f
r
o
m
th
e
v
er
y
b
as
ic
r
eq
u
ir
e
m
en
ts
,
th
en
it
will
b
e
th
e
s
o
f
twar
e
r
e
q
u
ir
em
e
n
ts
.
W
e
u
s
ed
p
y
th
o
n
an
d
An
ac
o
n
d
a.
T
h
is
is
a
p
r
er
e
q
u
is
ite
f
o
r
p
er
f
o
r
m
i
n
g
t
h
is
p
r
o
ce
d
u
r
e,
p
o
s
t
wh
ich
we
m
o
v
ed
to
d
ata
p
r
e
-
p
r
o
ce
s
s
in
g
p
ar
t.
Data
p
r
e
-
p
r
o
ce
s
s
in
g
is
all
ab
o
u
t
p
r
e
p
ar
in
g
th
e
r
aw
d
ata
an
d
m
ak
in
g
it
s
u
itab
le
f
o
r
m
ac
h
in
e
lear
n
in
g
m
o
d
el.
T
h
is
is
th
e
v
e
r
y
f
ir
s
t
s
tep
in
d
ev
elo
p
i
n
g
a
Ma
ch
in
e
L
ea
r
n
in
g
Mo
d
el.
An
d
to
m
ak
e
it
h
ap
p
en
we
n
ee
d
to
im
p
o
r
t
all
th
e
n
ec
ess
ar
y
lib
r
ar
ies
wh
ich
p
er
f
o
r
m
s
th
e
o
p
er
atio
n
s
u
n
d
e
r
Data
Pre
-
p
r
o
ce
s
s
in
g
.
I
n
n
ex
t
s
tep
s
,
p
r
o
ce
s
s
in
g
o
f
C
AN
lo
g
f
iles
h
ap
p
en
s
i.e
.
o
u
tp
u
t
d
ir
ec
to
r
y
g
ets
cr
ea
ted
if
it
d
o
esn
’
t
ex
is
t
an
d
th
en
iter
ates
th
r
o
u
g
h
C
SV
f
iles
in
i
n
p
u
t
d
ir
ec
to
r
y
.
E
ac
h
f
ile
is
r
e
ad
in
wh
ile
h
an
d
lin
g
p
a
r
s
in
g
er
r
o
r
s
,
a
n
d
it
is
also
en
s
u
r
ed
th
at
all
th
e
ex
p
ec
ted
c
o
lu
m
n
s
ex
is
t a
n
d
th
en
clea
n
ed
d
ataset
is
s
av
ed
.
I
n
Fu
r
th
er
s
tep
s
,
class
lab
e
l
s
ar
e
ass
ig
n
ed
i.e
.
all
th
e
p
r
o
ce
s
s
ed
C
SV
f
i
les
ar
e
r
ea
d
an
d
it
is
en
s
u
r
ed
th
at
co
lu
m
n
s
tr
u
ctu
r
e
is
co
n
s
is
ten
t.
I
t
r
ea
d
s
Pro
ce
s
s
ed
C
SV
f
ile
s
,
ad
d
s
a
n
ew
class
C
o
lu
m
n
.
Her
e
Def
au
lt
is
n
o
r
m
al,
s
ets
a
s
p
ec
if
ic
class
lab
els
b
ased
o
n
th
e
f
ile
n
am
e
an
d
th
e
f
la
g
v
alu
e
(
T
f
o
r
Attack
s
)
.
T
h
en
C
o
m
b
in
es
all
d
ataset
in
to
o
n
e
an
d
s
av
es
it.
On
ce
th
is
i
s
d
o
n
e,
th
en
d
ataset
is
co
m
b
in
e
d
an
d
s
av
ed
wh
ic
h
ar
e
in
Hex
ad
ec
im
al
f
o
r
m
.
No
w,
th
e
Hex
ad
ec
im
al
v
alu
es
ar
e
co
n
v
er
ted
to
Dec
im
al
Valu
es
an
d
t
h
en
we
g
et
th
e
u
p
d
ated
d
ataset.
N
o
w,
we
p
er
f
o
r
m
a
s
an
ity
o
n
d
ataset
to
ch
ec
k
if
th
e
s
tr
u
ct
u
r
e
is
p
r
o
p
er
o
r
n
o
t.
I
n
n
ex
t step
s
,
we
p
er
f
o
r
m
class
d
is
tr
ib
u
tio
n
v
is
u
aliza
tio
n
Fig
u
r
e
6
.
Fig
u
r
e
5
.
Ar
c
h
itectu
r
e
o
f
a
h
y
b
r
id
m
o
d
el
-
C
NN+
L
STM
+G
R
U
I
n
n
ex
t
s
tep
s
,
we
en
c
o
d
e
ca
teg
o
r
ical
v
ar
iab
les
(
Flag
,
cla
s
s
)
in
to
n
u
m
er
ical
v
al
u
es
an
d
th
en
we
p
er
f
o
r
m
Featu
r
e
c
o
r
r
elatio
n
a
n
aly
s
is
i.e
.
co
m
p
u
tes
an
d
v
is
u
alize
s
f
ea
tu
r
e
co
r
r
elatio
n
s
as
s
h
o
wn
in
Fig
u
r
e
7
.
Mo
v
in
g
n
ex
t,
we
p
er
f
o
r
m
Activ
ity
o
f
Dr
o
p
p
i
n
g
u
n
n
ec
es
s
ar
y
f
ea
tu
r
es
in
wh
ich
we
r
em
o
v
e
all
th
e
lo
w
r
elev
an
ce
f
ea
tu
r
es.
At
t
h
is
s
tag
e
we
h
av
e
o
u
r
f
in
al
d
ataset
r
ea
d
y
f
o
r
u
s
e,
n
o
w
we
s
ep
a
r
ate
th
e
d
ataset
in
to
test
s
et
an
d
tr
ain
s
et
i.e
.
we
ar
e
s
ep
ar
atin
g
f
ea
tu
r
es
(
x
)
f
r
o
m
la
b
els
(
y
)
.
Her
e,
we
ar
e
s
ep
ar
atin
g
th
e
d
ataset
in
a
p
r
o
p
o
r
tio
n
o
f
8
0
%
f
o
r
tr
ain
in
g
an
d
2
0
%
f
o
r
test
in
g
.
O
n
ce
t
h
is
is
co
m
p
leted
,
we
will
ev
al
u
ate
all
th
e
ex
is
tin
g
m
o
d
els an
d
its
ac
cu
r
ac
y
b
ef
o
r
e
we
m
o
v
ed
to
o
u
r
m
o
d
el.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
4
,
Dec
em
b
er
20
26
:
1
7
4
6
-
1
7
5
8
1752
Fig
u
r
e
6
.
C
lass
d
is
tr
ib
u
tio
n
v
i
s
u
aliza
tio
n
Fig
u
r
e
7
.
Featu
r
e
co
r
r
elat
io
n
a
n
aly
s
is
Po
s
t
th
is
we
test
all
th
e
ex
is
tin
g
m
o
d
els
i.e
.
,
ANN,
C
NN,
R
NN
an
d
o
th
er
Hy
b
r
i
d
Mo
d
els to
test
th
eir
Acc
u
r
ac
y
.
A
n
d
o
n
t
h
e
b
asis
o
f
th
at
we
d
ec
id
ed
th
at
wh
at
ar
e
th
e
n
e
u
r
al
n
etwo
r
k
s
th
at
ca
n
g
iv
e
o
u
r
m
o
d
el
a
b
etter
f
ac
e.
So
,
o
n
ce
we
C
o
m
p
leted
th
e
Valid
atio
n
we
f
o
r
m
ed
a
Hy
b
r
id
Mo
d
el
o
f
C
NN,
L
STM
an
d
GR
U.
C
NN
f
u
lf
ils
o
u
r
n
ee
d
o
f
f
ea
tu
r
e
E
x
tr
ac
tio
n
with
L
STM
an
d
GR
U
m
ee
ts
th
e
r
eq
u
ir
e
m
en
t
o
f
Seq
u
e
n
tial
L
ea
r
n
in
g
.
B
elo
w
is
th
e
R
esu
lt
Attain
ed
f
r
o
m
o
u
r
M
o
d
el
T
ab
l
e
2
.
T
ab
le
2
.
R
esu
lts
ac
h
iev
ed
f
r
o
m
h
y
b
r
id
m
o
d
el
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
sc
o
r
e
S
u
p
p
o
r
t
A
c
c
u
r
a
c
y
0
.
8
5
1
3
2
3
7
6
5
M
a
c
r
o
A
v
e
r
a
g
e
0
.
1
7
0
.
2
0
0
.
1
8
1
3
2
3
7
6
5
W
e
i
g
h
t
e
d
A
v
e
r
a
g
e
0
.
7
3
0
.
8
5
0
.
7
9
1
3
2
3
7
6
5
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Dev
elo
p
in
g
a
ma
ch
in
e
le
a
r
n
in
g
-
b
a
s
ed
a
lg
o
r
ith
m
fo
r
s
ec
u
r
ed
r
o
u
te
in
… (
S
h
w
eta
n
s
h
P
r
iya
d
a
r
s
h
i
)
1753
Sin
ce
th
e
Attain
ed
Acc
u
r
ac
y
was
n
o
t
m
ee
tin
g
o
u
r
r
e
q
u
ir
em
en
t,
we
Per
f
o
r
m
ed
Hy
p
er
-
p
ar
am
eter
T
u
n
in
g
.
T
h
e
Pro
ce
s
s
o
f
Op
tim
izatio
n
with
DM
OA
was
p
er
f
o
r
m
e
d
.
Her
e
h
y
p
er
-
p
a
r
am
et
er
s
(
n
eu
r
o
n
s
in
ea
ch
L
ay
er
)
a
r
e
o
p
tim
ized
u
s
in
g
D
MO
A.
T
h
en
all
t
h
e
E
v
al
u
atio
n
C
o
m
b
in
atio
n
s
ar
e
tr
ac
k
ed
i.
e.
,
a
d
ictio
n
ar
y
an
d
C
SV
lo
g
ar
e
u
s
ed
to
tr
ac
k
p
r
e
v
io
u
s
l
y
e
v
alu
ated
ar
ch
itectu
r
e
s
,
p
r
ev
e
n
tin
g
r
ed
u
n
d
a
n
t
co
m
p
u
tatio
n
s
.
An
d
o
n
ce
th
is
is
d
o
n
e,
we
p
er
f
o
r
m
M
o
d
el
tr
ain
in
g
,
wh
er
e
E
ac
h
ca
n
d
id
ate
m
o
d
el
is
tr
ain
ed
f
o
r
5
ep
o
ch
s
u
s
in
g
ea
r
l
y
s
to
p
p
in
g
to
p
r
ev
en
t
o
v
er
f
itti
n
g
.
At
last
,
Acc
u
r
ac
y
is
m
ea
s
u
r
ed
o
n
th
e
test
s
et
to
g
u
id
e
th
e
o
p
tim
izatio
n
p
r
o
ce
s
s
T
ab
le
3
.
T
ab
le
3
.
R
esu
lts
p
o
s
t h
y
p
er
-
p
a
r
am
eter
tu
n
in
g
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
R
e
c
a
l
l
F1
-
S
c
o
r
e
0
.
9
9
9
7
9
7
0
.
9
9
9
7
9
7
0
.
9
9
9
7
9
7
0
.
9
9
8
2
3
3
3
.
1
0
.
M
a
t
hem
a
t
ic
a
l
ex
press
io
ns
a
nd
s
y
m
bo
ls
Po
s
t
th
at
p
er
f
o
r
m
an
ce
ev
alu
a
tio
n
is
d
o
n
e
b
y
m
ea
s
u
r
in
g
th
e
Acc
u
r
ac
y
o
n
th
e
test
s
et
to
g
u
id
e
th
e
o
p
tim
izatio
n
p
r
o
ce
s
s
.
Af
ter
o
p
tim
izatio
n
,
th
e
b
est
-
p
er
f
o
r
m
i
n
g
m
o
d
el
is
r
elo
ad
ed
f
o
r
ev
al
u
atio
n
as
s
h
o
wn
in
Fig
u
r
e
8
.
Fig
u
r
e
8
.
C
lass
if
icatio
n
p
lo
t f
o
r
o
u
r
h
y
b
r
id
m
o
d
el
E
x
p
lan
atio
n
:
−
: Raw
o
u
tp
u
t
(
lo
g
its
)
f
r
o
m
y
o
u
r
m
o
d
el
f
o
r
class
(
co
m
p
u
te
d
as
=
ℎ
(
)
+
)
−
: E
x
p
o
n
e
n
tial o
f
th
at
v
alu
e
(
m
ak
es e
v
er
y
th
in
g
p
o
s
itiv
e)
−
Den
o
m
in
ato
r
: Su
m
o
f
e
x
p
o
n
e
n
tials
o
f
all
class
s
co
r
es
"
y
"
"
y
"
"
j
"
"
=
"
"
e
^
z
_
j
"
"
∑
_
(
k
=
1
)
^
5
e
^
z
_
k
"
"
,
whe
r
e
z
=
w
_
o
h
^
(
d
)
+
b
_
o
"
th
e
f
in
al
p
r
e
d
ictio
n
v
ec
t
o
r
is
"
ŷ"
"
=
[
"
"
y
"
"
1
"
"
,
"
"
y
"
"
2
"
"
,
"
"
y
"
"
3
"
"
,
"
"
y
"
"
4
"
"
,
"
"
y
"
"
5
"
"
]
,
"
∑
"
y
_
j
"
"
=
1"
z=
W
o
h
(
d
)
+b
o
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2
2
5
2
-
8
7
7
6
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
,
Vo
l.
15
,
No
.
4
,
Dec
em
b
er
20
26
:
1
7
4
6
-
1
7
5
8
1754
Pro
ce
s
s
o
f
lin
ea
r
tr
an
s
f
o
r
m
atio
n
:
ℎ
(
)
: Fin
al
h
id
d
en
lay
e
r
o
u
t
p
u
t (
f
r
o
m
C
NN
+
L
STM
+
GR
U)
: Weig
h
t m
atr
ix
: Bi
as
T
h
is
p
r
o
d
u
ce
s
th
e
lo
g
its
(
z
v
al
u
es)
b
ef
o
r
e
So
f
tMa
x
.
Fin
al
p
r
ed
ictio
n
v
ec
to
r
̂
=
[
1
,
2
,
3
,
4
,
5
]
T
h
is
is
th
e
f
in
al
o
u
tp
u
t
.
E
ac
h
=
p
r
o
b
ab
ilit
y
o
f
class
4.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
4
.
1
.
Why
DM
O
A
?
T
h
e
DM
OA
is
a
v
er
y
latest
m
etah
eu
r
is
tic
o
p
tim
izatio
n
al
g
o
r
ith
m
wh
ic
h
is
to
tally
in
s
p
ir
ed
b
y
t
h
e
f
o
r
ag
in
g
b
e
h
av
io
r
o
f
d
war
f
m
o
n
g
o
o
s
e.
Similar
ly
,
we
h
av
e
o
t
h
er
alg
o
r
ith
m
s
lik
e
g
r
ey
wo
lf
w
h
ale
o
p
tim
izatio
n
,
p
ar
ticle
s
war
m
o
p
tim
izatio
n
alg
o
r
ith
m
,
g
e
n
eti
c
alg
o
r
ith
m
,
an
d
an
t
co
lo
n
y
[
4
7
]
.
W
e
ca
n
n
o
t
s
ay
th
at
DM
OA
is
th
e
m
o
s
t
s
u
p
er
io
r
o
p
tim
izatio
n
Alg
o
r
ith
m
o
f
all
th
e
ex
is
tin
g
o
n
es
f
o
r
d
ee
p
lear
n
in
g
b
u
t
as
p
er
o
u
r
r
e
q
u
ir
em
e
n
t
an
d
u
s
e
it
is
q
u
ite
p
o
wer
f
u
l
f
o
r
co
n
tin
u
o
u
s
o
p
tim
izatio
n
,
tu
n
in
g
o
f
h
y
p
e
r
p
ar
am
eter
s
a
n
d
th
is
alg
o
r
ith
m
h
av
e
s
h
o
wn
b
est
r
e
s
u
lts
in
en
g
in
ee
r
in
g
o
p
tim
izatio
n
task
s
.
T
h
e
d
war
f
m
o
n
g
o
o
s
e
is
s
p
ec
ialized
in
ex
p
lo
r
in
g
,
s
ea
r
ch
in
g
wid
e
r
e
g
io
n
s
o
f
th
e
s
o
l
u
tio
n
s
p
ac
es a
n
d
it f
o
cu
s
es o
n
th
e
b
est ca
n
d
id
at
e
s
o
lu
tio
n
.
4
.
2
.
Co
m
pa
riso
n
t
a
ble
(
K
e
y
F
ind
ing
s
)
I
n
th
is
r
esear
ch
we
h
av
e
n
o
t
o
n
ly
d
e
v
elo
p
ed
a
Hy
b
r
id
m
o
d
el
b
u
t
also
wo
r
k
e
d
o
n
all
th
e
ex
is
tin
g
lo
o
p
h
o
les
f
o
u
n
d
in
p
r
ev
io
u
s
m
o
d
els.
I
n
o
r
d
er
to
d
ev
elo
p
a
Hy
b
r
id
m
o
d
el
wh
ich
is
th
e
co
m
b
in
atio
n
o
f
C
NN+
L
STM
+G
R
U
i.e
.
C
N
Ns
f
o
r
f
ea
tu
r
e
e
x
tr
ac
tio
n
,
with
L
STM
an
d
GR
Us
to
p
r
o
ce
s
s
tem
p
o
r
al
d
ep
en
d
e
n
cies
an
d
s
eq
u
en
tial
L
ea
r
n
in
g
,
we
h
av
e
also
s
tu
d
ie
d
in
d
iv
i
d
u
ally
th
e
C
NN,
ANN,
R
NN,
L
STM
.
T
h
e
in
d
iv
id
u
al
s
tu
d
y
o
f
m
o
d
els
h
e
lp
ed
u
s
i
n
k
n
o
win
g
th
e
p
o
s
itiv
e
p
ar
t
an
d
also
th
e
lo
o
p
h
o
les
wh
ich
m
a
k
es
th
ese
m
o
d
els
o
u
td
ate
d
with
ti
m
e.
As
Au
to
n
o
m
o
u
s
ca
r
s
ar
e
g
r
o
wi
n
g
to
p
ic
an
d
th
er
e
ar
e
lo
t
m
o
r
e
th
in
g
s
to
c
o
m
e,
l
o
t
m
o
r
e
tech
n
o
lo
g
ies
ar
e
ab
o
u
t
to
co
m
e
an
d
m
an
y
m
o
r
e
to
g
e
t
o
u
td
ated
.
So
,
th
is
f
ield
ca
n
n
ev
er
h
av
e
o
p
tim
al
s
o
lu
tio
n
s
f
o
r
its
p
r
o
b
lem
s
,
at
l
ea
s
t in
n
ea
r
f
u
tu
r
e
we
ca
n
n
o
t i
m
ag
i
n
e
it.
4
.
3
.
L
im
it
a
t
io
ns
a
nd
s
co
pe
f
o
r
f
uture
re
s
ea
rc
h
As
it
is
alr
ea
d
y
k
n
o
wn
th
at
A
u
to
n
o
m
o
u
s
ca
r
s
ar
e
em
er
g
i
n
g
to
p
ic
an
d
with
ev
er
y
p
ass
in
g
d
ay
th
in
g
s
,
tech
n
o
lo
g
ies,
in
f
r
astru
ctu
r
e,
m
o
d
els,
lev
el
o
f
Au
to
m
atio
n
,
etc.
is
ch
an
g
in
g
an
d
p
ar
allelly
th
e
ty
p
es
o
f
cy
b
e
r
-
Attack
s
ar
e
also
ch
an
g
in
g
.
So
,
Ou
r
Pro
p
o
s
ed
m
o
d
el
ca
n
b
e
s
u
itab
le
f
o
r
th
e
C
u
r
r
en
t
d
a
y
Scen
ar
io
b
u
t
it
ca
n
n
o
t
b
e
g
u
ar
an
tee
d
th
at
th
is
will
b
e
ab
le
to
m
ee
t
all
th
e
f
u
tu
r
e
C
y
b
er
A
ttack
s
an
d
Up
co
m
in
g
ch
allen
g
es.
Ou
r
Mo
d
el
ca
n
n
o
t
b
e
an
o
p
tim
al
s
o
lu
tio
n
to
all
th
e
Attack
s
i
n
ter
m
s
o
f
Au
to
n
o
m
o
u
s
Veh
icles
an
d
with
n
ew
tech
n
o
lo
g
ies
ar
r
iv
in
g
in
m
ar
k
et
th
is
n
ee
d
s
to
u
p
g
r
ad
ed
.
I
f
w
e
tak
e
ex
am
p
le
o
f
DM
OA
wh
ich
was
lau
n
ch
ed
in
2
0
2
2
an
d
as
th
e
p
er
th
e
cu
r
r
e
n
t
s
ce
n
ar
io
an
d
r
eq
u
ir
em
en
t,
we
f
o
u
n
d
it
to
a
p
er
f
ec
t
m
eta
h
eu
r
is
tic
Alg
o
r
ith
m
.
B
u
t in
co
m
in
g
d
ay
s
th
is
will b
e
d
ef
in
itely
ch
a
n
g
in
g
an
d
t
h
is
m
o
d
el
will b
e
ter
m
ed
as o
u
td
a
ted
.
4
.
4
.
Achiev
ed
re
s
ults
W
e
h
av
e
n
o
t
ju
s
t
f
ix
ed
th
e
c
u
r
r
en
t
Pr
o
b
lem
s
b
u
t
also
ap
p
lied
th
at
latest
alg
o
r
ith
m
i.e
.
,
we
h
av
e
ap
p
lied
o
p
tim
izatio
n
with
Dw
ar
f
Mo
n
g
o
o
s
e
Alg
o
r
ith
m
o
n
Hy
b
r
id
m
o
d
el
as
s
h
o
wn
in
T
a
b
le
4
.
T
h
is
h
elp
ed
u
s
in
attain
in
g
m
u
c
h
h
ig
h
er
ac
c
u
r
ac
y
th
a
n
th
e
p
r
ev
i
o
u
s
ex
is
tin
g
m
o
d
els
as sh
o
wn
in
Fig
u
r
e
9
.
T
h
e
h
y
p
er
p
ar
a
m
eter
tu
n
in
g
p
er
f
o
r
m
e
d
o
v
e
r
th
e
h
y
b
r
id
m
o
d
el
will
twea
k
th
e
p
er
f
o
r
m
a
n
ce
o
f
m
o
d
el
wh
ich
lead
s
to
o
p
tim
al
r
esu
lt
s
as
s
h
o
wn
in
Fig
u
r
e
1
0
.
B
asically
,
wh
at
h
ap
p
en
s
in
s
id
e
h
y
p
er
p
a
r
am
eter
is
r
ea
lly
in
ter
esti
n
g
,
it
is
a
k
in
d
o
f
ex
p
er
im
en
tal
m
eth
o
d
till
th
e
b
est
v
alu
es
ar
e
ac
h
iev
e
d
.
So
,
we
ca
n
ca
ll
a
g
o
o
d
h
y
p
er
p
ar
am
eter
tu
n
in
g
wh
e
n
th
e
o
v
er
all
p
er
f
o
r
m
a
n
ce
f
r
o
m
th
e
Dee
p
lea
r
n
in
g
m
o
d
el
is
h
ig
h
as
p
er
t
h
e
allo
tted
m
etr
ics as sh
o
wn
in
Fig
u
r
e
1
1
.
T
ab
le
4
.
C
o
m
p
a
r
is
o
n
b
etwe
en
3
m
o
d
els
M
o
d
e
l
n
a
m
e
A
c
c
u
r
a
c
y
P
r
e
c
i
s
i
o
n
C
N
N
M
o
d
e
l
0
.
9
6
0
.
9
6
A
N
N
M
o
d
e
l
0
.
8
5
0
.
8
5
H
y
b
r
i
d
M
o
d
e
l
0
.
9
9
0
.
9
9
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
t J I
n
f
&
C
o
m
m
u
n
T
ec
h
n
o
l
I
SS
N:
2252
-
8
7
7
6
Dev
elo
p
in
g
a
ma
ch
in
e
le
a
r
n
in
g
-
b
a
s
ed
a
lg
o
r
ith
m
fo
r
s
ec
u
r
ed
r
o
u
te
in
… (
S
h
w
eta
n
s
h
P
r
iya
d
a
r
s
h
i
)
1755
Fig
u
r
e
9
.
C
o
n
f
u
s
io
n
m
atr
i
x
is
v
is
u
alize
d
u
s
in
g
s
ea
b
o
r
n
h
ea
t
m
ap
to
an
al
y
s
e
p
r
ed
ictio
n
er
r
o
r
s
ac
r
o
s
s
class
e
s
Fig
u
r
e
10
.
C
o
m
p
ar
is
o
n
o
f
d
if
f
er
en
t m
etr
ics,
p
r
ec
is
io
n
,
r
ec
all
an
d
f
1
s
co
r
e
f
o
r
th
e
m
o
d
els
Fig
u
r
e
1
1
.
E
n
d
r
esu
lt f
o
r
p
r
o
p
o
s
ed
m
o
d
el
5.
CO
NCLU
SI
O
N
Dee
p
lear
n
in
g
p
lay
s
an
im
p
o
r
tan
t
r
o
le
in
d
etec
tin
g
co
m
p
l
ex
d
atasets
an
d
d
is
tin
g
u
is
h
in
g
b
etwe
en
attac
k
ed
d
ata
an
d
g
en
u
in
e
d
at
a,
m
ak
in
g
th
em
m
o
r
e
r
eliab
le
,
ac
cu
r
ate,
an
d
f
aster
f
o
r
in
v
e
s
tig
atin
g
is
s
u
es.
I
n
th
is
s
tu
d
y
,
we
ad
d
r
ess
th
e
ch
allen
g
e
o
f
cy
b
er
-
attac
k
s
wh
ich
d
iv
er
ts
th
e
r
o
u
te
o
f
AVs
o
r
lead
s
to
lo
s
s
o
f
life
an
d
in
f
r
astru
ctu
r
e
d
u
e
to
d
ata
tam
p
er
in
g
.
T
h
e
h
y
b
r
id
m
o
d
e
l
wh
ich
we
ar
e
s
u
g
g
esti
n
g
h
er
e
f
o
r
Au
to
n
o
m
o
u
s
v
eh
icles m
ee
ts
th
e
I
SO stan
d
ar
d
.
Evaluation Warning : The document was created with Spire.PDF for Python.