T
E
L
K
O
M
N
I
K
A
T
elec
o
m
m
un
ica
t
io
n Co
m
pu
t
ing
E
lect
ro
nics
a
nd
Co
ntr
o
l
Vo
l.
23
,
No
.
5
,
Octo
b
er
20
25
,
p
p
.
1
291
~1
3
0
3
I
SS
N:
1
6
9
3
-
6
9
3
0
,
DOI
: 1
0
.
1
2
9
2
8
/TE
L
KOM
NI
KA.
v
23
i
5
.
2
6
7
3
8
1291
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//jo
u
r
n
a
l.u
a
d
.
a
c.
id
/in
d
ex
.
p
h
p
/TELK
OM
N
I
K
A
O
ptimi
zed h
uma
n det
e
ction in
NL
O
S sce
na
rio
s usin
g
hybrid
dimens
io
na
lity re
duction a
nd
SVM
wit
h
UWB sig
na
l
s
E
no
ch
Ada
m
a
J
iy
a
1
,
I
lesa
nm
i
B
a
njo
O
luwa
f
em
i
2
,
E
mm
a
nu
el
Su
n
da
y
Ak
in
Aj
is
eg
iri
1
1
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
a
n
d
I
n
f
o
r
mat
i
o
n
E
n
g
i
n
e
e
r
i
n
g
,
C
o
l
l
e
g
e
o
f
E
n
g
i
n
e
e
r
i
n
g
,
La
n
d
mar
k
U
n
i
v
e
r
si
t
y
,
K
w
a
r
a
,
N
i
g
e
r
i
a
2
D
e
p
a
r
t
me
n
t
o
f
El
e
c
t
r
i
c
a
l
a
n
d
El
e
c
t
r
o
n
i
c
E
n
g
i
n
e
e
r
i
n
g
,
F
a
c
u
l
t
y
o
f
E
n
g
i
n
e
e
r
i
n
g
,
E
k
i
t
i
S
t
a
t
e
U
n
i
v
e
r
si
t
y
,
E
k
i
t
i
,
N
i
g
e
r
i
a
Art
icle
I
nfo
AB
S
T
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
Oct
22
,
2
0
2
4
R
ev
is
ed
J
u
l
11
,
2
0
2
5
Acc
ep
ted
Au
g
1
,
2
0
2
5
Trap
p
e
d
v
ict
im
l
o
c
a
li
z
a
ti
o
n
in
se
a
rc
h
a
n
d
re
sc
u
e
(S
AR)
o
p
e
ra
ti
o
n
s
is
e
sp
e
c
ially
d
iffi
c
u
l
t
in
n
o
n
-
li
n
e
-
of
-
sig
h
t
(NLOS)
c
o
n
d
it
i
o
n
s,
w
h
e
re
trad
it
io
n
a
l
tec
h
n
iq
u
e
s
fa
il
d
u
e
to
d
e
b
ris
a
n
d
sig
n
a
l
d
isto
rti
o
n
.
Ultra
-
wi
d
e
b
a
n
d
(UWB)
NLOS
sig
n
a
l
d
a
tas
e
ts
o
ffe
r
a
p
ro
m
isin
g
a
lt
e
r
n
a
ti
v
e
b
u
t
a
re
o
ften
h
ig
h
-
d
ime
n
sio
n
a
l
a
n
d
n
o
isy
.
Th
is
st
u
d
y
p
r
o
p
o
se
s
a
n
o
p
t
imiz
e
d
d
ime
n
sio
n
a
li
ty
re
d
u
c
ti
o
n
fra
m
e
wo
rk
c
o
m
b
i
n
in
g
a
n
a
d
a
p
ti
v
e
h
u
m
a
n
p
re
se
n
c
e
d
e
tec
to
r
(AH
P
D)
with
g
e
n
e
ti
c
a
l
g
o
rit
h
m
s
(G
A)
a
n
d
in
d
e
p
e
n
d
e
n
t
c
o
m
p
o
n
e
n
t
a
n
a
ly
sis
(ICA),
fo
ll
o
we
d
b
y
su
p
p
o
rt
v
e
c
to
r
m
a
c
h
in
e
(S
VM)
c
las
sifica
t
io
n
.
Th
e
a
p
p
ro
a
c
h
is
tes
ted
o
n
a
p
u
b
l
ic
NLOS
d
a
tas
e
t
c
o
m
p
r
isin
g
2
3
,
5
2
2
d
y
n
a
m
ic
in
sta
n
c
e
s,
e
a
c
h
with
2
5
6
si
g
n
a
l
sa
m
p
les
p
e
r
a
tt
rib
u
te,
sim
u
lati
n
g
c
o
m
p
lex
S
AR
sc
e
n
a
rio
s
in
c
lu
d
in
g
r
u
b
b
le
a
n
d
d
y
n
a
m
ic
o
b
sta
c
les
.
Th
e
re
su
l
ts
in
d
ica
te
th
a
t
t
h
e
AH
P
D+
G
A+
S
VM
m
o
d
e
l
re
a
c
h
e
d
a
n
a
c
c
u
ra
c
y
o
f
8
5
.
7
8
%
,
se
n
siti
v
it
y
o
f
8
0
.
0
0
%
,
a
n
d
sp
e
c
ifi
c
it
y
o
f
9
6
.
4
6
%
,
w
h
ich
is
b
e
tt
e
r
th
a
n
t
h
e
A
HPD+
ICA
+
S
VM
m
o
d
e
l
th
a
t
h
a
d
a
n
a
c
c
u
ra
c
y
o
f
7
9
.
2
0
%
,
se
n
siti
v
it
y
o
f
7
3
.
0
7
%
,
a
n
d
sp
e
c
ifi
c
it
y
o
f
8
1
.
0
5
%
.
T
h
e
se
fin
d
in
g
s
d
e
m
o
n
stra
te
th
e
fr
a
m
e
wo
rk
’
s
ro
b
u
stn
e
ss
a
n
d
sc
a
lab
il
it
y
,
m
a
k
in
g
it
a
str
o
n
g
c
a
n
d
id
a
te
f
o
r
re
a
l
-
ti
m
e
h
u
m
a
n
d
e
tec
ti
o
n
in
d
isa
ste
r
re
c
o
v
e
ry
m
is
sio
n
s.
K
ey
w
o
r
d
s
:
Dim
en
s
io
n
ality
No
n
-
lin
e
-
of
-
s
ig
h
t
Sear
ch
an
d
r
escu
e
Su
p
p
o
r
t
v
ec
to
r
m
ac
h
in
e
Ultr
a
-
wid
eb
an
d
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
:
E
n
o
ch
Ad
a
m
a
J
iy
a
Dep
ar
tm
en
t o
f
E
lectr
ical
an
d
I
n
f
o
r
m
atio
n
E
n
g
in
ee
r
in
g
,
C
o
lleg
e
o
f
E
n
g
in
ee
r
i
n
g
,
L
a
n
d
m
ar
k
Un
iv
er
s
ity
Km
.
4
I
p
et
u
R
o
ad
,
PMB
1
0
0
1
,
Om
u
-
ar
an
,
Kwa
r
a
,
Nig
er
ia
E
m
ail:
jiy
a.
ad
am
a@
lm
u
.
ed
u
.
n
g
1.
I
NT
RO
D
UCT
I
O
N
A
n
o
tab
le
c
h
allen
g
e
in
th
e
an
aly
s
is
o
f
s
ea
r
ch
a
n
d
r
escu
e
(
S
AR
)
o
p
er
atio
n
s
lies
in
th
e
a
g
g
r
eg
atio
n
o
f
in
f
o
r
m
atio
n
o
r
ig
i
n
atin
g
f
r
o
m
f
ast,
in
d
ep
en
d
e
n
t,
an
d
s
ev
e
r
e
-
im
p
ac
t
s
o
u
r
ce
s
[
1
]
.
A
p
r
o
m
i
n
en
t
ch
ar
ac
ter
is
tic
o
f
th
e
d
ataset
is
its
ex
ten
s
iv
e
s
am
p
le
s
ize,
wh
ich
en
co
m
p
a
s
s
es
a
co
n
s
id
er
ab
le
q
u
an
tity
o
f
r
ed
u
n
d
a
n
t
an
d
s
u
p
er
f
lu
o
u
s
n
o
is
y
f
ea
tu
r
es
[
2
]
,
[
3
]
.
T
h
e
p
r
esen
ce
o
f
th
ese
ar
b
itra
r
y
in
p
u
t
s
ig
n
als
d
etr
ac
ts
f
r
o
m
th
e
e
f
f
icac
y
o
f
class
if
icatio
n
-
b
ased
lear
n
in
g
ap
p
r
o
ac
h
es
[
4
]
.
Me
th
o
d
s
f
o
r
m
i
n
im
izin
g
d
im
en
s
io
n
ality
h
a
v
e
b
ee
n
em
p
lo
y
ed
o
n
n
u
m
er
o
u
s
o
cc
asio
n
s
.
E
x
tr
ac
t
in
g
r
elev
a
n
t
d
is
cr
im
in
ativ
e
s
u
b
s
ets
f
r
o
m
th
e
d
ata
r
e
p
r
esen
tatio
n
r
ed
u
ce
s
co
m
p
u
tatio
n
al
d
em
an
d
s
an
d
e
n
h
an
ce
s
th
e
p
r
ec
is
io
n
o
f
class
i
f
icatio
n
p
r
e
d
ictio
n
s
[
5
]
.
I
n
th
e
r
ea
lm
o
f
d
ataset
an
aly
s
is
,
th
e
p
h
en
o
m
en
a
o
f
m
o
d
el
o
v
er
f
itti
n
g
a
n
d
is
s
u
es
r
elate
d
to
h
ig
h
-
d
im
en
s
io
n
al
d
ata
ar
e
r
ec
o
g
n
is
ed
as
s
ig
n
if
ica
n
t
o
b
s
tacle
s
to
o
p
tim
al
class
if
icatio
n
p
e
r
f
o
r
m
an
ce
.
T
h
e
s
tr
u
ctu
r
e
co
n
s
is
ts
o
f
a
h
ig
h
-
d
im
en
s
io
n
al
in
p
u
t
s
p
ac
e,
o
f
te
n
d
e
n
o
ted
as
th
e
cu
r
s
e
o
f
d
im
en
s
i
o
n
ality
.
Nu
m
er
o
u
s
d
im
en
s
io
n
ality
r
ed
u
ctio
n
s
tr
ateg
ies
h
av
e
b
ee
n
u
s
ed
in
th
e
liter
atu
r
e
to
o
v
er
c
o
m
e
th
e
cu
r
s
e
o
f
d
im
en
s
io
n
ality
ch
allen
g
es
[
6
]
.
T
o
u
n
co
v
er
h
i
d
d
en
elem
e
n
ts
an
d
im
p
r
o
v
e
t
h
e
in
ter
p
r
eta
b
ilit
y
o
f
th
e
im
ag
es,
th
e
b
est
f
ea
tu
r
e
co
m
b
in
atio
n
m
u
s
t
b
e
ch
o
s
en
[
7
]
,
[
8
]
.
T
h
e
g
o
al
o
f
d
im
e
n
s
io
n
ality
r
ed
u
ctio
n
is
to
id
en
tif
y
a
tr
iv
ial
s
u
b
s
et
o
f
d
ata
th
at
m
ay
im
p
r
o
v
e
p
r
ed
ictio
n
p
er
f
o
r
m
an
ce
,
wh
ich
w
ill
aid
en
g
in
ee
r
s
an
d
SAR
i
n
lo
ca
lis
atio
n
an
d
d
ec
is
io
n
-
m
ak
in
g
[
9
]
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
TEL
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
23
,
No
.
5
,
Octo
b
e
r
20
25
:
1
2
9
1
-
1
3
0
3
1292
T
h
e
is
s
u
es
s
u
r
r
o
u
n
d
in
g
th
e
cu
r
s
e
o
f
d
im
e
n
s
io
n
ality
h
a
v
e
b
ee
n
co
v
e
r
ed
b
y
n
u
m
er
o
u
s
wr
iter
s
.
Alth
o
u
g
h
m
eta
h
eu
r
is
tics
h
av
e
also
b
ee
n
p
r
esen
ted
,
m
eth
o
d
s
f
o
r
co
llectin
g
n
o
-
lin
e
-
of
-
s
ig
h
t
(
NL
OS)
s
ig
n
al
d
ata
s
u
b
s
ets
ar
e
ch
allen
g
e
d
b
y
s
tr
o
n
g
c
o
r
r
elatio
n
s
,
h
ig
h
p
r
o
ce
s
s
in
g
r
ates,
an
d
p
r
o
l
o
n
g
ed
co
m
p
u
tatio
n
d
u
r
atio
n
s
[1
0
]
,
[
1
1
]
.
Me
th
o
d
i
ca
lly
id
en
tify
i
n
g
t
h
e
id
ea
l
s
u
b
s
et
o
f
d
ata
attr
ib
u
tes
is
a
m
ajo
r
p
r
o
b
lem
.
T
h
e
wid
ely
r
ec
o
g
n
ized
m
eth
o
d
s
f
o
r
r
ed
u
cin
g
d
im
en
s
io
n
ality
in
f
ea
tu
r
e
ex
tr
ac
tio
n
(
FE)
[1
2
]
,
[
1
3
]
(
s
u
p
er
v
is
ed
a
n
d
u
n
s
u
p
er
v
is
ed
)
an
d
f
ea
t
u
r
e
s
el
ec
tio
n
s
tr
ateg
ies,
in
clu
d
i
n
g
f
ilter
,
wr
ap
p
e
r
,
a
n
d
em
b
ed
d
ed
m
eth
o
d
s
[1
4
]
h
av
e
r
eso
lv
ed
p
e
r
f
o
r
m
an
ce
en
h
an
c
em
en
t;
h
o
wev
e
r
,
f
u
r
th
er
a
d
v
a
n
ce
m
en
ts
in
h
y
b
r
id
m
o
d
els
a
n
d
o
p
tim
is
atio
n
ar
e
r
eq
u
ir
ed
to
ac
h
iev
e
ev
e
n
b
ette
r
r
esu
lts
[1
5
]
-
[
17
]
.
T
h
e
o
b
j
e
c
t
i
v
e
i
s
t
o
i
d
e
n
t
i
f
y
t
h
e
b
e
s
t
s
e
l
e
ct
e
d
p
o
r
t
i
o
n
o
f
d
a
t
a
a
t
t
r
i
b
u
t
es
a
i
m
e
d
at
m
a
n
a
g
i
n
g
h
i
g
h
-
d
i
m
e
n
s
i
o
n
a
l
o
p
t
i
m
i
s
at
i
o
n
p
r
o
b
l
e
m
s
a
n
d
p
r
o
v
i
d
e
f
e
a
s
i
b
l
e
s
o
l
u
t
i
o
n
s
[
6
]
,
[
1
3
]
.
Alth
o
u
g
h
GA
h
as
b
ee
n
wid
ely
u
tili
s
ed
an
d
is
g
o
o
d
at
i
d
en
tif
y
in
g
th
e
b
est
-
p
er
f
o
r
m
in
g
f
ea
tu
r
e
g
r
o
u
p
s
with
in
h
ig
h
-
d
im
en
s
io
n
al
d
ata
s
ets,
it
is
co
m
p
u
tatio
n
ally
c
o
s
tly
an
d
s
u
s
ce
p
tib
le
to
o
v
er
f
itt
in
g
.
T
o
g
et
ar
o
u
n
d
th
is
lim
itatio
n
,
o
p
tim
is
atio
n
tech
n
iq
u
es
h
av
e
b
ee
n
ap
p
lie
d
to
p
r
o
v
id
e
im
p
r
o
v
ed
o
u
tc
o
m
es
in
ter
m
s
o
f
ch
o
o
s
in
g
th
e
b
est
s
elec
ted
f
ea
tu
r
e
g
r
o
u
p
s
an
d
th
e
p
r
ec
is
io
n
o
f
class
if
icatio
n
[
18
]
.
A
leg
itima
te
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
e
th
at
h
as
b
ee
n
wid
ely
u
s
ed
as
a
ca
p
ab
le
s
tan
d
ar
d
m
eth
o
d
f
o
r
e
x
tr
ac
tin
g
g
r
o
u
p
s
o
f
f
ea
tu
r
e
s
am
p
les
u
s
ed
f
o
r
class
if
icatio
n
p
u
r
p
o
s
es
is
in
d
ep
en
d
en
t
co
m
p
o
n
en
t
an
aly
s
is
(
I
C
A)
(
l
in
ea
r
)
,
wh
ich
h
as
r
ec
en
tly
attr
ac
ted
m
o
r
e
atten
ti
o
n
[
19
]
.
T
h
e
h
y
b
r
i
d
ap
p
r
o
ac
h
’
s
r
em
ar
k
ab
le
r
esu
lts
an
d
ad
v
a
n
tag
es
d
em
o
n
s
tr
ate
its
v
alu
e
in
ad
d
r
ess
in
g
d
im
en
s
io
n
al
is
s
u
es
th
at
h
in
d
er
class
if
icatio
n
.
I
d
e
n
tify
in
g
o
r
ca
teg
o
r
izin
g
NL
OS
s
ig
n
al
d
ata
an
d
th
e
an
aly
s
is
o
f
ex
p
r
e
s
s
io
n
d
ata
d
e
p
en
d
o
n
th
e
cr
ea
t
io
n
o
f
ef
f
ec
tiv
e
m
o
d
els
th
at
ar
e
s
im
p
le
to
u
s
e
an
d
co
m
p
u
te
q
u
ick
ly
[2
0
]
.
Nu
m
er
o
u
s
s
tu
d
ies
h
av
e
b
ee
n
co
n
d
u
cte
d
an
d
r
ep
o
r
te
d
in
th
e
liter
atu
r
e
[2
1
]
,
[
2
2
]
.
Ho
wev
e
r
,
g
iv
e
n
th
e
p
r
ev
alen
ce
o
f
b
u
ild
i
n
g
co
llap
s
es
an
d
tr
ap
p
ed
v
ictim
s
in
W
es
t
Af
r
ica,
th
ese
s
tu
d
ies
n
ee
d
to
b
e
im
p
r
o
v
ed
to
aid
in
m
ak
in
g
d
ec
is
io
n
s
r
eg
ar
d
i
n
g
th
e
r
ed
u
ctio
n
o
f
v
ictim
m
o
r
ta
lity
in
th
e
r
eg
io
n
[2
3
]
.
T
h
e
c
o
m
m
o
n
ly
em
p
lo
y
e
d
co
n
v
en
tio
n
al
tar
g
et
lo
ca
tio
n
an
d
class
if
icatio
n
(
T
L
C
)
m
e
th
o
d
s
d
e
p
en
d
h
ea
v
ily
o
n
u
n
d
er
s
tan
d
in
g
s
ig
n
al
b
eh
av
io
r
a
n
d
s
u
r
r
o
u
n
d
in
g
en
v
ir
o
n
m
e
n
tal
co
n
d
itio
n
s
,
d
esp
ite
th
eir
ef
f
ec
tiv
en
ess
in
co
n
tr
o
lled
s
itu
atio
n
s
.
B
ec
au
s
e
th
e
m
an
u
al
ca
lib
r
atio
n
p
r
o
ce
d
u
r
e
ca
n
o
cc
asio
n
all
y
b
e
tim
e
-
co
n
s
u
m
in
g
,
it
is
n
o
t
ap
p
r
o
p
r
iate
f
o
r
er
r
atic,
ex
tr
em
e
s
itu
atio
n
s
lik
e
ea
r
th
q
u
ak
e
d
eb
r
is
.
I
n
ad
d
it
io
n
,
T
L
C
m
o
d
es
e
x
h
ib
it
r
e
d
u
ce
d
p
er
f
o
r
m
an
ce
,
esp
ec
ially
wh
en
d
is
tin
g
u
is
h
in
g
lo
w
-
r
e
f
lectiv
e
o
r
s
tatio
n
ar
y
tar
g
ets,
s
in
ce
th
ey
lack
th
e
s
o
p
h
is
ticated
m
eth
o
d
s
f
o
r
elim
in
atin
g
o
r
m
in
im
izi
n
g
s
ig
n
al
in
ter
f
er
en
ce
n
ee
d
e
d
to
m
an
ag
e
th
e
g
r
ea
ter
d
eg
r
ee
o
f
o
b
s
tr
u
ctio
n
n
atu
r
ally
p
r
esen
t
in
n
o
n
-
lin
e
-
of
-
s
ig
h
t
s
ce
n
ar
io
s
.
T
o
m
o
r
e
cl
o
s
ely
r
esem
b
le
h
u
m
a
n
-
g
e
n
er
a
ted
co
n
ten
t,
it
m
u
s
t
m
ak
e
s
ev
er
al
ch
a
n
g
es th
at
ca
n
ad
d
co
m
p
lex
ity
a
n
d
d
i
v
er
s
ity
.
T
o
an
ticip
ate
NL
OS
d
ata
s
ig
n
als,
th
is
p
ap
e
r
s
u
g
g
ests
a
h
y
b
r
id
d
im
en
s
io
n
ality
r
ed
u
ctio
n
m
eth
o
d
.
Hy
b
r
id
s
y
s
tem
s
o
u
tp
er
f
o
r
m
e
d
co
n
v
en
tio
n
al
m
eth
o
d
s
b
ase
d
o
n
s
ce
n
ar
io
-
s
p
ec
if
ic
p
a
r
am
eter
is
atio
n
in
ea
ch
in
s
tan
ce
.
I
n
cr
ea
s
ed
ad
a
p
tab
ilit
y
ac
r
o
s
s
a
r
an
g
e
o
f
s
ettin
g
s
will
r
esu
lt
f
r
o
m
th
e
u
s
e
in
v
o
lv
in
g
ad
a
p
tiv
e
n
o
is
e
s
u
p
p
r
ess
io
n
,
s
elf
-
ad
ju
s
tin
g
p
a
r
am
eter
tu
n
in
g
,
an
d
c
o
n
tin
u
o
u
s
f
ea
tu
r
e
o
p
tim
is
atio
n
.
W
h
e
n
co
m
p
ar
e
d
to
o
th
e
r
s
y
s
tem
s
,
th
e
h
y
b
r
id
s
o
lu
tio
n
s
h
o
wn
h
er
e
h
as
s
ev
er
al
ad
v
an
ta
g
es
o
v
er
tr
ad
itio
n
al
tr
ac
k
i
n
g
l
o
ca
lis
atio
n
s
ettin
g
s
.
Ho
wev
er
,
th
e
s
u
g
g
ested
u
n
d
e
r
-
r
u
b
b
le
ad
ap
tiv
e
h
u
m
an
p
r
esen
ce
d
etec
to
r
(
AHPD)
m
eth
o
d
elim
in
ates
th
e
n
ee
d
f
o
r
i
n
tr
icate
m
ath
e
m
atica
l
ca
l
cu
latio
n
s
o
r
p
a
r
am
eter
s
ea
r
ch
es
b
y
co
m
b
i
n
in
g
a
g
en
etic
al
g
o
r
ith
m
(
GA)
with
I
C
A
to
f
lex
ib
ly
r
esp
o
n
d
to
v
a
r
iatio
n
s
in
e
n
v
ir
o
n
m
en
tal
c
o
n
d
itio
n
s
.
T
h
is
is
in
lin
e
with
n
e
w
r
esear
ch
f
in
d
in
g
s
th
at
em
p
h
asis
e
th
e
n
ec
ess
ity
o
f
m
ac
h
in
e
lear
n
in
g
-
in
f
u
s
ed
f
le
x
ib
le
m
o
d
els
to
attain
h
ig
h
er
g
en
er
alis
atio
n
s
in
a
r
an
g
e
o
f
s
itu
atio
n
s
.
Fo
llo
win
g
th
e
m
eth
o
d
s
,
th
e
GA
an
d
I
C
A
ar
e
ap
p
lied
af
ter
p
er
tin
en
t
d
ata
s
u
b
s
ets
ar
e
ex
tr
a
cted
u
s
in
g
an
AHPD
p
s
eu
d
o
co
d
e
th
at
f
ilter
s
o
u
t
n
o
is
e
an
d
p
e
r
m
its
au
to
m
ated
ad
ju
s
tm
en
ts
in
a
m
p
litu
d
e
to
id
e
n
tify
h
id
d
en
co
m
p
o
n
en
ts
.
AHPD
with
GA
an
d
AHPD
with
I
C
A
co
m
b
in
atio
n
s
u
n
d
er
r
u
b
b
le
ar
e
class
if
ied
u
s
in
g
s
u
p
p
o
r
t
v
ec
to
r
m
ac
h
in
es
(
SVM)
o
n
an
NL
OS
d
ataset.
T
o
h
elp
en
g
in
ee
r
s
an
d
SAR
team
s
m
ak
e
b
etter
ju
d
g
m
en
ts
,
th
is
ef
f
o
r
t
s
ee
k
s
to
m
in
im
ize
ch
allen
g
es
in
p
r
ed
ictio
n
,
in
cl
u
d
in
g
c
o
m
p
u
tatio
n
al
ex
p
en
s
es,
o
b
tain
in
g
p
er
tin
e
n
t
p
o
r
tio
n
s
o
f
th
e
d
ataset,
an
d
in
ter
r
elatio
n
s
h
ip
s
am
o
n
g
v
ar
ia
b
les.
T
h
e
o
t
h
er
p
o
r
tio
n
s
o
f
t
h
is
s
tu
d
y
co
n
s
is
t
o
f
ex
is
tin
g
r
ese
ar
ch
,
r
ele
v
an
t
m
ater
ials
,
a
n
d
m
eth
o
d
o
l
o
g
y
,
th
e
f
in
d
in
g
s
,
d
i
s
cu
s
s
io
n
s
,
an
d
th
e
co
n
clu
s
io
n
s
.
2.
ADAP
T
I
VE
A
L
G
O
RI
T
H
M
F
O
R
H
UM
AN
P
RE
SE
N
CE
DE
T
E
C
T
I
O
N
I
N
UND
E
R
-
R
UB
B
L
E
E
NVI
RO
NM
E
N
T
S
T
h
e
em
itted
u
ltra
-
wid
eb
a
n
d
(
UW
B
)
p
u
ls
e
i
s
s
ig
n
if
ican
tly
wea
k
en
ed
,
alter
ed
,
a
n
d
b
o
u
n
ce
d
b
ac
k
s
ev
er
al
tim
es
d
u
e
t
o
th
e
n
atu
r
e
o
f
th
e
r
u
b
b
le.
R
ad
ar
f
in
d
s
it
ch
allen
g
in
g
to
d
etec
t
s
m
al
l
f
ea
tu
r
es
am
id
th
e
n
o
is
e
b
ec
au
s
e
o
f
th
ese
is
s
u
es.
Pre
p
r
o
ce
s
s
in
g
th
e
r
aw
d
ata
g
ath
er
ed
f
r
o
m
th
e
e
n
v
ir
o
n
m
e
n
t
u
s
in
g
p
r
o
ce
s
s
in
g
tech
n
iq
u
es
is
th
e
f
ir
s
t
s
tep
in
r
em
o
v
in
g
th
ese
b
ar
r
ier
s
.
Fo
r
t
h
e
p
r
o
p
o
s
ed
s
tu
d
y
,
Alg
o
r
ith
m
1
s
h
o
ws
a
f
lex
ib
le,
d
y
n
am
ic
s
y
s
tem
f
o
r
d
etec
tin
g
h
u
m
an
p
r
esen
ce
.
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
Op
timiz
ed
h
u
ma
n
d
etec
tio
n
in
N
LOS
s
ce
n
a
r
io
s
u
s
in
g
h
yb
r
id
d
imen
s
io
n
a
lity
…
(
E
n
o
ch
A
d
a
ma
Jiya
)
1293
Algorithm 1
.
Proposed
flexible adaptive human presence detector
Step 1: Main loop
First step: 512 * n_pulses
First step: 256 n_samples
Step 1.3: dtpulse = 0.014
In step 1.4, breath_freq = [0.2, 0.7].
Step 1.5: amp_thresh = 0.5
Step 2: Determine the amplitude
’
s minimal threshold
Step 2.1: noise_thresh = 0.3
In step 3.0, set the noise threshold.
Step 3.1: The NFFT is 1024.
Step 3.2: For n_measures, use 10.
Step 4.0: Replace the actual quantity of measurements
Step 4: actual_targets = np.random.choice([0,1],n_measures)
Step 5: Replace the existing target values
In Step 5.1, CF = np.zeros((2,2)).
Step 6.0: The feature of noise filtering
In step 6.1, define
filter_noise(data,noise_thresh):
If noise_thresh > np.max(data) in step 6.2, then
6.3: filtered_data = data
-
np.median(data)
Step 6.4: Should it happen that
Data = filtered_data in step 6.5
In step 7.6, return filtered_data.
Step 7.0: The method by which the measure detects the presence of humans
Step 7.1 is defined as Def
verify_human_presence_in_measure(measure,amp_thresh,breath_freq,noise_thresh,n_pulses,dtpul
se,NFFT,n_samples).
Step 8.0: Subtract the row and column averages.
Measure in Step 8.
1 [:,np.newaxis]
-
Mdiff_measure
-
np.mean(axis=0), + np.mean(measure),
np.mean(measure,axis=1)
In step 9.0, apply noise filtering.
Step 9.1: Mfiltered_measure = np.zeros_like(Mdiff_measure)
Step 9.2: For i in range(Mdiff_measure.shape[1]),
The formula for step 9.3 is Mfiltered_measure[:,i] =
filter_noise(Mdiff_measure[:,i],noise_thresh).
Step 10.0: Use the FFT to determine the amplitude spectrum
Step 10.1: n_samples = Mfft_measure* / np.abs(np.fft.fft(Mfiltered_measure,NFFT,axis=0))
Step 11.0: Find t
he maximum amplitude and pulse index
Step 11.1: max_amp,pulse_idxmax_amp = np.max(Mfft_measure),np.argmax(Mfft_measure)
Step 11.2: i,_ = np.unravel_index(pulse_idxmax_amp,Mfft_measure.shape)
Step 12.0: Ascertain whether the potential target
Step 12.1 if max_amp > amp_thresh
Step 12.2: pred_freq = (i
-
1) / (n_pulses * dtpulses)
In case breath_freq[0] <= pred_freq <= breath_freq[1], as stated in clause 12.3,
Step 12.4: Respond Truthfully
Step12.5: Return the false
Step 13.0: Update the Confusion Mat
rix
Step 13.2: def update_cf(predicted_target,actual_target,cf) if actual_target:
13.3: Is predicted_target supposed to be true?
Step 13.4: Increase cf[0,0] by 1
Step 13.5: Should it happen that
Step 13.6: cf[1,0] + 1
Step 13.7: Should it happen that
13.8: Should predicted_target come true:
Step 13.9: Increase cf[0,1] by 1
Step 13.10: In the absence of
13.11 Step: 1 + cf[1,1]
Step 13.12: Return with vigor cf
Step 13.13: For k in range(n_measures):
Step 13.14: Measure = np.random.rand(n_pulses,n
_samples)
Step 14.0: Use authentic measurement information
In step 14.1, actual_target = actual_targets[k].
The presence of humans in the measure
(measure,amp_thresh,breath_freq,noise_thresh,n_pulses,dtpulse,NFFT,n_samples) is confirmed
by the predicted_target function at step 14.2.
14.3: CF = update_cf(predicted_target,actual_target,CF)
In step 14.4, print("Confusion Matrix:").
In step 14.5, print(CF).
Fro
m
s
ev
er
al
p
h
ases
o
f
a
p
p
r
o
ac
h
th
at
ass
ess
th
e
p
o
s
s
ib
il
ity
o
f
h
u
m
an
p
r
esen
ce
,
th
e
s
u
g
g
ested
m
eth
o
d
f
in
d
s
h
u
m
an
w
h
er
ea
b
o
u
ts
.
T
h
e
alg
o
r
ith
m
ca
r
ef
u
lly
an
aly
s
es
ev
er
y
m
ea
s
u
r
em
e
n
t
to
ascer
tain
th
e
p
r
o
b
a
b
ilit
y
o
f
h
u
m
an
p
r
esen
c
e
af
ter
n
o
is
e
f
ilter
in
g
.
T
h
is
th
o
r
o
u
g
h
a
n
aly
s
is
g
u
ar
a
n
tees
th
e
d
etec
tio
n
s
y
s
tem
’
s
r
esil
ien
ce
an
d
ef
f
icac
y
in
a
r
a
n
g
e
o
f
s
itu
atio
n
s
an
d
ap
p
licatio
n
s
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
TEL
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
23
,
No
.
5
,
Octo
b
e
r
20
25
:
1
2
9
1
-
1
3
0
3
1294
An
I
n
tel
C
o
r
e
5
C
PU
co
n
s
is
t
in
g
o
f
a
6
4
-
b
it
o
p
er
atin
g
s
y
s
t
em
an
d
1
6
GB
o
f
R
AM
is
u
s
ed
f
o
r
th
e
test
in
g
in
th
is
r
esear
ch
.
T
h
e
alg
o
r
ith
m
was
wr
itten
u
s
in
g
th
e
Py
th
o
n
e
n
v
ir
o
n
m
en
t.
T
o
v
er
if
y
co
m
p
a
r
ab
le
tr
ain
in
g
an
d
test
in
g
p
er
f
o
r
m
a
n
ce
s
o
f
th
e
ex
p
er
im
en
ts
in
t
er
m
s
o
f
ac
cu
r
ac
y
an
d
s
en
s
itiv
ity
,
am
o
n
g
o
th
e
r
m
etr
ics,
th
e
co
n
f
u
s
io
n
m
atr
ice
s
wer
e
em
p
lo
y
ed
as th
e
class
if
icatio
n
ev
alu
atio
n
[
7
]
.
3.
M
AT
E
R
I
AL
S AN
D
M
E
T
H
O
D
T
h
is
s
ec
tio
n
ex
p
lain
s
th
e
m
eth
o
d
o
lo
g
y
u
s
ed
f
o
r
th
e
in
v
esti
g
atio
n
an
d
c
o
n
cu
r
r
en
tly
.
G
iv
es
an
all
-
in
clu
s
iv
e
d
is
cu
s
s
io
n
o
f
th
e
m
ater
ials
u
s
ed
.
Fig
u
r
e
1
p
r
esen
ts
th
e
tech
n
iq
u
e
f
lo
w
ch
ar
t
.
W
h
ile
T
ab
le
1
d
is
p
lay
s
th
e
d
ataset
d
escr
ip
tio
n
.
Fig
u
r
e
1
.
Pro
p
o
s
ed
u
n
d
er
-
r
u
b
b
le
AHPD
f
o
r
NL
OS d
ata
h
u
m
an
d
etec
tio
n
3
.
1
.
Da
t
a
s
et
T
h
e
in
v
esti
g
atio
n
m
ad
e
u
s
e
o
f
th
e
Un
iv
er
s
ity
o
f
B
o
lo
g
n
a
’
s
f
r
ee
ly
av
ailab
le
NL
OS
s
ig
n
al
d
ataset
co
llectio
n
f
o
r
h
u
m
an
d
etec
tio
n
b
u
r
ied
u
n
d
e
r
d
eb
r
is
.
I
t
is
co
m
p
o
s
ed
o
f
v
ar
io
u
s
ty
p
es
o
f
d
e
tr
itu
s
,
r
an
g
in
g
f
r
o
m
g
lass
ce
m
en
t
-
b
ased
m
ater
ials
an
d
wo
o
d
,
a
n
d
a
n
o
r
ie
n
t
e
d
h
u
m
an
b
o
d
y
co
n
s
is
tin
g
2
5
6
s
am
p
les
an
d
2
3
,
5
5
2
o
cc
u
r
r
e
n
ce
s
as d
escr
ib
ed
in
T
a
b
le
1
.
T
ab
le
1
.
Featu
r
es
of
th
e
d
atase
t
D
a
t
a
s
e
t
I
n
st
a
n
c
e
s
To
t
a
l
n
u
m
b
e
r
o
f
s
a
mp
l
e
s
c
o
l
l
e
c
t
e
d
p
e
r
o
b
serv
e
d
w
i
n
d
o
w
D
y
n
a
mi
c
r
a
d
a
r
v
a
l
u
e
s
2
3
,
5
5
2
2
5
6
3
.
2
.
M
et
ho
do
lo
g
y
SAR
,
o
r
th
r
o
u
g
h
-
wall
h
u
m
a
n
d
etec
tio
n
,
is
a
wid
ely
u
s
ed
m
eth
o
d
f
o
r
lo
ca
tin
g
v
ictim
s
,
p
ar
ticu
lar
ly
th
o
s
e
h
id
d
en
b
y
d
e
b
r
is
.
T
h
e
a
b
ilit
y
to
d
etec
t
s
ev
er
al
asp
ec
ts
o
f
h
id
d
e
n
d
ata
is
a
k
ey
elem
e
n
t
o
f
cu
r
r
en
t
h
ig
h
-
th
r
o
u
g
h
p
u
t
g
en
etic
f
ac
to
r
tech
n
o
lo
g
ies.
I
t
r
ea
cts
to
v
ar
io
u
s
tr
ain
in
g
s
o
lu
tio
n
s
an
d
s
itu
atio
n
s
,
g
en
er
atin
g
s
u
f
f
icien
t
s
eq
u
en
ci
n
g
d
ata
[2
4
]
,
a
n
d
it
d
etec
ts
s
u
b
tle
c
h
an
g
es
o
cc
u
r
r
in
g
in
t
h
e
c
o
n
d
itio
n
s
o
f
b
u
r
ied
v
ictim
s
,
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
Op
timiz
ed
h
u
ma
n
d
etec
tio
n
in
N
LOS
s
ce
n
a
r
io
s
u
s
in
g
h
yb
r
id
d
imen
s
io
n
a
lity
…
(
E
n
o
ch
A
d
a
ma
Jiya
)
1295
p
r
o
v
id
i
n
g
g
r
ea
ter
in
s
ig
h
t
f
o
r
in
m
ates,
r
escu
e
ef
f
o
r
ts
,
an
d
im
p
r
o
v
e
d
d
etec
tio
n
s
[2
5
]
.
Peo
p
le
b
eh
in
d
b
a
r
r
ier
s
ca
n
n
o
w
b
e
class
if
ied
an
d
ev
al
u
ated
th
an
k
s
to
th
e
h
u
m
an
d
etec
tio
n
p
r
ed
ictio
n
o
f
NL
OS sig
n
al
d
ata
[2
6
]
.
Du
e
to
th
e
h
ig
h
-
d
im
en
s
io
n
al
d
ataset,
th
e
d
etec
tio
n
p
r
o
b
le
m
is
a
k
ey
o
n
e
th
at
lead
s
to
u
n
f
itti
n
g
co
n
clu
s
io
n
s
f
o
r
UW
B
d
ata
s
ig
n
als
o
b
tain
ed
f
r
o
m
s
en
s
o
r
s
.
An
NL
OS
s
ig
n
al
d
ataset
is
u
s
ed
in
th
is
in
v
esti
g
atio
n
.
T
h
e
Py
th
o
n
ap
p
licatio
n
is
u
s
ed
to
an
aly
s
e
t
h
e
d
ata
s
am
p
les.
T
h
e
AHPD
r
ec
eiv
es
th
e
s
am
p
les.
Af
ter
th
at,
a
s
m
aller
s
am
p
le
is
co
llected
an
d
f
e
d
in
to
t
h
e
GA
an
d
I
C
A
in
d
ep
e
n
d
en
tly
.
T
r
ain
i
n
g
an
d
test
in
g
s
ets
ar
e
cr
ea
ted
u
s
in
g
th
e
r
e
d
u
ce
d
d
ata.
SVM
is
u
s
ed
f
o
r
class
if
icatio
n
.
3
.
3
.
Di
m
ens
io
na
lity
red
uct
io
n
Dim
en
s
io
n
ality
r
ed
u
ctio
n
is
a
wid
ely
u
s
ed
m
eth
o
d
f
o
r
g
ettin
g
r
id
o
f
ex
tr
a
n
eo
u
s
f
e
atu
r
es
an
d
u
n
d
esira
b
le
n
o
is
e.
T
h
e
h
ig
h
-
d
im
en
s
io
n
al
f
ea
tu
r
es
in
th
e
NL
OS
d
ataset
ar
e
co
m
p
u
tatio
n
ally
in
ten
s
iv
e,
wh
ic
h
h
in
d
er
s
th
e
ef
f
ec
tiv
en
ess
o
f
class
if
ica
tio
n
m
eth
o
d
s
.
Dim
en
s
io
n
ality
r
ed
u
ctio
n
s
tr
ate
g
ies
ar
e
cr
u
cial
f
o
r
r
em
o
v
in
g
d
u
p
licatio
n
a
n
d
co
llectin
g
u
n
n
ec
ess
ar
y
f
ea
tu
r
es
th
at
r
ed
u
ce
ac
tiv
ity
ef
f
icien
cy
b
y
r
ed
u
ci
n
g
th
e
s
am
p
le
-
to
-
f
ea
tu
r
e
r
atio
s
.
T
h
is
ap
p
r
o
ac
h
r
e
d
u
ce
s
th
e
lik
elih
o
o
d
o
f
o
v
er
f
itti
n
g
.
On
e
im
p
o
r
tan
t
tech
n
iq
u
e
is
f
ea
tu
r
e
ex
tr
ac
tio
n
an
d
co
llectio
n
,
wh
ich
l
o
wer
s
th
e
d
im
e
n
s
io
n
ality
[
27
]
,
[
28
]
.
3
.
4
.
F
e
a
t
ure
s
elec
t
io
n
Mo
d
el
test
in
g
a
n
d
tr
ain
in
g
d
e
p
en
d
o
n
tech
n
o
lo
g
ies
s
u
ch
as NL
OS
s
ig
n
al
d
ata,
w
h
ich
p
r
o
d
u
ce
u
n
iq
u
e
an
d
p
er
tin
en
t
f
ea
t
u
r
e
I
Ds
f
o
r
t
r
an
s
cr
ip
t
s
eq
u
en
ce
s
.
T
o
im
p
r
o
v
e
class
if
icatio
n
p
er
f
o
r
m
an
ce
,
f
ea
tu
r
e
s
elec
tio
n
is
cr
u
cial.
Featu
r
e
s
elec
tio
n
less
en
s
th
e
d
etr
im
en
tal
im
p
ac
ts
o
f
d
im
en
s
io
n
ality
an
d
m
ak
es
it
p
o
s
s
ib
le
to
ch
o
o
s
e
p
er
tin
en
t
co
m
p
o
n
e
n
ts
f
o
r
cl
ass
if
icatio
n
m
o
d
el
p
er
f
o
r
m
a
n
ce
s
b
y
elim
in
atin
g
u
n
n
ec
es
s
ar
y
an
d
d
u
p
licate
ch
ar
ac
ter
is
tics
[2
9
]
,
[
3
0
]
.
I
t
s
u
p
p
o
r
ts
th
e
lear
n
in
g
p
r
o
ce
s
s
th
r
o
u
g
h
o
u
t
th
e
ca
teg
o
r
is
atio
n
p
h
ase
an
d
im
p
r
o
v
es
th
e
s
u
cc
ess
m
o
d
el.
F
o
r
e
x
a
m
p
l
e
,
b
o
t
h
s
u
p
e
r
v
i
s
e
d
a
n
d
u
n
s
u
p
e
r
v
i
s
e
d
d
e
c
i
s
i
o
n
-
m
a
k
i
n
g
l
e
a
r
n
i
n
g
a
r
e
u
s
e
d
i
n
t
h
e
m
a
s
s
i
v
e
i
n
f
o
r
m
a
t
i
o
n
f
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
m
e
t
h
o
d
t
h
a
t
c
o
m
b
i
n
e
s
w
a
l
l
a
n
d
S
A
R
d
a
t
a
.
T
h
e
p
r
e
d
i
c
t
i
o
n
m
o
d
e
l
’
s
e
f
f
i
c
a
c
y
w
i
l
l
b
e
i
n
c
r
e
a
s
e
d
b
y
u
s
i
n
g
c
a
r
e
f
u
l
l
y
c
h
o
s
e
n
o
p
t
i
m
a
l
r
a
n
k
a
t
t
r
i
b
u
t
e
s
t
h
a
t
c
o
m
m
u
n
i
c
a
t
e
p
r
i
o
r
i
t
y
f
o
r
c
a
t
e
g
o
r
i
s
a
t
i
o
n
j
o
b
s
.
O
n
e
e
f
f
e
c
t
i
v
e
m
e
t
h
o
d
k
n
o
w
n
a
s
a
f
i
l
t
e
r
,
w
r
a
p
p
e
r
,
o
r
e
m
b
e
d
d
e
d
t
y
p
e
i
s
t
h
e
c
o
l
l
e
c
t
i
o
n
o
f
f
e
a
t
u
r
e
s
e
l
e
c
t
i
o
n
[3
1
]
,
[
3
2
]
.
3
.
5
.
G
enet
ic
a
lg
o
rit
h
m
E
n
g
in
e
o
p
tim
is
atio
n
p
r
o
b
lem
s
ar
e
an
aly
s
ed
u
s
in
g
a
g
e
n
etic
alg
o
r
ith
m
,
wh
ich
is
an
e
v
o
lu
tio
n
ar
y
m
eth
o
d
f
o
r
ch
o
o
s
in
g
p
e
r
tin
en
t
f
ea
tu
r
es
b
ased
o
n
wr
a
p
p
er
s
.
T
h
e
p
er
s
is
ten
ce
o
f
th
e
r
ig
h
ti
s
t
p
ar
ad
ig
m
-
b
ased
g
en
etic
alg
o
r
ith
m
s
is
b
u
ilt
o
n
r
ea
l
b
eh
av
io
u
r
s
co
n
n
ec
ted
to
h
u
m
a
n
h
er
ed
itar
y
elem
en
ts
.
Gen
etic
alg
o
r
ith
m
s
in
clu
d
e
p
r
im
a
r
y
p
o
p
u
latio
n
ad
v
an
ce
s
,
f
itn
ess
ev
alu
atio
n
,
p
ar
en
t selec
tio
n
,
cr
o
s
s
o
v
er
,
a
n
d
m
u
tatio
n
[3
3
]
,
[
3
4
]
.
A
GA
i
s
an
ex
p
lo
r
ato
r
y
d
is
co
v
er
y
tech
n
iq
u
e
ch
ar
ac
ter
is
ed
b
y
a
s
tr
aig
h
tf
o
r
war
d
p
r
o
c
ed
u
r
e
th
at
g
en
er
ates
a
v
alu
e
ap
p
r
o
p
r
iate
f
o
r
th
e
p
r
im
ar
y
o
b
jectiv
e
o
f
c
o
m
p
u
tin
g
f
av
o
u
r
a
b
le
f
in
d
i
n
g
s
b
y
u
s
in
g
a
m
o
d
el
o
f
r
an
d
o
m
l
y
g
e
n
er
ated
r
esu
lts
.
I
n
g
e
n
er
al,
p
r
o
p
er
ty
s
ets
th
at
ar
e
r
ep
r
esen
ted
as
b
in
ar
y
s
tr
i
n
g
s
o
f
0
s
an
d
1
s
co
m
p
r
is
e
wr
ec
k
ag
e
o
r
r
u
b
b
le
[3
5
]
.
E
v
en
th
o
u
g
h
g
e
n
etic
alg
o
r
ith
m
s
ar
e
h
ig
h
ly
s
en
s
itiv
e
to
th
e
b
eg
in
n
in
g
p
o
p
u
latio
n
,
th
ey
ex
h
ib
it
an
o
p
tim
ality
d
ef
icit.
Alth
o
u
g
h
it
h
as
b
ee
n
r
ev
ea
led
to
y
ield
s
u
f
f
icien
t
em
in
en
ce
s
o
lu
tio
n
s
to
im
p
r
o
v
e
it
f
o
r
N
L
OS
s
am
p
lin
g
,
th
e
q
u
ality
o
f
its
o
u
tp
u
t
d
ec
lin
es
as
th
e
p
r
o
b
lem
d
im
en
s
io
n
s
in
cr
ea
s
e
[3
6
]
.
3
.
6
.
F
e
a
t
ure
ex
t
r
a
ct
io
n
Fin
d
in
g
s
ig
n
if
ican
t
tr
aits
,
attr
ib
u
tes,
o
r
s
tr
u
ctu
r
es
in
d
ata
i
s
k
n
o
wn
as
f
ea
tu
r
e
ex
tr
ac
tio
n
.
Fin
d
in
g
p
atter
n
s
an
d
p
u
b
lic
e
v
en
ts
in
an
ass
em
b
ly
o
f
i
d
en
tific
atio
n
s
ar
e
two
ex
am
p
les
o
f
f
ea
tu
r
e
ex
tr
ac
tio
n
s
tr
ateg
ies
[
37
]
.
Featu
r
e
ex
tr
ac
tio
n
is
u
s
ed
to
o
b
tain
an
ad
d
itio
n
al
d
et
ailed
p
ictu
r
e
o
f
th
e
f
ea
tu
r
es
wh
ile
wo
r
k
in
g
with
d
ata
th
at
co
n
tain
s
d
im
en
s
io
n
al
lo
ad
s
.
T
h
e
cu
r
s
e
o
f
d
im
en
s
io
n
ality
ca
n
b
e
less
en
ed
b
y
em
p
lo
y
i
n
g
f
ea
tu
r
e
ex
tr
ac
tio
n
to
is
o
late
r
ev
o
lu
tio
n
ar
y
f
ea
tu
r
e
v
ar
iab
les.
I
n
p
ar
ticu
lar
,
th
er
e
ar
e
two
m
ain
ca
teg
o
r
ies
o
f
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
i
q
u
es:
lin
ea
r
(
s
u
p
p
o
s
in
g
a
l
o
w
-
d
im
en
s
io
n
al
d
ep
ictio
n
r
esu
ltin
g
f
r
o
m
h
ig
h
-
d
im
en
s
io
n
al
f
ea
tu
r
es,
co
m
p
a
r
ab
ly
I
C
A)
an
d
n
o
n
-
lin
ea
r
(
ass
u
m
in
g
d
ata
o
n
a
lo
w
-
d
im
en
s
io
n
al
s
u
b
s
p
a
ce
,
lik
e
PC
A)
f
o
r
a
non
-
lin
ea
r
r
elatio
n
s
h
ip
b
etwe
en
f
ea
tu
r
es
[
1
9
]
,
[
38
]
.
3
.
7
.
I
nd
ependent
co
m
po
nen
t
a
na
ly
s
is
B
y
s
ep
ar
atin
g
m
u
ltiv
ar
iate
s
ig
n
als
in
to
d
is
tin
ct
n
o
n
-
Gau
s
s
ian
co
m
p
o
n
en
ts
f
o
r
s
tatis
t
ically
in
d
ep
en
d
en
t
co
m
p
o
n
e
n
ts
,
I
C
A
ca
n
ass
is
t
in
r
ev
ea
lin
g
h
i
d
d
en
f
ea
t
u
r
es
f
r
o
m
m
u
ltid
im
en
s
io
n
al
d
ata.
I
C
A
em
b
ellis
h
es
th
e
d
ata
b
y
d
eleti
n
g
o
r
alter
in
g
th
e
r
elev
an
t
in
f
o
r
m
atio
n
t
o
f
in
d
a
r
elatio
n
s
h
ip
am
o
n
g
s
t
th
e
b
its
o
f
in
f
o
r
m
atio
n
[3
9
]
.
A:
I
C
A
ad
o
p
ts
o
p
in
io
n
B
as
a
s
tr
aig
h
t
-
lin
e
co
m
b
in
atio
n
o
f
th
e
in
d
iv
id
u
al
p
ar
ts
.
I
f
B
r
elate
s
to
th
e
c
o
lu
m
n
s
o
f
C
,
th
en
d
ef
i
n
e
th
e
f
u
n
d
am
en
tal
c
h
ar
ac
ter
i
s
tic,
th
e
in
d
ep
e
n
d
en
t
weig
h
te
d
m
atr
ix
R
,
v
ec
to
r
s
o
f
o
b
s
er
v
atio
n
X.
=
,
=
(
1
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
TEL
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
23
,
No
.
5
,
Octo
b
e
r
20
25
:
1
2
9
1
-
1
3
0
3
1296
I
C
A
h
as
b
ee
n
wid
ely
a
p
p
lied
in
in
f
o
r
m
atio
n
r
et
r
iev
al,
r
ec
o
g
n
itio
n
,
th
r
o
u
g
h
-
wall
ap
p
licatio
n
s
,
an
d
SAR
[4
0
]
,
[
4
1
]
.
GA
is
a
n
o
n
-
lin
ea
r
o
p
tim
izatio
n
tech
n
iq
u
e
th
at
r
ed
u
ce
s
th
e
n
u
m
b
er
a
n
d
d
im
en
s
io
n
ality
o
f
f
ea
tu
r
es.
Alth
o
u
g
h
GA
is
in
h
er
en
tly
n
o
n
-
lin
ea
r
,
p
r
e
p
r
o
ce
s
s
in
g
en
h
an
ce
s
p
er
f
o
r
m
an
ce
an
d
allo
ws
I
C
A
to
o
p
er
ate
as a
lin
ea
r
tec
h
n
iq
u
e
[4
2
]
.
4.
CL
AS
SI
F
I
CAT
I
O
N
On
e
p
o
p
u
lar
s
u
p
er
v
is
ed
lear
n
i
n
g
tactic
i
n
d
ata
m
in
in
g
m
eth
o
d
s
is
class
if
icatio
n
.
I
t
en
tails
class
lab
el
ass
ig
n
m
en
t
an
d
p
r
ed
ictio
n
u
s
in
g
p
r
e
d
eter
m
in
e
d
class
lab
els
an
d
av
ailab
le
d
ata.
T
h
e
r
e
a
r
e
two
s
tep
s
in
th
e
ca
teg
o
r
is
atio
n
p
r
o
ce
d
u
r
e
[4
3
]
.
First,
a
class
lab
el
an
d
a
co
llectio
n
o
f
tr
ain
i
n
g
d
ata
a
r
e
u
s
ed
to
d
ev
el
o
p
a
class
if
icatio
n
m
o
d
el.
T
h
e
ac
cu
r
ac
y
o
f
th
e
SVM
class
if
ier
is
th
en
ass
ess
ed
b
y
u
s
in
g
t
h
is
m
o
d
el
to
p
r
ed
ict
class
lab
els f
o
r
d
ata
th
at
h
as n
o
t y
et
b
ee
n
o
b
s
er
v
ed
.
T
h
e
tex
t p
r
o
v
i
d
es d
ef
in
itio
n
s
f
o
r
th
e
u
tili
s
ed
eq
u
atio
n
s
.
4
.
1
.
Su
pp
o
rt
v
ec
t
o
r
ma
chine
B
y
id
en
tify
in
g
th
e
b
est
h
y
p
er
p
lan
e
in
th
e
i
n
p
u
t
s
p
ac
e
,
SVM
aim
s
to
s
ep
ar
ate
g
r
o
u
p
s
.
B
y
in
co
r
p
o
r
atin
g
th
e
k
er
n
el
n
o
tio
n
s
in
to
h
ig
h
-
d
im
en
s
io
n
al
wo
r
k
s
p
ac
es,
SVM,
a
lin
ea
r
class
if
ier
,
is
d
ev
elo
p
e
d
to
h
an
d
le
n
o
n
-
lin
ea
r
s
ce
n
a
r
io
s
.
Fo
r
n
o
n
-
lin
ea
r
s
ce
n
ar
io
s
,
SVM
u
s
es
a
k
er
n
el
to
tr
ain
th
e
d
ata
in
o
r
d
er
to
n
ar
r
o
w
th
e
s
p
r
ea
d
th
e
d
im
en
s
io
n
.
W
h
en
m
o
d
if
y
in
g
th
e
p
r
o
p
o
r
tio
n
s
,
SVM
s
h
o
u
ld
s
ea
r
ch
f
o
r
th
e
b
est
h
y
p
er
p
lan
e
th
a
t
ca
n
d
is
tin
g
u
is
h
o
n
e
class
f
r
o
m
an
o
th
er
[4
4
]
.
B
y
id
en
tif
y
in
g
t
h
e
b
est
h
y
p
er
p
lan
e
in
th
e
in
p
u
t
s
p
ac
e,
SVM
aim
s
to
s
ep
ar
ate
g
r
o
u
p
s
.
SVM,
a
lin
ea
r
class
if
ier
,
is
d
ev
elo
p
ed
b
y
co
m
b
in
in
g
th
e
k
er
n
el
co
n
ce
p
ts
in
h
ig
h
-
d
im
en
s
io
n
al
wo
r
k
s
p
ac
es
to
h
an
d
le
n
o
n
-
lin
ea
r
s
ce
n
ar
io
s
.
SVM
em
p
lo
y
s
a
k
er
n
el
to
tr
ain
th
e
d
ata
to
n
ar
r
o
w
d
is
tr
ib
u
te
th
e
d
im
en
s
io
n
f
o
r
n
o
n
-
lin
ea
r
s
itu
atio
n
s
.
SVM
ca
n
f
in
d
th
e
o
p
tim
al
h
y
p
er
p
lan
e
an
d
d
if
f
er
e
n
tiate
a
class
f
r
o
m
o
th
er
class
es b
y
ad
j
u
s
tin
g
th
e
p
r
o
p
o
r
tio
n
s
[4
5
]
.
T
h
e
Gau
s
s
ian
k
er
n
el
[
46
]
is
ass
o
ciate
d
with
th
e
g
en
er
al
as
s
u
m
p
tio
n
th
at
all
k
th
-
o
r
d
er
s
u
b
o
r
d
in
ates
ar
e
s
m
o
o
th
.
T
o
d
escr
ib
e
p
r
ev
io
u
s
lear
n
in
g
c
h
allen
g
es,
k
e
r
n
els
th
at
co
n
tr
o
l
a
ce
r
tain
p
r
io
r
d
ata
r
ec
u
r
r
e
n
ce
m
ater
ial
ca
n
b
e
co
n
s
tr
u
cted
.
All
o
f
th
e
p
o
ly
n
o
m
ial
ex
ten
s
io
n
s
o
f
th
e
x
co
m
p
o
n
e
n
ts
ar
e
in
clu
d
ed
in
th
e
tr
an
s
latio
n
o
f
ea
ch
in
p
u
t
v
ec
to
r
,
x
,
in
t
o
an
in
f
in
ite
-
d
im
en
s
io
n
al
v
ec
to
r
[
47
]
.
Ad
d
in
g
d
im
en
s
io
n
s
to
NL
O
S
s
ig
n
al
d
ata
is
a
m
ajo
r
ch
allen
g
e
to
s
tr
aig
h
tf
o
r
war
d
,
t
r
u
s
two
r
th
y
r
esear
ch
tech
n
iq
u
es.
W
h
en
lear
n
in
g
co
m
p
le
x
s
tr
ateg
ie
s
o
n
m
u
ltip
le
le
v
els
th
at
ar
e
in
f
lu
e
n
ce
d
b
y
m
o
r
p
h
o
lo
g
ical
p
r
o
ce
s
s
es
th
at
ar
e
o
f
in
ter
est,
it
is
im
p
er
ati
v
e
to
em
p
l
o
y
tr
a
d
itio
n
al
way
s
.
Mo
s
t
tr
ad
itio
n
al
ap
p
r
o
ac
h
es
f
o
r
h
a
n
d
lin
g
h
ig
h
-
d
im
en
s
io
n
al
d
ata,
lik
e
th
e
N
L
OS
s
ig
n
al
d
ata,
h
av
e
s
ev
er
al
p
r
o
b
lem
s
.
W
h
en
a
p
o
r
tio
n
o
f
d
ata
f
r
o
m
o
n
e
o
p
er
a
tio
n
is
ad
d
ed
to
th
e
in
p
u
t o
f
an
o
th
er
,
th
e
ap
p
licatio
n
o
f
d
if
f
er
en
t d
im
en
s
io
n
ality
r
ed
u
ctio
n
tec
h
n
iq
u
es
ca
n
p
r
o
v
id
e
s
p
ec
ial
ad
v
an
tag
es.
Feat
u
r
e
ex
tr
ac
tio
n
tech
n
iq
u
es
o
f
t
en
em
p
lo
y
f
ea
tu
r
e
s
elec
tio
n
o
r
r
ed
u
n
d
an
t
s
ig
n
al
d
ata
d
eletio
n
to
ch
o
o
s
e
th
e
o
r
ig
in
al
s
u
b
s
et
o
f
d
ata,
r
esp
e
ctiv
ely
,
s
o
f
ac
ilit
at
e
f
ea
tu
r
e
s
elec
tio
n
.
I
t
m
ay
b
e
ad
v
an
tag
e
o
u
s
to
ex
tr
ac
t
p
r
im
ar
y
s
u
b
s
et
f
ea
tu
r
es
an
d
co
m
b
in
e
m
an
y
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
i
q
u
es
[
38
]
,
[
48
]
.
T
h
is
wo
r
k
p
r
o
p
o
s
ed
an
ef
f
icie
n
t
d
im
en
s
io
n
r
e
d
u
ctio
n
tech
n
i
q
u
e
f
o
r
NL
OS
s
ig
n
al
d
ata
class
if
icatio
n
.
T
h
is
m
eth
o
d
h
as
en
o
r
m
o
u
s
p
r
o
m
is
e
f
o
r
tr
ac
k
in
g
d
o
wn
,
id
e
n
tify
in
g
,
an
d
lo
ca
tin
g
v
ictim
s
wh
o
ar
e
co
n
ce
aled
b
en
ea
th
t
h
e
g
r
o
u
n
d
.
Ho
wev
er
,
th
e
s
tr
u
ct
u
r
es
b
ec
o
m
e
m
o
r
e
a
p
p
ar
en
t
wh
en
th
e
d
im
en
s
io
n
al
ity
is
r
ed
u
ce
d
.
Data
is
s
t
ill
d
if
f
icu
lt
to
h
an
d
le,
t
h
o
u
g
h
,
an
d
ex
is
tin
g
alg
o
r
ith
m
s
r
eq
u
ir
e
im
p
r
o
v
e
m
en
t
to
ex
h
ib
it
th
e
r
ig
h
t
ch
ar
ac
ter
is
tics
.
Alth
o
u
g
h
th
e
f
u
s
io
n
s
tr
ateg
y
o
f
f
e
r
s
b
en
e
f
its
,
it
also
n
ec
es
s
itates
th
e
u
s
e
o
f
b
en
ef
icial
m
o
d
ellin
g
tech
n
iq
u
es.
T
h
is
wo
r
k
p
r
o
p
o
s
ed
an
ef
f
icie
n
t
d
im
en
s
io
n
r
e
d
u
ctio
n
tech
n
i
q
u
e
f
o
r
NL
OS
s
ig
n
al
d
ata
class
if
icatio
n
.
T
h
is
m
eth
o
d
h
as
en
o
r
m
o
u
s
p
r
o
m
is
e
f
o
r
tr
ac
k
in
g
d
o
wn
,
id
e
n
tify
in
g
,
an
d
lo
ca
tin
g
v
ictim
s
wh
o
ar
e
co
n
ce
aled
b
en
ea
th
t
h
e
g
r
o
u
n
d
.
Ho
wev
er
,
th
e
s
tr
u
ct
u
r
es
b
ec
o
m
e
m
o
r
e
a
p
p
ar
en
t
wh
en
th
e
d
im
en
s
io
n
al
ity
is
r
ed
u
ce
d
.
Data
is
s
t
ill
d
if
f
icu
lt
to
h
an
d
le,
t
h
o
u
g
h
,
an
d
ex
is
tin
g
alg
o
r
ith
m
s
r
eq
u
ir
e
im
p
r
o
v
e
m
en
t
to
ex
h
ib
it
th
e
r
ig
h
t
ch
ar
ac
ter
is
tics
.
Alth
o
u
g
h
th
e
f
u
s
io
n
s
tr
ateg
y
o
f
f
e
r
s
b
en
e
f
its
,
it
also
n
ec
es
s
itates
th
e
u
s
e
o
f
b
en
ef
icial
m
o
d
ellin
g
tech
n
iq
u
es.
T
h
e
AHPD
cy
p
h
er
tex
t,
f
ea
tu
r
es
ch
o
s
en
,
ch
ar
ac
ter
is
tics
elim
in
ated
,
an
d
th
e
class
o
f
th
e
class
ar
e
th
e
f
o
u
r
s
tep
s
th
at
h
a
v
e
b
ee
n
s
u
g
g
ested
p
r
io
r
to
th
e
class
if
icati
o
n
tech
n
iq
u
e.
T
h
e
p
r
o
p
o
s
ed
h
y
b
r
id
s
y
s
tem
f
o
r
th
e
NL
OS
d
ata
h
u
m
an
d
etec
tio
n
ar
ch
itectu
r
e,
wh
ic
h
p
r
e
d
icts
v
ictim
s
tr
ap
p
ed
b
eh
in
d
d
eb
r
is
u
s
in
g
th
e
NL
OS
s
ig
n
al
d
ataset,
is
s
h
o
wn
in
Fig
u
r
e
1
.
Fo
u
r
s
u
b
s
y
s
tem
s
m
ak
e
u
p
th
e
f
r
am
ewo
r
k
:
o
n
e
f
o
r
class
-
b
ased
f
ea
tu
r
e
ex
tr
ac
tio
n
,
o
n
e
f
o
r
AHDP
p
s
eu
d
o
co
d
e,
o
n
e
f
o
r
f
ea
tu
r
e
co
l
lectin
g
,
an
d
o
n
e
f
o
r
class
if
icatio
n
.
T
h
e
f
u
n
ctio
n
s
elec
tio
n
s
u
b
-
s
y
s
tem
em
p
lo
y
s
AHPD
p
s
eu
d
o
co
d
e,
wh
ich
f
ilter
s
n
o
is
e
u
s
in
g
th
e
f
ir
s
t
m
eth
o
d
an
d
u
tili
s
es
GA
to
ev
alu
ate
th
e
f
itn
ess
to
id
en
t
if
y
an
id
ea
l
s
u
b
s
et.
I
C
A
is
u
s
ed
b
y
th
e
f
u
n
cti
o
n
e
x
tr
ac
tio
n
s
u
b
s
y
s
tem
d
u
e
to
its
d
ata
p
r
o
jectio
n
o
f
ef
f
icien
cy
,
in
v
ar
ian
ce
,
an
d
im
p
er
tin
en
t
o
r
d
er
in
g
.
SVM
is
u
s
ed
to
class
if
y
r
esear
ch
s
tan
d
ar
d
s
.
T
h
e
d
y
n
am
ic
p
r
o
p
er
ties
o
f
t
h
e
h
u
m
a
n
d
etec
tin
g
alg
o
r
ith
m
,
wh
ich
o
f
f
er
m
an
y
s
ea
r
c
h
ar
ea
s
th
at
in
d
ep
en
d
en
tly
an
d
co
n
cu
r
r
en
tl
y
r
ev
iew
th
e
b
est
r
esu
lt
to
p
r
o
d
u
ce
a
g
o
o
d
r
esu
lt,
ar
e
wh
at
m
ak
e
it
im
p
o
r
tan
t
to
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
Op
timiz
ed
h
u
ma
n
d
etec
tio
n
in
N
LOS
s
ce
n
a
r
io
s
u
s
in
g
h
yb
r
id
d
imen
s
io
n
a
lity
…
(
E
n
o
ch
A
d
a
ma
Jiya
)
1297
o
p
tim
is
e.
T
o
r
e
d
u
ce
th
e
n
u
m
b
er
o
f
f
ea
tu
r
es
in
th
is
s
tu
d
y
wh
ile
p
r
eser
v
i
n
g
d
is
cr
im
in
atin
g
q
u
alities
,
GA
f
ea
tu
r
es
wer
e
u
s
ed
.
R
ed
u
ce
d
d
ata
ar
e
co
n
v
er
ted
in
to
late
n
t
co
m
p
o
n
en
ts
u
s
in
g
th
e
b
est
f
ea
tu
r
e
ex
tr
ac
tio
n
tech
n
iq
u
e.
Sad
ly
,
th
is
r
e
d
u
ce
s
its
p
r
o
d
u
ctiv
ity
an
d
in
v
alid
at
es
b
o
th
d
im
en
s
io
n
ality
r
e
d
u
ct
io
n
tech
n
iq
u
es
f
o
r
th
e
d
ataset.
5.
RE
SU
L
T
S AN
D
D
I
SCU
SS
I
O
N
Usi
n
g
a
p
u
b
licly
av
ailab
le
d
ataset
co
n
tain
in
g
2
3
,
5
5
2
s
am
p
les
an
d
2
5
6
o
cc
u
r
r
en
ce
s
,
as
s
h
o
wn
in
T
ab
le
2
,
th
is
r
esear
ch
p
r
esen
ts
a
Py
th
o
n
-
b
ased
to
o
l f
o
r
NL
O
S
s
ig
n
al
d
ataset
class
if
icat
io
n
[
2
3
]
.
A
t
o
tal
o
f
4
8
1
f
ea
tu
r
es
wer
e
ex
tr
ac
ted
f
r
o
m
t
h
e
d
y
n
am
ic
d
ataset
as
r
elev
an
t
f
ea
tu
r
es.
T
h
e
AHPD
ap
p
r
o
ac
h
was
em
p
lo
y
e
d
to
f
ilter
o
u
t n
o
is
e
an
d
r
ef
in
e
f
ea
tu
r
e
s
elec
tio
n
,
th
er
eb
y
en
h
an
ci
n
g
class
if
icatio
n
ac
cu
r
ac
y
an
d
ef
f
icien
cy
.
Fig
u
r
e
2
illu
s
tr
ates th
e
p
r
o
p
o
s
ed
p
r
ed
ic
tio
n
f
r
am
ewo
r
k
f
o
r
h
u
m
an
d
et
ec
tio
n
d
ata
an
aly
s
is
.
T
ab
le
2
.
C
o
m
p
a
r
ativ
e
ap
p
r
o
ac
h
es
M
e
t
h
o
d
s e
mp
l
o
y
e
d
A
c
c
u
r
a
c
y
(
%)
C
N
N
+
st
a
c
k
e
d
-
LST
M
[
49
]
8
2
.
1
4
A
H
P
D
+
G
A
+
B
a
g
g
e
d
e
n
s
e
mb
l
e
[5
0
]
8
5
.
6
9
F
E+
S
V
M
[5
1
]
8
3
.
0
0
S
V
M
+
E
n
sem
b
l
e
[5
1
]
8
1
.
0
0
A
H
P
D
+
G
A
+
S
V
M
(
p
r
o
p
o
s
e
d
mo
d
e
l
)
8
5
.
7
8
Fig
u
r
e
2
.
Pro
p
o
s
ed
c
o
m
p
lete
f
r
am
ewo
r
k
f
lo
w
f
o
r
h
u
m
an
d
et
ec
tio
n
d
ata
an
aly
s
is
A
0
.
5
th
r
esh
o
ld
was
ap
p
lied
as
a
d
ec
is
io
n
b
o
u
n
d
ar
y
in
cla
s
s
if
icatio
n
task
s
to
d
eter
m
in
e
wh
eth
er
a
d
etec
ted
s
ig
n
al
co
r
r
esp
o
n
d
s
to
h
u
m
an
b
r
ea
t
h
in
g
.
T
h
is
th
r
esh
o
ld
in
g
m
ec
h
an
is
m
en
s
u
r
es
th
a
t
o
n
ly
s
ig
n
als
with
s
u
f
f
icien
t
co
n
f
id
e
n
ce
ar
e
class
if
ied
as
b
r
ea
th
in
g
,
th
er
eb
y
i
m
p
r
o
v
i
n
g
d
etec
tio
n
r
eliab
ilit
y
.
B
y
im
p
lem
en
tin
g
th
is
th
r
esh
o
ld
,
th
e
s
y
s
tem
ef
f
e
ctiv
ely
m
in
im
izes
f
alse
p
o
s
iti
v
es,
wh
ich
co
u
ld
m
is
id
en
tif
y
n
o
n
-
h
u
m
an
s
ig
n
als
s
u
ch
as
m
ac
h
in
er
y
v
i
b
r
atio
n
s
o
r
en
v
ir
o
n
m
en
tal
n
o
is
e
as
b
r
ea
th
in
g
.
Ad
d
itio
n
ally
,
f
alse
n
eg
ativ
es,
wh
ich
o
cc
u
r
wh
en
r
ea
l
h
u
m
an
b
r
ea
th
in
g
i
s
n
o
t
d
etec
ted
,
ar
e
r
ed
u
ce
d
b
y
o
p
tim
izin
g
t
h
e
s
y
s
tem
’
s
s
en
s
itiv
ity
with
in
th
e
s
p
ec
if
ied
f
r
eq
u
en
cy
r
an
g
e.
Fu
r
th
er
m
o
r
e
,
th
e
b
r
ea
th
f
r
eq
u
en
cy
r
a
n
g
e
was
s
et
at
[
0
.
2
,
0
.
7
]
Hz
to
en
h
an
ce
SAR
ef
f
ec
tiv
en
ess
,
p
ar
ticu
lar
ly
in
d
is
aster
s
ce
n
ar
io
s
wh
er
e
v
ictim
s
m
ay
h
av
e
wea
k
en
ed
o
r
ir
r
eg
u
lar
b
r
ea
th
in
g
p
atter
n
s
.
I
n
d
iv
id
u
als
tr
ap
p
ed
u
n
d
er
r
u
b
b
le
d
u
e
to
s
tr
u
ctu
r
al
co
llap
s
es
m
ay
ex
p
e
r
ien
ce
s
lo
w
o
r
r
ap
id
b
r
ea
t
h
in
g
as
a
r
esu
lt
o
f
tr
au
m
a
,
p
a
n
ic,
o
r
o
x
y
g
en
d
ep
r
iv
atio
n
.
B
y
ex
p
an
d
in
g
th
e
b
r
ea
t
h
f
r
eq
u
e
n
cy
r
an
g
e,
t
h
e
s
y
s
tem
b
ec
o
m
es
m
o
r
e
ad
ap
tiv
e
to
d
iv
er
s
e
p
h
y
s
io
lo
g
ical
co
n
d
itio
n
s
,
th
er
eb
y
i
m
p
r
o
v
i
n
g
its
r
o
b
u
s
tn
ess
in
r
ea
l
-
wo
r
ld
s
ea
r
ch
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
TEL
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
23
,
No
.
5
,
Octo
b
e
r
20
25
:
1
2
9
1
-
1
3
0
3
1298
r
escu
e
ap
p
licatio
n
s
.
T
h
e
s
elec
ted
r
an
g
e
e
n
s
u
r
es
th
at
ev
en
s
u
b
tle
r
esp
ir
ato
r
y
s
ig
n
als
ar
e
d
etec
ted
,
in
cr
ea
s
in
g
th
e
lik
elih
o
o
d
o
f
s
u
cc
ess
f
u
l h
u
m
an
p
r
esen
ce
i
d
en
tific
atio
n
in
co
m
p
lex
a
n
d
n
o
is
y
en
v
i
r
o
n
m
en
ts
.
T
h
e
d
ataset
was
ev
alu
ated
u
s
in
g
s
tate
-
of
-
th
e
-
ar
t
m
ac
h
in
e
lear
n
in
g
class
if
ier
s
,
with
an
em
p
h
asis
o
n
o
p
tim
izin
g
h
y
p
er
p
ar
am
eter
s
f
o
r
im
p
r
o
v
ed
p
er
f
o
r
m
an
ce
.
GA
-
b
ased
f
ea
tu
r
e
s
elec
tio
n
was
ap
p
lied
to
r
e
f
in
e
th
e
d
ataset
b
y
s
elec
tin
g
o
n
ly
th
e
m
o
s
t
r
elev
an
t
f
ea
tu
r
es,
th
u
s
r
ed
u
cin
g
d
i
m
en
s
io
n
al
ity
an
d
en
h
an
ci
n
g
class
if
icatio
n
ac
cu
r
ac
y
.
T
h
r
o
u
g
h
th
is
o
p
tim
izatio
n
p
r
o
ce
s
s
,
t
h
e
id
ea
l
n
u
m
b
er
o
f
n
eig
h
b
o
r
s
(
n
_
n
eig
h
b
o
r
s
)
was
id
en
tifie
d
as
1
1
f
o
r
th
e
SVM
class
if
ier
.
T
h
is
v
alu
e
p
r
o
v
id
ed
th
e
b
est
tr
ad
e
-
o
f
f
b
etwe
en
b
ias
an
d
v
ar
ian
ce
,
en
s
u
r
in
g
a
b
alan
ce
b
etwe
en
m
o
d
el
co
m
p
le
x
ity
an
d
g
en
er
aliza
tio
n
.
T
h
e
u
s
e
o
f
GA
f
u
r
th
er
e
n
h
an
ce
d
class
if
icatio
n
p
er
f
o
r
m
an
ce
b
y
o
p
tim
izin
g
f
ea
tu
r
e
s
elec
tio
n
a
n
d
e
n
s
u
r
in
g
r
o
b
u
s
t
g
e
n
er
aliza
t
io
n
ac
r
o
s
s
d
y
n
a
m
ic
d
atasets
.
T
h
is
s
tu
d
y
u
n
d
er
s
co
r
es
th
e
ef
f
e
ctiv
en
ess
o
f
in
teg
r
atin
g
d
im
e
n
s
io
n
ality
r
e
d
u
cti
o
n
tech
n
iq
u
es
with
m
ac
h
in
e
lear
n
in
g
class
if
ier
s
to
im
p
r
o
v
e
NL
OS
h
u
m
an
d
et
ec
tio
n
in
s
ea
r
ch
an
d
r
escu
e
o
p
er
atio
n
s
.
Fig
u
r
e
3
s
h
o
ws th
e
co
n
f
u
s
io
n
m
atr
ices f
o
r
AHPD+
GA+
SVM
.
Fig
u
r
e
3
.
C
o
n
f
u
s
io
n
m
atr
i
x
f
o
r
AHPD
with
GA
an
d
SVM
(
d
y
n
am
ic)
T
P=1
3
2
9
; T
N=
1
6
5
7
;
FP
=1
5
4
; FN=
3
4
1
T
h
e
SVM
class
if
icatio
n
tech
n
iq
u
e
with
v
alid
atio
n
was
u
s
ed
to
p
r
ed
ict
th
e
o
u
tco
m
es
o
f
th
e
ex
tr
ac
ted
f
ea
tu
r
es.
T
o
ev
alu
ate
p
r
e
d
ictio
n
m
o
d
els,
th
is
ap
p
r
o
ac
h
s
ep
ar
ates
th
e
g
iv
en
s
am
p
le
in
to
tr
ain
in
g
an
d
test
in
g
s
ets.
Per
f
o
r
m
an
ce
in
d
icato
r
s
wer
e
th
en
u
s
ed
to
ass
ess
th
e
SVM
co
n
f
u
s
io
n
m
atr
ix
.
A
d
d
it
io
n
ally
,
th
e
f
ilter
ed
AHPD
f
ea
tu
r
es
wer
e
s
u
b
jecte
d
to
th
e
GA
m
eth
o
d
a
n
d
t
h
e
I
C
A
v
ar
iab
les.
T
h
e
c
o
n
f
u
s
io
n
m
atr
ix
was
ev
alu
ated
f
o
llo
win
g
th
e
class
if
icatio
n
o
f
th
e
laten
t
f
ea
tu
r
es
u
s
in
g
SVM
with
cr
o
s
s
-
v
alid
atio
n
.
T
h
e
alg
o
r
ith
m
s
AHPD
+G
A+
SVM
an
d
AHPD+
I
C
A
+SVM
wer
e
test
ed
o
n
th
e
NL
OS
s
ig
n
al
d
ataset.
Fig
u
r
e
4
d
is
p
lay
th
e
R
OC
f
o
r
I
C
A
-
SVM
p
er
f
o
r
m
an
ce
Fig
u
r
e
4
.
A
R
OC
cu
r
v
e
o
f
th
e
SVM
attr
ib
u
tes with
I
C
A
Evaluation Warning : The document was created with Spire.PDF for Python.
T
E
L
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l Co
n
tr
o
l
Op
timiz
ed
h
u
ma
n
d
etec
tio
n
in
N
LOS
s
ce
n
a
r
io
s
u
s
in
g
h
yb
r
id
d
imen
s
io
n
a
lity
…
(
E
n
o
ch
A
d
a
ma
Jiya
)
1299
Nu
m
er
o
u
s
r
esear
c
h
er
s
h
av
e
em
p
lo
y
e
d
m
ac
h
i
n
e
lear
n
i
n
g
ap
p
r
o
ac
h
es
to
ad
d
r
ess
f
u
n
d
am
en
tal
class
if
icatio
n
p
r
o
b
lem
s
an
d
d
ev
elo
p
r
eliab
le
d
etec
tio
n
a
n
d
p
r
ed
ictio
n
s
tr
ateg
ies
f
o
r
id
en
tify
in
g
tr
ap
p
ed
v
ictim
s
.
Fig
u
r
e
5
d
ep
icts
th
e
co
n
f
u
s
io
n
m
atr
ices
f
o
r
AHPD+
I
C
A+
SVM.
T
h
e
r
esu
lts
p
r
esen
ted
in
Fig
u
r
e
6
d
em
o
n
s
tr
ate
a
n
im
p
r
o
v
em
en
t
o
v
e
r
th
e
p
r
ev
io
u
s
m
eth
o
d
.
C
o
m
p
ar
ed
t
o
th
e
s
tate
-
of
-
t
h
e
-
ar
t,
th
e
ac
cu
r
ac
y
im
p
r
o
v
e
d
(
T
ab
le
2
)
.
W
h
en
a
GA
f
o
r
f
ea
tu
r
e
s
elec
tio
n
was
u
s
ed
in
co
n
j
u
n
ctio
n
with
an
SVM
class
if
ier
,
th
e
AHPD
s
y
s
tem
d
em
o
n
s
tr
ated
s
ig
n
if
ican
t p
er
f
o
r
m
a
n
ce
im
p
r
o
v
em
en
ts
,
p
ar
ticu
lar
ly
f
o
r
th
e
d
y
n
am
ic
d
ataset.
Fig
u
r
e
5
.
C
o
n
f
u
s
io
n
m
atr
i
x
f
o
r
AHPD
with
I
C
A
an
d
SVM
(
d
y
n
am
ic
d
ata)
T
P=1
2
0
2
; T
N=
1
5
5
5
; FP=2
8
1
;
FN=4
4
3
Fig
u
r
e
6
.
Per
f
o
r
m
an
c
e
m
etr
ics cla
s
s
if
icatio
n
r
esu
lt f
o
r
th
e
ex
p
er
im
en
t
T
h
e
AHPD+
GA+
S
VM
m
o
d
el
ac
h
iev
ed
th
e
h
ig
h
est
p
er
f
o
r
m
an
ce
,
with
an
ac
cu
r
ac
y
o
f
8
5
.
7
8
%.
T
h
is
im
p
lies
th
at
it
h
as
a
s
tr
o
n
g
ab
ilit
y
to
co
r
r
ec
tly
p
r
e
d
ict
b
o
t
h
th
e
p
r
esen
ce
an
d
a
b
s
en
ce
o
f
h
u
m
a
n
s
tr
ap
p
ed
b
eh
in
d
r
u
b
b
le.
T
h
e
h
i
g
h
ac
c
u
r
ac
y
s
u
g
g
ests
th
at
GA
ef
f
e
ctiv
ely
s
elec
ts
th
e
m
o
s
t
r
ele
v
an
t
f
ea
tu
r
es
wh
ile
r
em
o
v
in
g
r
ed
u
n
d
a
n
t
o
r
n
o
is
y
d
ata
ca
u
s
ed
b
y
NL
OS
s
ig
n
a
l
r
ef
lectio
n
s
,
lead
in
g
to
im
p
r
o
v
ed
class
if
icatio
n
ef
f
icien
cy
.
T
h
e
ab
ilit
y
to
co
r
r
ec
tly
d
etec
t
th
e
p
r
esen
ce
o
f
tr
ap
p
ed
v
ictim
s
(
s
en
s
itiv
ity
)
is
8
0
.
0
0
%.
T
h
is
m
ea
n
s
th
e
s
y
s
tem
is
ef
f
ec
tiv
e
in
m
in
i
m
izin
g
f
alse
n
eg
ativ
es,
w
h
ich
is
cr
u
cial
in
SAR
m
is
s
io
n
s
wh
er
e
th
e
in
ab
ilit
y
t
o
d
etec
t a
tr
ap
p
ed
in
d
iv
id
u
al
co
u
ld
h
av
e
life
-
th
r
ea
te
n
in
g
co
n
s
eq
u
en
ce
s
.
T
h
e
m
o
d
e
l
i
s
r
o
b
u
s
t
a
g
a
i
n
s
t
f
a
ls
e
a
la
r
m
s
,
a
s
it
d
e
m
o
n
s
t
r
a
t
es
a
h
i
g
h
c
a
p
ab
i
l
i
t
y
(
9
6
.
4
6
%
)
i
n
c
o
r
r
e
c
tl
y
i
d
en
t
i
f
y
i
n
g
n
o
n
-
h
u
m
a
n
p
r
e
s
e
n
c
e
(
s
p
e
c
i
f
i
ci
t
y
)
,
m
a
k
i
n
g
i
t
h
i
g
h
l
y
r
e
l
i
a
b
l
e
f
o
r
r
e
a
l
-
w
o
r
ld
d
e
p
l
o
y
m
e
n
t
i
n
d
i
s
a
s
t
e
r
s
c
e
n
ar
i
o
s
.
T
h
i
s
r
e
d
u
c
e
s
u
n
n
e
c
e
s
s
a
r
y
r
e
s
o
u
r
c
e
a
l
l
o
c
a
t
i
o
n
,
e
n
s
u
r
i
n
g
t
h
a
t
r
e
s
c
u
e
t
e
a
m
s
f
o
c
u
s
o
n
a
c
t
u
a
l
h
u
m
a
n
p
r
e
s
e
n
c
e
r
a
t
h
e
r
t
h
a
n
f
a
l
s
e
d
et
e
c
ti
o
n
s
.
F
i
g
u
r
e
6
co
n
tain
s
th
e
lis
t o
f
p
er
f
o
r
m
an
ce
ev
al
u
atio
n
r
esu
lts
f
r
o
m
th
e
ex
p
er
im
en
ts
.
0
20
40
60
80
1
0
0
1
2
0
Ac
c
u
ra
c
y
(%
)
S
e
n
sit
iv
it
y
(%
)
S
p
e
c
ifi
c
it
y
(%
)
P
re
c
isio
n
(%
)
Re
c
a
ll
(%
)
F
1
-S
c
o
re
(%
)
P
er
ce
n
tag
e
v
alu
es
(
%)
P
er
f
o
r
m
an
ce
Me
tr
ics
Cla
s
s
if
ica
t
io
n Re
s
ult
AH
P
D + GA
a
n
d
AH
P
D + ICA + S
VM
Cla
ss
ifi
c
a
ti
o
n
AH
P
D
+
GA
+
S
V
M
AH
P
D + GA
a
n
d
AH
P
D + ICA + S
VM
Cla
ss
ifi
c
a
ti
o
n
AH
P
D
+
I
C
A
+
S
VM
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
1
6
9
3
-
6
9
3
0
TEL
KOM
NI
KA
T
elec
o
m
m
u
n
C
o
m
p
u
t E
l
C
o
n
tr
o
l
,
Vo
l.
23
,
No
.
5
,
Octo
b
e
r
20
25
:
1
2
9
1
-
1
3
0
3
1300
T
h
e
p
r
ec
is
io
n
o
f
8
0
.
8
2
%
in
d
i
ca
tes
th
at
8
0
.
8
2
%
o
f
t
h
e
m
o
d
el’
s
h
u
m
an
p
r
esen
ce
class
if
icatio
n
s
wer
e
co
r
r
ec
t,
s
h
o
win
g
t
h
at
th
e
p
r
ed
ictio
n
s
m
ad
e
wer
e
g
en
e
r
al
ly
r
eliab
le.
T
h
e
b
ala
n
ce
b
et
wee
n
p
r
ec
is
io
n
an
d
s
en
s
itiv
ity
is
es
s
en
tial
in
ap
p
licatio
n
s
wh
er
e
b
o
th
m
in
im
izi
n
g
f
alse
p
o
s
itiv
es
an
d
m
a
x
im
izin
g
tr
u
e
p
o
s
itiv
es
ar
e
n
ec
ess
ar
y
.
T
h
e
r
ec
all
(
9
0
.
0
0
%)
is
th
e
h
ig
h
est
am
o
n
g
all
th
e
ev
alu
ated
m
et
r
ics,
h
ig
h
lig
h
tin
g
th
at
th
e
m
o
d
el
s
u
cc
ess
f
u
lly
id
en
tifie
d
a
h
ig
h
p
r
o
p
o
r
tio
n
o
f
tr
u
e
h
u
m
an
p
r
esen
ce
ca
s
es.
T
h
is
is
p
ar
ticu
lar
l
y
im
p
o
r
tan
t
in
SAR
o
p
er
atio
n
s
,
wh
er
e
m
is
s
in
g
a
tr
ap
p
ed
v
ictim
co
u
l
d
d
elay
r
escu
e
ef
f
o
r
ts
an
d
r
ed
u
ce
s
u
r
v
iv
al
ch
an
ce
s
.
T
h
e
F1
-
s
co
r
e
o
f
8
0
.
4
1
%
co
n
f
ir
m
s
s
tr
o
n
g
class
if
icatio
n
p
er
f
o
r
m
a
n
c
e,
s
h
o
win
g
a
g
o
o
d
b
alan
ce
b
e
twee
n
p
r
ec
is
io
n
an
d
r
ec
all.
T
h
is
m
ea
n
s
th
at
th
e
m
o
d
el
m
ain
tain
s
o
v
er
all
ef
f
e
ctiv
en
ess
,
m
ak
in
g
it
s
u
itab
le
f
o
r
co
m
p
lex
,
n
o
is
y
en
v
ir
o
n
m
en
ts
wh
er
e
ac
cu
r
ate
h
u
m
an
p
r
esen
ce
d
etec
tio
n
is
cr
itical.
T
h
e
AHPD+
I
C
A+
S
VM
m
o
d
el
d
em
o
n
s
tr
ated
s
lig
h
tly
lo
wer
p
er
f
o
r
m
an
ce
,
ac
h
ie
v
in
g
an
a
cc
u
r
ac
y
o
f
7
9
.
2
0
%.
T
h
is
s
u
g
g
ests
th
at
wh
ile
I
C
A
is
s
o
m
ewh
at
ef
f
ec
tiv
e
in
ex
tr
ac
tin
g
m
ea
n
in
g
f
u
l
f
ea
tu
r
es,
it
m
ay
r
etain
s
o
m
e
r
ed
u
n
d
an
t
o
r
n
o
is
y
co
m
p
o
n
e
n
ts
,
lead
in
g
to
r
e
d
u
ce
d
class
if
icatio
n
ac
cu
r
ac
y
.
T
h
e
7
3
.
0
7
%
s
en
s
itiv
ity
in
d
icate
s
th
at
th
e
m
o
d
el
is
les
s
ef
f
ec
tiv
e
th
an
th
e
GA
-
b
ased
m
o
d
el
in
d
etec
tin
g
h
u
m
an
p
r
e
s
en
ce
,
wh
ich
co
u
ld
in
cr
ea
s
e
th
e
lik
elih
o
o
d
o
f
f
al
s
e
n
eg
ativ
es.
T
h
e
m
o
d
el
h
as
a
ten
d
en
c
y
(
8
4
.
6
9
%)
t
o
m
is
c
lass
if
y
n
o
n
-
h
u
m
an
s
ig
n
als
as
h
u
m
an
p
r
esen
ce
(
s
p
ec
if
icity
)
,
p
o
te
n
tially
lead
in
g
to
m
o
r
e
f
alse
alar
m
s
.
T
h
e
p
r
ec
is
io
n
o
f
8
1
.
0
5
%
s
u
g
g
ests
th
at
wh
en
th
e
m
o
d
el
d
o
es
class
if
y
h
u
m
an
p
r
esen
ce
,
it
is
r
elativ
ely
co
n
f
id
e
n
t
in
its
p
r
ed
ictio
n
.
Ho
wev
er
,
th
e
r
ec
all
(
7
7
.
8
3
%)
in
d
icate
s
th
at
th
e
m
o
d
el
is
l
ess
ef
f
ec
tiv
e
at
ca
p
tu
r
in
g
all
i
n
s
tan
ce
s
o
f
h
u
m
an
p
r
esen
ce
.
T
h
e
F1
-
s
co
r
e
o
f
7
6
.
8
5
% su
g
g
ests
a
wea
k
er
o
v
er
al
l b
alan
ce
b
etwe
en
p
r
ec
is
io
n
an
d
r
ec
all.
T
h
e
p
er
f
o
r
m
an
ce
o
f
AHPD+
I
C
A+
SVM
s
u
g
g
ests
th
at
I
C
A
is
less
ef
f
ec
tiv
e
th
an
GA
in
f
ea
t
u
r
e
ex
tr
ac
tio
n
f
o
r
h
u
m
an
d
ete
ctio
n
in
NL
OS
en
v
ir
o
n
m
e
n
ts
.
Ho
wev
er
,
th
e
s
u
p
er
io
r
p
er
f
o
r
m
an
ce
o
f
AHPD+
GA+
SVM
in
d
icate
s
t
h
at
u
s
in
g
a
GA
f
o
r
f
ea
tu
r
e
s
el
ec
tio
n
en
h
an
ce
s
th
e
m
o
d
el
’
s
ab
ilit
y
to
d
is
tin
g
u
is
h
b
etwe
en
r
elev
an
t
an
d
ir
r
elev
a
n
t
f
ea
tu
r
es,
lead
i
n
g
to
im
p
r
o
v
ed
g
e
n
er
aliza
tio
n
a
n
d
r
o
b
u
s
tn
ess
in
class
if
y
in
g
h
u
m
an
p
r
esen
ce
in
c
h
allen
g
in
g
NL
OS c
o
n
d
itio
n
s
.
T
h
e
ac
cu
r
ac
y
o
f
th
e
GA
-
SVM
h
y
b
r
id
in
cr
ea
s
ed
s
ig
n
if
ican
tly
(
b
y
6
.
5
8
%)
f
r
o
m
7
9
.
2
0
%
to
8
5
.
7
8
%
in
a
co
m
p
lex
a
n
d
d
y
n
am
ic
s
ce
n
a
r
io
co
m
p
a
r
ed
to
I
C
A.
Sen
s
itiv
ity
r
o
s
e
f
r
o
m
7
3
.
0
7
%
to
8
0
.
0
0
%,
d
em
o
n
s
tr
atin
g
GA
’
s
ab
ilit
y
to
o
p
tim
ize
f
ea
t
u
r
e
s
elec
tio
n
.
Ad
d
itio
n
ally
,
th
e
F1
-
s
co
r
e
(
8
0
.
4
1
)
im
p
r
o
v
ed
,
in
d
icatin
g
a
b
etter
b
alan
ce
b
etwe
en
p
r
ec
is
io
n
a
n
d
r
ec
all.
T
h
ese
r
esu
lts
em
p
h
asize
th
e
im
p
o
r
tan
ce
o
f
f
ea
tu
r
e
s
elec
tio
n
in
d
ea
lin
g
with
co
m
p
lex
s
ce
n
ar
i
o
s
wh
er
e
ir
r
elev
an
t f
ea
tu
r
es c
o
u
ld
o
b
s
c
u
r
e
im
p
o
r
tan
t p
atter
n
s
.
T
h
e
AHPD
with
GA
an
d
SV
M
ap
p
ea
r
s
to
b
e
a
r
eliab
le
te
ch
n
iq
u
e
f
o
r
th
r
o
u
g
h
-
wall
d
et
ec
tio
n
.
T
o
f
u
r
th
er
im
p
r
o
v
e
p
er
f
o
r
m
a
n
ce
,
f
u
tu
r
e
wo
r
k
s
h
o
u
ld
f
o
cu
s
o
n
n
o
is
e
r
ed
u
ctio
n
s
tr
ateg
ies
an
d
ad
d
itio
n
al
f
ea
tu
r
e
r
ef
in
em
en
t
tec
h
n
iq
u
es.
T
h
ese
f
in
d
in
g
s
h
ig
h
lig
h
t
th
e
p
o
te
n
tial
o
f
GA
-
en
h
an
ce
d
SVM
m
o
d
els
f
o
r
NL
OS
h
u
m
an
d
etec
tio
n
,
p
ar
ticu
lar
l
y
in
s
ea
r
ch
an
d
r
escu
e
ap
p
li
ca
tio
n
s
wh
er
e
ac
cu
r
ate
v
ictim
id
en
tific
atio
n
is
cr
itical.
6.
CO
M
P
ARA
T
I
V
E
ANA
L
YS
I
S
T
h
is
wo
r
k
p
r
o
v
id
es
a
b
ett
er
way
to
m
ak
e
o
b
s
er
v
atio
n
s
th
an
m
o
r
e
tr
ad
itio
n
al
a
p
p
r
o
ac
h
es.
Fu
r
th
er
m
o
r
e
,
it
ca
n
p
r
o
v
id
e
a
m
o
r
e
p
r
ec
is
e
ass
es
s
m
en
t
o
f
h
u
m
an
d
etec
tio
n
an
d
lo
ca
lizatio
n
d
u
r
in
g
s
ea
r
ch
-
an
d
-
r
escu
e
o
p
er
atio
n
s
.
T
ab
le
2
s
h
o
ws h
o
w
th
is
s
tu
d
y
co
m
p
ar
es to
o
th
er
ap
p
r
o
ac
h
es th
at
h
a
v
e
b
ee
n
r
e
p
o
r
ted
in
th
e
liter
atu
r
e.
7.
CO
NCLU
SI
O
N
T
h
is
s
tu
d
y
p
r
o
p
o
s
es
a
h
y
b
r
id
d
im
en
s
io
n
ality
r
ed
u
ctio
n
ap
p
r
o
ac
h
co
m
b
i
n
in
g
AHPD,
GA,
an
d
I
C
A,
in
teg
r
ated
with
SVM
clas
s
if
ic
atio
n
,
to
en
h
an
ce
v
ictim
lo
ca
l
izatio
n
in
NL
OS
s
ce
n
ar
io
s
f
o
r
SAR
o
p
er
atio
n
s
.
T
h
e
AHPD+
GA+
SVM
m
o
d
el
ac
h
iev
ed
s
u
p
e
r
io
r
p
er
f
o
r
m
an
ce
with
an
ac
cu
r
ac
y
o
f
8
5
.
7
8
%,
d
em
o
n
s
tr
atin
g
its
p
o
ten
tial a
s
a
s
ca
lab
le
an
d
r
o
b
u
s
t so
lu
tio
n
f
o
r
r
ea
l
-
tim
e
d
is
aster
r
esp
o
n
s
e.
Desp
ite
p
r
o
m
is
in
g
r
esu
lts
,
th
e
m
o
d
el
was
test
ed
o
n
co
n
tr
o
ll
ed
d
atasets
,
an
d
its
r
ea
l
-
tim
e
ad
ap
tab
ilit
y
in
u
n
s
tr
u
ct
u
r
ed
en
v
ir
o
n
m
e
n
ts
r
em
ain
s
to
b
e
v
alid
ated
.
L
im
i
tatio
n
s
in
clu
d
e
p
o
te
n
tial
co
m
p
u
tatio
n
al
o
v
er
h
ea
d
an
d
ch
allen
g
es
with
s
en
s
o
r
r
eliab
ilit
y
in
p
r
ac
tical
s
ce
n
ar
io
s
.
Fu
tu
r
e
r
esear
ch
s
h
o
u
ld
ex
p
lo
r
e
r
ea
l
-
wo
r
ld
test
in
g
,
lig
h
tweig
h
t
m
o
d
el
o
p
tim
izatio
n
,
d
ee
p
lear
n
in
g
in
te
g
r
atio
n
,
an
d
m
u
lti
-
s
en
s
o
r
f
u
s
io
n
to
im
p
r
o
v
e
th
e
s
y
s
tem
’
s
r
o
b
u
s
tn
ess
an
d
d
ep
lo
y
ab
ilit
y
in
ac
tu
al
SAR
m
is
s
io
n
s
.
ACK
NO
WL
E
DG
E
M
E
NT
S
T
h
e
au
th
o
r
e
x
p
r
ess
es
g
r
atitu
d
e
to
L
a
n
d
m
ar
k
Un
iv
e
r
s
ity
f
o
r
p
r
o
v
id
in
g
all
th
e
m
ater
ials
r
e
q
u
ir
ed
f
o
r
th
is
r
esear
ch
.
Evaluation Warning : The document was created with Spire.PDF for Python.