Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
8
, No
.
6
,
Decem
ber
201
8,
pp. 5
227~
5237
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v8
i
6
.
pp
5227
-
52
37
5227
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Throu
gh the Wal
l, Reco
gn
ize Moving T
ar
gets Bas
ed on Mi
cro
-
Dopple
r Si
gnatures
Thamir
Rashe
d Saeed
,
M
ahmuod
H
amz
a Al
-
Muifr
aj
e
,
Ghufra
n
M.
Hatem
Rada
r
R
ese
ar
ch Group(R
2
G),
De
par
tment
of
E
lec
tri
c
al
Engi
n
ee
rin
g,
Univer
si
t
y
of
Te
chno
log
y
,
Ira
q
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ja
n
21
, 2
01
8
Re
vised
Ju
l
2
6
,
201
8
Accepte
d
J
ul
29
, 2
01
8
Rada
r is
a
prom
i
sing d
evi
c
e
for
d
et
e
ct
ion
and
r
ecogniti
on
of
invi
s
ibl
e
m
oving
obje
c
ts,
W
her
e,
the
m
ic
ro
-
Doppler
fre
qu
ency
shift
ca
used
b
y
m
oving
the
obje
c
t'
s
p
art
s
ha
ve
bee
n
rep
r
ese
n
te
d
as
an
attra
ct
i
ve
feature
in
th
e
rec
ogni
ti
on
proc
ess.
In
spite
of
tha
t
,
no
thoro
ugh
ana
l
y
sis of
h
um
an
m
ovement
b
y
bi
c
y
c
le
and
no
discri
m
i
nat
e
from
th
e
ru
nning
one
throu
gh
the
wal
l
in
t
he
literature
.
Thi
s
pape
r
pre
sents
a
m
at
hemati
cal
m
odel
of
bicy
cle
m
ovement,
the
n,
th
e
rec
ogni
ti
on
of
m
oving
obje
ct
s
through
the
wall
.
W
her
e
th
ree
hum
an
m
ovements;
wal
king,
runn
ing
an
d
on
a
bi
c
y
cle
h
a
ve
be
en
r
ec
ogni
z
ed
through
two
t
y
pes
of
wal
l
wood
and
cor
k
.
The
the
o
retical
ana
l
y
sis
and
m
e
asure
d
was
give
n
a
re
cognition
98.
7%
for
hum
an
walki
n
g
on
his
fe
et,
99%
of
th
e
passenge
r
on
a
bicy
c
le
,
and
98
%
of
th
e
pe
rson
running
h
av
e
b
ee
n
ac
hi
eved
without
wall
s
.
W
hil
e,
a
95.
4%
,
96.
2%
,
and
9
5%
rec
ogni
ti
on
have
be
en
gai
ned
from
walki
ng,
m
oving
b
y
bicy
cle
and
running
with
wooden
wall
and
94%,
94.
8%
,
an
d
93.
3%
respe
ctively
wi
th
cor
k
wall
.
2
.
4
GH
z
a
s
a
det
e
ct
o
r
and
SV
M a
s c
la
s
sifie
r are
used.
Ke
yw
or
d:
Mi
cro
-
dopple
r fre
qu
e
ncy
Object
recog
niti
on
Thro
ugh wall
detect
ion
Copyright
©
201
8
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Tham
ir Rashed Saeed
,
Dep
a
rtm
ent o
f El
ect
rical
En
gi
neer
i
ng,
Un
i
ver
sit
y o
f Te
ch
no
l
og
y,
Ir
aq.
Em
a
il
:
tha
m
ir_
rash
e
d@y
ah
oo.co
m
1.
INTROD
U
CTION
In
t
he
la
st
de
c
ade,
radar
has
dev
el
op
e
d
a
s
a
prom
isi
ng
in
novative
[
1]
f
or
detect
ing
an
d
recog
nizing
the
m
ov
in
g
obje
ct
s
beh
in
d
the
wall
s
[2
]
-
[
4].
In
this
co
ntext,
the
wall
s
are
com
plex
prop
a
ga
ti
on
en
vir
on
m
ents
that
intr
oduce
refract
ion,
at
te
nu
at
io
n,
ri
ng
i
ng
a
nd
m
ulti
pat
h
to
the
rad
a
r
s
ign
al
[2
]
.
T
he
r
efore,
t
he
knowle
dge
of
m
ov
in
g
ob
j
ect
s
ca
n
be
a
us
e
fu
l
t
oo
l
in
sur
veill
ance,
secu
rity
,
m
il
it
ary
co
m
b
at
,
searc
h
a
nd
r
escue
op
e
rati
ons,
m
e
dical
fiel
ds,
urban
wa
rf
a
re,
a
nti
-
te
rror
ism
,
la
w
e
nfor
c
em
e
nt,
ea
rth
qu
a
ke
survi
vor
dete
ct
ion
,
trap
ped
hum
an
locat
ion,
ho
m
el
and
secu
rity
,
an
d
the
co
nce
al
ed
wea
pon
de
te
ct
ion
[
5]
-
[
8]
.
I
n
this
co
ntex
t,
the
adv
a
ntage
s
of
ra
da
r
c
ompare
d
t
o
th
e
oth
e
r
sens
ors
is
t
he
a
bili
ty
of
operati
on
in
di
ff
e
ren
t
env
i
ronm
ents [9]
-
[
11]
.
The
ex
pected
beh
a
vior
of
th
e
il
lu
m
inate
d
m
ov
ing
obj
ect
s
by
rad
ar
sig
na
l
has
m
od
ulate
d
this
sign
al
at
the
received
ph
ase
ca
us
e
d
by
the
tim
e
-
var
yi
ng
delay
be
tween
the
rad
a
r
an
d
that
obj
e
ct
.
Than
a
Dopp
l
e
r
fr
e
qu
e
ncy
has
been
i
ntrod
uce
d
cause
d
by
th
is
m
otion
,
w
hile,
the
m
ic
ro
-
Dop
pler
w
hich
is
caused
by
m
ic
ro
-
m
ot
ion
-
[2
]
,
[
12]
.
In
case
of
hu
m
an
as
a
non
-
rigid
body,
wh
e
re
his
m
ic
r
o
-
m
otion
is
introd
uced
by
his
par
ts;
the
arm
s,
le
gs
,
torso,
an
d
ot
he
rs.
All
these
par
ts
are
c
ontr
ibu
te
d
t
o
a
uniqu
e
si
gn
at
ur
e
cause
d
by
it
s
m
ic
ro
-
Dop
pler
w
hic
h
is
m
od
ulate
d
on
the
m
ain
D
oppler
whic
h
is
create
d
by
the
torso
[10].
Theref
ore,
the
recog
niti
on
of
m
ov
ing
ob
j
ect
s
is
achieve
d
by
it
s
m
ic
ro
-
Dopp
le
r
(MD
)
fe
a
tures,
w
her
e
,
the
m
ic
ro
-
m
oti
on
has
a
uniq
ue
si
gna
ture
at
dif
fer
e
nt
ra
nges
[13],
[
14
]
.
This
sig
natu
re
will
be
reduce
d
to
m
i
nim
iz
e
the
ef
f
or
ts
of
dig
it
al
sign
al
proc
es
sin
g
ap
plica
ti
on
s
[
15]
-
[
18]
.
It
is
oft
en
be
nef
ic
ia
l,
the
c
ho
ic
e
of
t
he
f
re
qu
e
ncy
of
op
e
r
at
ion
dep
e
nds
on
th
e
app
li
cat
ion,
sp
eci
fical
ly
on
the
barrier
ty
pe,
ta
rg
et
po
sit
io
n
be
hind
the
wall
,
sta
ndoff
requirem
ent, a
nd r
es
olu
ti
on
r
equ
i
rem
ents, all
o
f
which
are som
ewh
at
inter
relat
ed
[
8].
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8: 5227
-
5
237
5228
As
f
or
t
he
th
r
ough
-
wall
ap
pl
ic
at
ion
s,
w
hic
h
is
re
pr
ese
nte
d
an
is
a
pp
e
al
ing
r
esearc
h
a
rea
there
is
gr
eat
resea
rc
h
value
f
or
hu
m
an
detect
ion
and
r
eco
gnit
ion
us
in
g
ra
da
r
te
chnolo
gy.
Wh
e
re,
L
ui
et
al
.
[19]
stud
ie
d
the
m
otion
detect
io
n
and
recog
niti
on
base
d
on
M
D
thr
ough
the
ho
m
og
e
neous
non
-
m
agn
et
ic
wal
ls,
wh
e
re
t
hey
sho
w
that
t
he
si
gnat
ur
es
a
re
unal
te
red
as
co
rro
borate
d
by
Ra
m
et
al
.
I
n
[20].
I
n
this
co
ntext,
Chen
et
al
.
in
[10],
[
21
]
prese
nted
a
recog
niti
on
s
tud
y
thr
ou
gh
t
wo
dif
fer
e
nt
10
cm
-
thick
w
al
ls,
con
c
rete
blo
c
ks
and
plasti
c,
f
or
m
edical
app
li
ca
ti
on
.
In
[6
]
was
c
on
ce
ntrat
ed
to
the
thr
ou
gh
t
he
wall
s
cl
assifi
cat
ion
ac
cur
acy
by
us
in
g
the
s
upport
ve
ct
or
m
achine
f
or
S
-
ba
nd
at
long
range.
Wh
il
e
[
22
]
,
is
e
xtracted
the
ef
fected
m
ic
r
o
sign
al
by
th
res
ho
l
d
se
gm
entat
ion
,
a
nd
the
tr
ai
nin
g
phase
is
done
by
us
i
ng
SV
M
with
ac
cur
acy
is
gain
ed
by
95.4 %.
Howe
ver,
de
spi
te
these
stu
dies,
the
re
is
sti
ll
a
ga
p
i
n
the
di
sti
nction
betw
een
the
detect
ion
an
d
t
he
recog
niti
on
t
hroug
h
the
wall
for
t
he
r
unni
ng
pe
rs
on
fro
m
that
whose
m
ov
ing
by
bic
yc
le
and
t
her
e
is
no
m
at
he
m
at
ic
a
l
m
od
el
of
t
hese
m
ov
e
m
ents.
I
nd
ee
d
t
he
i
ncrea
sing
of
th
e
us
a
ge
of
t
he
non
-
m
agn
et
ic
m
at
erial
li
ke
w
ood
a
nd
cork
as
wall
'
s
i
ns
ulati
on
la
ye
r
especial
ly
in
the
ho
s
pital
s,
w
hile
there
is
a
gap
in
the
rese
arc
h
area
of
detect
ion an
d rec
ogni
ti
on
th
r
ough th
ese la
y
ers.
Ther
e
f
or
e,
this
pa
per
f
oc
us
es
on
the
rec
ognit
ion
of
t
he
hu
m
an
act
ivit
ie
s
th
rou
gh
the
w
oo
d
a
nd
c
ork
la
ye
rs.
Also
,
t
he
m
a
the
m
at
ical
analy
sis
of
t
he
hum
an
m
ov
ing
by
a
bicy
cl
e
is
pr
esent.
T
hen,
the
eff
ect
on
the
m
ic
ro
wa
ve pr
opagati
on t
hro
ugh t
his lay
er,
a
nd the
rec
ogniti
on
of the
hum
an
m
ov
em
ent h
ave
b
ee
n p
res
ented
2.
MICRO
-
DOP
PLE
R
EFFE
CT I
N RAD
A
R
[
14]
To
analy
ze
the
m
ov
in
g
ob
j
ect
at
po
int
P
with
m
ov
ing
fr
e
quency
fv
at
dis
ta
nce
R
with
di
sp
la
cem
ent
Dv as s
how
n
i
n
Fi
gure
1
,
the
m
at
he
m
at
ic
s o
f
it
s m
ic
ro
-
Dop
p
le
r
e
ff
ect
is;
The displac
em
ent fu
nction i
s
D(t)
=
D
v
sin(2
π
fv
t
)
co
s β cos α
p
(ass
um
ing
α
=
0
a
nd
β
p
=
0)
The ran
ge f
un
c
ti
on
var
ie
s
wit
h
ti
m
e d
ue
to
the tar
get m
ic
ro
-
m
otion
is;
R(
t
)
=
R +
D(t
)
.
Figure
1. Ge
om
et
ry f
or r
a
dar an
d
m
ov
in
g p
oin
t
obj
ect
The
rad
a
r recei
ved sig
nal
bec
om
es [
1]
(
)
=
(
(
2
+
(
)
)
)
(
)
=
4
(
)
(1)
wh
e
re;
ρ
-
bac
ks
cat
te
rin
g
c
oe
ff
ic
ie
nt.
f
o
-
ca
rr
ie
r
fre
qu
e
ncy.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Th
r
ough t
he
W
all, Rec
ognize
Movin
g Tar
get
s Ba
s
ed
on Mic
ro
-
D
oppler
Signa
t
ur
es
(
T
hami
r Ra
s
he
d S
aee
d
)
5229
λ
-
ca
rr
ie
r wav
el
eng
th
.
Substi
tuti
ng th
e eq
uation o
f
R
(t)
in
equati
on
(1),
a
nd the
n d
erive its
ph
a
se
to pr
oduce t
he m
ic
ro
-
D
opple
r
as;
=
cos
cos
cos
(
)
(2)
In this c
on
te
xt,
the
rou
nd
-
trip
return tim
e
τ
R
for
a tar
get locate
d
at
a
ra
ng
e
of
R i
s g
ive
n by
[
8];
=
2
(3)
3.
WALL EF
FE
CTS
In
th
r
ough
-
the
-
wall
ra
dar
des
ign
a
nd
it
s
ret
urn
sig
nal
anal
ysi
s,
the
pro
pe
rtie
s
of
that
w
al
l
rep
rese
nt
crit
ic
al
f
act
or
s w
hic
h
aff
ect
e
d on
the att
enu
a
ti
on
, d
is
per
si
on, r
e
flect
ion
, a
nd
r
e
fr
act
i
on
of the
return signal
and
then
on
the
D
opple
r
sign
at
ur
e
s
[23],
[10].
T
he
eff
ect
of
th
e
wall
s
on
the,
Tx/Rx
bel
ow
X
ba
nd,
ra
dar
sign
al
s
are
co
ncen
tr
at
ed
in
two
po
i
nts
;
the
first
on
e
is
the
eff
ect
on
the
sign
al
phas
e,
this
eff
ect
is
si
m
ple
and
co
nst
ant
because
o
f
th
e w
al
l i
s stat
ion
ary. Th
e sec
on
d effect i
s o
n
th
e
m
axi
m
u
m
d
e
te
ct
able ran
ge
, w
he
re,
the sig
na
l
-
to
-
no
ise
rati
o
(S
/
R)
of
t
he
MD
sign
at
ur
e
is
de
crease
cau
sed
by
the
wall
at
te
nu
at
io
n
[2
]
,
[
24
]
.
In
this
c
on
te
xt,
the
pro
per
ti
es
of
the
wall
w
hich
is
pr
od
uced
t
he
at
te
nu
at
io
n
are;
the
pe
rm
i
t
ti
vit
y
(ε)
,
perm
eabil
it
y
(µ)
,
and
t
he
cond
uctivit
y
(
σ),
w
her
e
,
al
l
these
pros
per
it
ie
s
are
var
ie
d
on
the
f
reque
nc
y
of
the
i
nci
den
t
wa
ve.
I
f
these
m
at
erial'
s p
rop
erti
es are
know
n,
t
hen the att
e
nu
at
io
n
c
onsta
nt can be
app
roxim
a
te
d
as [2
3]
, [
25]
;
=
√
√
1
+
(
1
)
2
−
1
2
for
lo
w
-
lossy
m
at
erial
as w
ood ca
n ap
pro
xi
m
at
e as
=
0
.
5
√
=
0
.
5
√
=
(4)
wh
il
e the
inci
de
nt w
a
ve velo
c
it
y t
hr
ough t
he
m
at
erial
is[25
]
;
=
√
1
+
√
1
+
(
)
2
2
wh
e
re;
=
,
=
2
,
=
1
−
∴
=
/
√
=
4
10
-
7
H/m
.
=
8.854
18782 ×
10
-
12
m
-
3
kg
-
1
s
4
A
2
Ob
se
r
ving
the
D
oppler
featu
res
of
t
he
ref
l
ect
ed
sig
nal
e
nab
le
s
the
cl
utter
to
be
easi
l
y
su
pp
resse
d
because
,
in
ge
ner
al
,
tra
diti
onal
cl
utter
el
e
ments
su
c
h
as
wall
s,
fu
r
nitur
e
,
and
oth
er
sta
ti
on
a
ry
obj
ect
s
will
no
t
exh
i
bit a D
oppl
er s
hi
ft.
It
is
wo
rt
h
noti
ng
that,
when
t
he
wall
is
con
s
tructed
f
r
om
the
wood
or
glass
or
co
r
k
or
dry
wall
,
the
at
te
nu
at
io
n
is
relat
ively
s
m
all.
Con
s
e
quentl
y,
the
fr
e
quen
cy
sel
ect
ion
is
i
m
po
rtant
f
or
app
li
cat
ions
be
cause
the
m
at
erial
p
roper
ti
es
are
dep
e
ndent
on
it
.
[2
3].
H
oweve
r,
if
a
sta
ti
on
ary
ta
r
get
vib
rates,
ro
ta
te
s,
or
m
aneu
ve
rs,
it
s
struct
ur
al
par
t
s
are
in
m
otion
,
a
nd
these
i
nduce
the
m
ic
ro
-
D
oppl
er
m
od
ulati
on
s
.
I
f
the
re
ar
e
=
1,
2,
.
.
.
,
str
uctu
res
with
,
bein
g
the
ra
dia
l
velocit
y
of
th
e
ℎ
str
uctu
re,
t
he
com
po
sit
e
m
i
cro
-
D
opple
r
sig
nal
has
fr
e
quency
com
pone
nts
at
2
V
r
,
i
f
o
/
c
∀
i
=
1,
2,
.
.
.
,
N
,
w
hi
ch
a
re
uniq
ue
to
the
s
pecific
m
otion
al
char
act
e
risti
cs o
f
the tar
get.
An
al
ysi
s
of
th
e
m
ic
ro
-
Dop
pler
sig
natu
res
i
n
the
joint
ti
m
e
-
f
reque
ncy
dom
ai
n
can
pro
vid
e
us
e
fu
l
inf
or
m
at
ion
f
or
ta
r
get
detect
ion
,
cl
assifi
cat
ion
,
an
d
rec
ogniti
on
[8
]
,
[
26]
.
T
her
e
f
or
e,
the
m
ic
ro
-
Dopp
le
r
sign
at
ur
es
f
r
om
the
scene
are
cal
culat
ed
by
app
ly
ing
a
ti
m
e
-
fr
e
qu
e
ncy
trans
form
s,
su
ch
as
the
Sho
r
t
-
Ti
m
e
Four
ie
r
T
ra
ns
f
or
m
(S
TFT)
w
it
h
a
Gau
s
sia
n
window,
t
o
the
scat
te
red
fi
el
ds
.
T
hen,
the
com
pu
ta
ti
on
of
this
trans
form
is
based
on
the
pro
per
sam
pling
r
at
e,
w
hile
this
sam
pling
rate
dep
e
nds
on
the
m
axi
m
u
m
fr
equ
e
nc
y
Dop
pler
s
hift,
wh
e
re,
it
d
e
pends on t
he
r
at
e
of m
otion
[9].
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8: 5227
-
5
237
5230
4.
E
X
PERI
MEN
TAL C
AS
E
S
TUD
Y
The
e
xperim
ental
work
f
or
de
te
ct
ion
a
nd
rec
ogniti
on
of
the
hum
an
m
ov
em
ents
be
hind
t
he
wall
ha
ve
been d
one
with
the s
pecific
pa
r
am
et
ers
as in
Table
1,
t
hese
par
am
et
ers
are;
Table
1.
E
xper
i
m
ental
Par
am
et
ers
No
.
Para
m
eter
Qu
ality
Ty
p
es
1
W
all thick
n
ess
2
0
c
m
wo
o
d
en
2
Bo
d
y
parts sp
eed
2
-
5
m
/sec
Hu
m
an
3
Dv
0
.01
m
4
U
0
m
5
V
2
-
5
m
Hu
m
an
6
W
1
.66
m
Hu
m
an
7
Carrie
r
f
requ
en
cy
2
.4 Gh
z
Rad
ar
Para
m
eters[
1
9
]
8
BW
4
0
0
M
h
z
9
Sa
m
p
lin
g
128
10
An
t.
Po
lariza
tio
n
Vertica
l
11
Sweep
T
i
m
e
FMC
W
-
S
awto
o
th
0
.5
m
sec
The
hum
an
body
m
od
el
has
been
base
d
on
the
m
od
el
wh
i
ch
is
pr
ese
nted
in
[7
]
,
[8
]
,
[
9]
as
sh
own
i
n
Figure
2.
(a)
Ra
dial veloc
it
ie
s f
or d
if
fere
nt bo
dy joints
in on
e
cyc
le
(b)D
i
ff
e
ren
t m
ov
i
ng bo
dy'
s
par
ts
Figure
2. H
uma
n body'
s p
arts
v
el
ociti
es
w
he
re;
=
relat
ive v
el
ocity
of
c
yc
le
=
aver
age
vel
ocity
/ heigh
t
of the t
high
=
3/
0.
83=
3.614
4
=
relat
ive len
gth
of
the cyc
le
=
1.346
√
=
2.558
9
=
/
--
durati
on
of
walkin
g
cy
cl
e
=
2.558
9/3
.
6155
=
0.7
07 sec
f
Doppler
=
2
v
R
/
λ
c
=
(
32
-
80) Hz
v
R
-
rate o
f
m
otion(2
-
5)m
/se
c.,
The
sam
pling
rate
is
directl
y
pr
op
or
ti
onal
to
the
m
axi
m
u
m
sp
eed
obser
ved.
It
fo
ll
ows
that,
fo
r
a
range
of
sp
ee
d
of
2
-
5
m
/s,
th
e
Dopp
le
r
sh
if
t
can
var
y
signi
ficantl
y
based
on
the
operati
ng
f
re
qu
e
ncy
of
th
e
rad
a
r.
F
or
insta
nce,
us
in
g
a
2.4
G
Hz
rad
a
r
re
su
lt
s
in
a
Dop
pl
er
sh
ift
of
(32
-
80)
Hz,
wh
ic
h
requires
a
sam
pling
rate (
64
-
160) s
a
m
ples/s. I
n
thi
s cont
ext,
the
m
od
el
o
f
t
he h
um
an
m
ov
in
g on a
bicy
cl
e as shown i
n
Fi
gu
re
3
is;
The
li
nea
r
m
ove
m
ent o
f
lag
=
D
m.
The
ci
rc
um
fer
e
nce
of a circl
e
C1
is
relat
ed
to
2
D
m
.
C1
=
K
C2
m.
=
K×π×
R
m.
The
s
pee
d of b
ic
yc
le
=
Z ×
K×
π×
R
m
/se
c.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Th
r
ough t
he
W
all, Rec
ognize
Movin
g Tar
get
s Ba
s
ed
on Mic
ro
-
D
oppler
Signa
t
ur
es
(
T
hami
r Ra
s
he
d S
aee
d
)
5231
w
he
re;
K
-
pe
rc
ent of
C
1/C2
.
Z
-
no. of
t
urn
/sec.
R
-
rad
i
us
of th
e
rear
wheel
.
L
=
u
c
os(β)
+
Y
c
os
(α
)
The
n
the
Do
pple
r
f
reque
ncy of the
hum
an
m
ov
i
ng on bic
yc
le
=
2
×
2
(
)
≅
2
=
6.315
2
Hz
5.
RESU
LT
S
AND DI
SCUS
S
ION
The
e
xp
e
rim
ents
hav
e
bee
n
done
i
n
ou
r
la
borat
or
y
by
us
i
ng
a
n
An
c
ort
ek
I
nc.
S
-
band
r
adar
kit
[27]
as
sh
ow
n
in
F
i
gure
4
(
a)
.
Wh
e
re
from
the
resu
lt
s
of
the
se
ex
per
im
ents,
the
abili
ty
of
detect
ion
an
d
rec
og
niti
on
of
the
m
ov
in
g
body
be
hind
th
e
woo
den
wa
ll
is
evi
den
t
a
nd
sh
ows
in
t
he
F
igures
(
5
-
9).
Wh
e
re
th
ree
di
ff
e
ren
t
hu
m
an
m
ov
em
ents
ha
ve
bee
n
exam
ined;
H
um
an
walki
ng,
r
unning
a
nd
m
ov
in
g
by
the
bicy
cl
e
as
exp
la
in
ed
in
F
igures
(
2,3)
,
with tw
o dif
fere
nt w
al
ls;
woo
den an
d
c
ork
a
s in
Fi
gure
3
a
nd 4.
Figure
3.
H
uma
n
m
ov
in
g on
bicy
cl
e m
od
el
(a)
Ra
dar
be
hind the
w
al
l
(b)
W
al
king
person on t
he othe
r
w
al
l si
de
Figure
4. The
e
xp
e
rim
ent in libr
at
or
y,
(a) t
he
r
a
dar
i
n woo
de
n wall
side,
(b
)
m
ov
in
g perso
n
in
anothe
r
woo
den w
al
l si
de
The
m
ov
in
g H
um
an
body
part
s,
ra
dial vel
ociti
es, and r
el
at
e
d Dop
pler fre
quencies
are
li
st
ed
in
the
Table
2
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8: 5227
-
5
237
5232
Table
2.
H
um
a
n
B
od
y
Parts
V
el
ociti
es and R
el
at
ed
D
oppler
Fr
e
quencies
H.B.P.
*
Sp
eed
m
/sec
Do
p
p
ler
f
requ
en
cy
(
Hz
)
Han
d
sp
eed
-
0
.7
—
+2
1
1
.2
—
32
Kn
ee
-
0
.7
—
+2
.1
1
1
.2
—
3
3
.6
An
k
le
-
0
.2
—
+4
.95
3
.2
—
7
9
.2
Toe
-
0
.8
—
+
4.8
1
2
.8
—
7
6
.8
In
the
be
ginnin
g,
the
la
bo
ratory
wall
at
5.
6
m
fr
om
the
rad
a
r
.
The
po
wer
frequ
e
ncy
intens
it
y
against
range
is
sh
ow
n
in
Fig
ur
e
5
(
a),
w
hich
is
re
la
te
d
to
the
lab
orat
ory
wall
and
the
respo
nse
of
the
pe
rs
on
in
the
la
borator
y
without
m
ov
in
g
at
2.8m
fr
om
the
rad
a
r.
Also,
th
e
Dop
pler
fr
e
quency
range
of
that
wall
an
d
per
s
on
app
ea
rs
i
n
the
Figure
5
(
b), a
nd it
'
s clea
r
the
Dop
pler
is ze
r
o becau
se the
r
e is n
o
m
otion
.
(a)
P
ower/f
re
quency agai
ns
t r
a
ng
e
(b)Tim
e agains
t fr
e
qu
e
ncy
Figure
5. L
ab
orat
or
y
wall
5.6
m
an
d
a stan
di
ng p
e
rs
on at 2
.
8m
f
ro
m
the r
a
dar
The
m
ov
em
en
t
of
the
per
s
on
an
d
his
par
t
s
le
ads
to
the
gen
e
rate
the
D
oppler
an
d
m
i
cro
-
D
opple
r,
wh
ic
h
is
relat
e
d
to
t
he
par
ts
ve
locit
ie
s
as
in
F
igure
6.
Whe
re,
the
vel
ociti
es
ha
ve
been
a
pp
ea
rin
g
as
i
n
Figure
6
(
a),
w
hic
h
re
pr
ese
nts
t
he
diff
e
ren
t
velocit
ie
s
of
th
e
Hu
m
an
body
par
ts
against
t
he
range.
I
n
this
co
nt
ext,
th
e
Dop
pler
fr
e
qu
ency
caused
by
these
velocit
ie
s
is
cl
eared
in
Figu
re
6
(
b),
wh
e
re
the
app
ea
ra
nce
of
these
Dop
pler wit
h d
iffer
e
nt ti
m
e d
el
ay
caused b
y
the m
ov
em
ent o
f
the
perso
n
a
gainst t
he
fi
xe
d rada
r.
(a)
Ra
dial Velo
ci
ty
ag
ai
ns
t ra
nge
for
m
ov
in
g perso
n
(b)Tim
e agains
t fr
e
qu
e
ncy
Figure
6.
Lab
orat
or
y
wall
5.6
m
an
d
a m
ov
in
g perso
n
at
2.8
m
f
ro
m
the r
ad
ar
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Th
r
ough t
he
W
all, Rec
ognize
Movin
g Tar
get
s Ba
s
ed
on Mic
ro
-
D
oppler
Signa
t
ur
es
(
T
hami
r Ra
s
he
d S
aee
d
)
5233
The
com
par
iso
n
betwee
n
Fig
ur
e
5
an
d
6
,
wh
e
re,
the
difference
is
cl
ear
wh
er
e
the
person
a
ppear
s
without
m
ov
e
m
ent
in
Fig
ur
e
5
, while
in
Fig
ur
e 6
t
he
pe
rs
on
is
in
m
otion
.
I
n
t
his
co
ntext
,
the
dis
persi
on o
f
t
he
ref
le
ct
ed
sig
na
l
caused
by
the
wood
e
n
wall
has
al
so
ap
pea
red
in
the
Fig
ure
7.
Wh
il
e,
Figure
8
re
pr
e
se
nts
the
respo
ns
e
a
nd
velocit
ie
s
ra
nge
distrib
utio
n
of
t
he
body
pa
rts
w
hich
is
de
te
ct
ed
thr
ough
that
wall
,
where
the
par
ts
w
hich
ha
ve
a
m
ov
em
ent
le
ss
than
12.
5
cm
fr
o
m
the
cor
e
of
the
body
(to
rso),
it
will
appear
a
s
one
m
ov
ing
obj
ect
,
b
eca
us
e the
op
erati
ng freq
ue
nc
y has
wav
el
e
ngth is
12.
5
cm
, and
the
noise
wh
ic
h
a
pp
ea
rs i
s b
y
wall
eff
ect
e
d o
n
th
e
sig
nal
prop
a
gatio
n
.
Figure
7
(a)
. F
i
xed p
e
rs
on b
e
hi
nd
w
ood wall
(2.8
m
)
from
r
adar
Figure
7 (b)
.
Radial
v
el
ociti
es again
st a ra
ng
e of
m
ov
ing
per
s
on
Fo
rt
unat
el
y,
th
e
hum
an
on
t
he
bicy
cl
e
is
m
ov
i
ng
his
le
gs
,
not
his
a
rm
,
t
her
e
fore,
t
his
note
has
bee
n
us
e
d
f
or r
ec
og
nizing bet
wee
n t
hese m
ov
em
e
nts. Also
, th
e
bi
cy
cl
e w
heels a
pp
ea
r
it
s
veloci
ty
as in
Fi
gure
8,
Figure
8. Ra
di
al
v
el
ociti
es ag
ai
ns
t a ra
ng
e
for
m
ov
in
g
a
person at
2.8 m
w
it
h
bicy
cl
e
beh
i
nd wo
od
W
al
l
from
rad
a
r
Anothe
r
wall
has
bee
n
e
xa
m
ined,
co
r
k
wall
wh
ic
h
is
us
ed
widely
in
m
any
app
li
cat
ion
s
as
a
n
isolat
ion
lay
er
.
Th
e
r
es
pons
e
of hum
an
m
ov
e
m
ents b
e
hind
the cor
k wall
is shown i
n
Fi
gu
re
9
.
The
diff
e
re
nce
in
res
ponse
s
be
tween
t
he
rec
ogniti
on
of
hum
an
m
ov
e
m
ents
beh
i
nd
the
woo
den
wall
and
the
c
ork
wall
is
cause
d
by
their
inter
nal
str
uctu
re.
Wh
e
re
t
he
int
ern
al
st
ru
ct
ur
e
of
the
w
ood
is
m
ore
su
it
able
f
or
det
ect
ion
at
2.
4
G
Hz
tha
n
a
c
ork
wall
.
Fig
u
re
10
sho
wn
t
he
ef
fect
of
t
he
inte
rn
al
wall
struct
ur
e
,
wh
e
re
t
he
de
viati
on
of
the
pr
op
a
gatio
n
wa
ve
thr
ough
t
hes
e
wall
s,
especi
al
ly
through
th
e
co
rk
wall
,
w
hile
the
disp
e
rsion
of the
pro
pag
at
io
n wa
ve
th
r
ough
the wo
od
e
n wa
ll
is cle
ar.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8: 5227
-
5
237
5234
(a)Dop
pler
fr
e
qu
e
ncy
with ti
m
e fo
r
hum
an
body
par
ts
duri
ng jo
gg
i
ng
(b)Do
pp
le
r
f
re
qu
e
ncy
with ti
m
e fo
r
p
a
rts
of the
hu
m
an
bo
dy whil
e m
ov
ing o
n a bicy
cl
e
Figure
9. D
oppl
er freque
ncy
with ti
m
e fo
r p
arts of t
he hu
m
an body t
hr
ough the
w
al
l
The
diff
e
re
nce
in
res
ponse
s
be
tween
t
he
rec
ogniti
on
of
hum
an
m
ov
e
m
ents
beh
i
nd
the
woo
den
wall
and
the
c
ork
wall
is
cause
d
by
their
inter
nal
str
uctu
re.
Wh
e
re
t
he
int
ern
al
st
ru
ct
ur
e
of
the
w
ood
is
m
ore
su
it
able
f
or
det
ect
ion
at
2.4
G
Hz
tha
n
a
c
ork
wall
.
Fig
ur
e
10
sho
wn
t
he
ef
fect
of
t
he
inte
rn
al
wall
struct
ur
e
,
wh
e
re
t
he
de
viati
on
of
the
pr
op
a
gatio
n
wa
ve
thr
ough
t
hes
e
wall
s,
especi
al
ly
through
th
e
co
rk
wall
,
w
hile
the
disp
e
rsion
of the
pro
pag
at
io
n wa
ve
th
r
ough
the wo
od
e
n wa
ll
is cle
ar.
Figure
10. T
he
eff
ect
of
wall
inn
e
r
st
ru
ct
ur
a
l
on the
pro
pa
ga
ti
on
wav
e
th
rough
it
, (a)
with
ou
t a
w
al
l,
(
b)
c
ork
wall
and
(c)wood
e
n wall
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Th
r
ough t
he
W
all, Rec
ognize
Movin
g Tar
get
s Ba
s
ed
on Mic
ro
-
D
oppler
Signa
t
ur
es
(
T
hami
r Ra
s
he
d S
aee
d
)
5235
Ther
e
f
or
e,
to
c
la
ssify
these
t
hree
m
otion
s
wi
th
a
nd
with
out
wall
a
t
hr
ee
S
VM
has
bee
n
us
e
d
as
in
pro
po
se
d
cl
ass
ifie
r,
an
d
the
cl
assifi
cat
ion
al
go
rithm
has
been
show
n
i
n
Figure
11.
Where,
acco
rd
i
ng
to
the
velo
ci
ty
of
the
le
gs
and
arm
as
in
Figure
2
and
T
a
ble
2
w
her
e
it
is
us
ed
as
featur
es
,
can
cl
assify
the
sign
a
l
from
hu
m
an
as;
walkin
g,
r
unning
or
m
ov
e
by
bicy
cl
e.
Wh
ere
,
if
the
ar
m
is
in
m
otion
the
bo
dy
is
w
al
kin
g
or
run
ning
an
d
cl
assify
betwee
n
these
tw
o
m
otion
s
by
the
t
orso
velocit
y
(m
ai
n
D
oppler
).
Well
,
if
the
a
r
m
is
no
t
m
ov
ed
an
d
t
he l
egs
are
m
ov
ed
then the
body
is a m
ov
e b
y
bi
cy
cl
e.
The
n
the
seco
nd
an
d
t
hir
d
c
la
ssifie
r
is
w
orki
ng
with
the
sp
ee
d
of
the
le
gs
(
K
nee
a
nd
Ankle)
a
s
featur
e
s
f
or
hum
an
m
otion
decisi
on
a
nd
cl
assifi
cat
ion
am
on
g;
wal
ki
ng
,
r
unni
ng
or
m
ov
e
by
bicy
cl
e.
Ther
e
f
or
e,
acc
ordin
g
to
the
hum
an
bo
dy
pa
r
ts
velocit
ie
s,
the
cl
assifi
cat
ion
of
dif
fer
e
nt
hum
an
m
ov
ing
sta
tus
has
bee
n
done
by
us
in
g
SV
M
as
a
cl
assi
fier
and
us
e
the
m
i
cro
-
D
opple
r
ef
fect
as
a
feature
,
and
the
re
sul
ts
as
sh
ow
n
in
Ta
ble 3
.
Figure
11. Clas
sifie
r
fl
ow
c
ha
rt
Table
3.
Rec
og
niti
on
of
Diff
e
r
ent Mo
ving B
odie
s
with T
wo
Wall
s Type
Mov
in
g
bo
d
y
Reco
.%
with
o
u
t wall
Reco
.
% with
wo
o
d
en
wall
Reco
.% with
Co
rk wall
W
alk
in
g
Hu
m
an
98
.7
9
5
.4
94
Hu
m
an
on
the
b
icy
cle
99
9
6
.2
9
4
.8
Ru
n
n
in
g
Hu
m
an
98
95
9
3
.3
6.
CONCL
US
I
O
N
Thro
ugh
-
wall
detect
ion
a
nd
cl
assifi
cat
ion
are
vital
in
m
any
app
li
cat
io
ns
.
Wh
il
e
ther
e
are
m
any
i
m
po
rtant
fact
or
s
,
ha
ve
bee
n
effe
ct
ed
on
the
detect
ion
and
cl
assifi
cat
ion
w
hic
h
are
con
cl
ud
e
d
f
r
om
this
exp
e
rim
ental
work.
The
se
f
act
or
s
are;
sig
nal
stren
gth
,
wa
ll
prosp
e
riti
es
(inter
nal
w
al
l
structur
e
),
detect
ion
distance,
car
rier
f
reque
ncy,
a
nd
the
am
ou
nt
of
dis
placem
ent
of
m
ov
ing
bo
dy
pa
rts.
Wher
e
2.4
G
Hz
has
bee
n
us
e
d,
t
he
detect
able
dis
placem
ent
in
our
e
xperim
ents
is
12.
5
cm
,
an
d
it
is
achieve
d
i
n
so
m
ebo
dies
pa
rts
i
n
hu
m
an
walki
ng,
r
unni
ng
a
nd
hu
m
an
on
th
e
bicy
cl
e.
Ther
ef
or
e
,
the
rec
ogniti
on
wh
ic
h
has
been
ga
ined
is
98.7%
of
the
hum
an
walking
and
98%
for
hum
an
ru
nnin
g,
wh
il
e,
99%
of
hu
m
an
on
the
bicy
cl
e
m
ov
em
ent
without
wall
s.
W
hile,
t
he
re
cogniti
on
w
hi
ch
is
g
ai
ned,
with
the
w
oode
n
wall
,
is
95
.4
%
for
a
hum
an
is
walkin
g,
and 95%
f
or
a
hum
a
n
is
r
unni
ng,
w
hile,
96.
2%
for
hu
m
an
m
ov
in
g
by
a b
ic
yc
le
.
In
a
nothe
r
w
ord,
th
e
recog
niti
on
th
r
ough
the
c
ork
wall
is
reduce
d
by
1.5
59%
f
rom
that
of
the
woo
den
wall
,
cause
d
by
it
s
inter
nal
structu
re.
In
t
hi
s
con
te
xt,
the cla
ssific
at
ion
is
base
d
on
the h
an
d,
Knee
an
d
A
nkle
m
ov
em
ent
as
a
hu
m
an
bo
dy
par
t
by
us
in
g
S
VM as a
classi
fier.
ACKN
OWLE
DGE
MENTS
We
w
ou
l
d
li
ke
to
thank
prof
.
Dr.
Jaw
ad
K.
Al
i
fo
r
hi
s
su
pp
or
t
an
d
adv
ic
e
in
accom
plishing
our wor
k
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8: 5227
-
5
237
5236
REFERE
NCE
S
[1]
Vince
nt
R.
Rad
zi
ck
i,
D
avi
d
Bo
utt
e
,
Paul
V.
T
a
y
lor,
and
Hua
Le
e
.
“
Standoff
CW
Rada
r
for
Through
-
the
-
W
a
ll
Dete
c
ti
on
of
Hu
m
an
Hea
rtb
ea
t
Signat
ur
e
”
s.
IEEE
Radar
Conf
ere
n
ce
(
RadarConf)
,
2016.
[2]
Shobha
Sundar
Ram
and
Ang
sh
ul
Majumdar.
“
Through
-
W
al
l
Propaga
t
ion
Eff
e
ct
s
on
Doppler
-
Enha
nc
ed
Fronta
l
Rada
r
Im
ag
es
of
Hum
ans
”
.
IE
EE R
adar Conf
ere
n
ce
2016
.
[3]
Ti
an
Jin,
Yongping
Song,
Bi
y
in
g
Lu
and
Zhi
m
in
Zhou.
“
A
nov
el
through
-
the
-
wall
image
form
at
ion
for
vir
tua
l
ape
rtur
e
rad
ar
with
elec
t
rom
agne
tic
wave
r
efr
ac
t
ion
eff
ec
t
co
rre
ction
”
.
URSI
Gene
ral
Assem
bly
and
S
ci
en
ti
f
ic
Symposium (
UR
SI
GASS)
,
2014
”
.
[4]
P.
Anil
Kum
ar,
B.
Anuradha
.
R
efl
e
ct
iv
ity
Para
m
et
er
Ext
ra
ction
From
RADAR
Im
age
s
Us
ing
B
ac
k
Propaga
t
ion
Algorit
hm
s
”
.
Int
ernati
onal
Journal
of
El
e
ct
rica
l
and
Computer
Engi
ne
ering
(
IJE
CE)
,
vol
8
,
N
o.
5,
Oct
.
,
p
art
1
,
2018.
[5]
Pin
-
Heng
Chen, Mahe
sh
C.
Shastr
y
,
Chi
eh
-
Ping Lai,
and
Ram
M.
Nara
y
ana
n
.
“
A
P
orta
bl
e
Re
al
-
Tim
e
Digit
a
l
Nois
e
Rada
r
S
y
st
em
for
Through
-
the
-
W
al
l
Im
agi
ng
”
.
IEE
E
Tr
ansact
ions
on
Geosci
e
nce
and
R
emote
Sensing
,
2012;
50(10).
[6]
Dus
ti
n
P.
Fair
c
hil
d
and
Ram
M.
Nara
y
a
nan
.
“
Micro
-
Doppler
Rada
r
Cla
ss
ifica
t
ion
of
Hum
a
n
Motions
unde
r
Vari
ous T
ra
ini
n
g
Scena
r
ios
”
.
SP
IE
Proc
ee
d
ings
Vol.
8734
,
2013
.
[7]
Xiaoxi
ang
L
iu,
Henr
y
L
e
ung,
an
d
George
A.
Lam
propoulos.
“
Eff
ec
ts of
Non
-
Uniform
Motion
in
Through
-
the
-
W
a
ll
SA
R
Im
agi
ng
”
.
I
EE
E
Tr
ansacti
o
ns on
Antennas
and
Propagati
on
.
2009;
57(
11)
.
[8]
Ram
M.
Nara
y
a
nan,
Sonn
y
Sm
it
h,
and
K
y
le
A.
Gall
agh
er.
“
A
Multi
fr
eque
nc
y
R
a
dar
S
y
st
em
for
Dete
c
ti
ng
Hum
ans
and
Chara
c
te
ri
z
ing
Hum
an
Ac
ti
vities
for
Short
-
Range
Throu
gh
-
W
al
l
and
L
ong
-
Range
Foli
age
Penet
r
at
ion
Applic
a
ti
ons
”
.
Hindawi
Publis
hing
Corpora
t
io
n,
Int
ernati
onal
Journal
of
Mi
crowave
S
ci
en
c
e
and
Techno
lo
gy
,
Volum
e
2014,
2
1
page
s,
2014
.
[9]
Jos
e
M.
Garc
ia
-
Rubia
,
Ozl
em
K
il
ic,
Vinh
Dang,
Quang
Nguy
en
,
and
Nghia
Tran.
“
Anal
y
sis
of
Moving
Hum
an
Micro
-
Doppler
Signat
ure
in
For
est
Env
ironment
s
”
.
Progress
In Electromagnetic
s R
ese
arch
,
Vol
.
148,
1{14,
2014
.
[10]
Matt
hew
Rit
ch
ie,
Matt
hew
As
h,
Qingcha
o
Chen
and
Kevin
Chet
t
y
.
“
Through
W
al
l
Rada
r
Cla
ss
ifica
t
ion
of
Hum
a
n
Micro
-
Doppler
Us
ing
Singula
r V
al
ue
De
compos
it
ion
Ana
l
y
s
is
”
.
Sensors
2016,
1
6,
1401
.
[11]
Pree
thi
K.
Mane
and
K.
Nara
sim
ha
Rao.
“
Granul
ar
Mobili
t
y
-
Fa
ctor
Anal
y
sis
Fra
m
ework
for
enr
ic
hing
Occ
up
an
c
y
Se
nsing
with
Do
pple
r
Rada
r
”
.
In
te
rnational
Jour
nal
of
El
e
ct
rica
l
and
Computer
Engi
ne
ering
(
IJE
CE)
,
Vol.
8,
No.
2,
pp
.
979~988,
April
2018.
[12]
Carmine
Cle
m
e
nte
,
Al
essio
Ba
ll
eri,
Kar
l
W
oodbridge
and
Jo
hn
J
Soragha
n.
“
Deve
lopments
in
ta
rg
et
m
ic
r
o
-
Doppler
signat
u
re
s
anal
y
s
is:
ra
dar
imaging
,
ul
tra
sound
and
t
hrough
-
the
-
wal
l
rad
ar
”
.
EURASIP
Journal
o
n
Adv
anc
es
in
Sig
nal
Proc
essing
2
013,
2013:47
[13]
Yinan
Yang,
Jia
j
in
L
ei
,
W
enxue
Zha
ng
and
Chao
Lu.
“
Ta
rge
t
Cl
a
ss
ifi
ca
ti
on
and
Pattern
Re
cogni
t
i
on
Us
ing
Micro
-
Doppler
Rada
r
Sig
nat
ure
s
”
.
IE
EE
Sevent
h
AC
IS
Inte
rnational
Confe
renc
e
on
Soft
ware
Engi
nee
ring,
Artifi
cial
Inte
lligen
ce,
Ne
t
working,
and
Pa
ralle
l/
Distribu
ted Computi
ng
,
20
06.
[14]
Carmine
Cle
m
e
nte
,
Al
essio
Balleri,
Karl
W
oodbridge
3
and
Joh
n
J
Soragha
n.
“
Deve
lopments
i
n
ta
rge
t
m
i
cro
-
Doppler
signat
u
res
anal
y
s
is:
ra
dar
imaging
,
ul
tra
sound
and
t
hrough
-
the
-
wal
l
rad
ar
”
.
EURASIP
Journal
o
n
Adv
anc
es
in
Sig
nal
Proc
essing
,
2013:47,
2013
.
[15]
Aw
s
Al
-
Qaisi,
Sale
h
A.
Khawa
tr
eh,
Ahm
ad
A.
Shara
dqah
,
Zi
ad
A.
Alqadi.
“
W
ave
File
Fea
ture
s
Ext
ra
ct
ion
Us
ing
Reduc
ed
LBP
”
.
Inte
rnational
Jo
urnal
of
El
e
ct
ri
c
al
and
Compute
r
Engi
nee
ring
(
IJE
CE)
,
vol
8
,
No.
5,
Oc
t.,
par
t1
,
2018.
[16]
Hus
sein
Ali
Aldel
f
y
,
Mahm
oo
d
Ham
za
Al
-
Mufraj
i
,
Tha
m
ir
R.
Saee
d
.
“
Im
prove
d
Ke
y
Fram
e
Ext
r
ac
t
ion
Us
ing
Discre
te
W
avele
t
Tra
nsform
with
Modifie
d
Thre
shold
Fact
or
”
.
TEL
KOMNIKA
(
Tele
communic
ati
on,
Computin
g
,
El
e
ct
ronivs
and
Control)
,
Vol.
16
,
No.2
,
pp
.
567~
572,
April
2018.
[17]
Ham
ee
d
R.
Farh
an,
Mahm
uod
H
.
Al
-
Muifra
je
an
d
Tha
m
ir
R
.
Sae
ed.
“
A
Novel
F
a
ce
R
ec
ogni
ti
on
Method
base
d
o
n
One
State
of
Disc
rete
Hidden
Markov
Model
”
.
IE
EE
Annual
Confe
ren
ce
on
New
Tr
ends
i
n
Information
&
Comm
unic
ati
ons Tec
hnolog
y App
li
cations
,
7
-
9
Marc
h
2017
[18]
Ham
ee
d
R.
Farh
an,
Mahm
uod
H.
Al
-
Muifra
j
e
an
d
Tha
m
ir
R.
Sae
ed.
“
Us
ing
onl
y
Two
Stat
es
of
Discre
t
e
HM
M
fo
r
High
-
Speed
Fa
ce
R
ec
ogni
ti
on
”
.
I
EE
E
Al
-
Sa
deq
Int
ernati
on
al
Confe
ren
ce
on
Mult
id
isci
pl
i
nary
in
IT
an
d
Comm
unic
ati
on
Sci
en
ce and
App
li
cations
,
IRAQ
(9
-
10)
Ma
y
,
20
16
[19]
X
Li
u,
H
Le
un
g,
G
La
m
propoulos.
“
Eff
ec
ts
o
f
non
-
uniform
m
oti
on
in
through
-
the
-
wal
l
SA
R
imaging
”
.
IE
EE
Tr
an
s.
Ant
ennas Propagation
.
57
(11),
3539
–
3548
(2009).
[20]
S
Ram,
C
Chri
stia
nson,
Y
Kim
,
H
Li
ng.
“
Si
m
ula
ti
on
and
a
naly
s
is
of
hum
an
Micro
-
Dopple
r
in
through
-
wall
envi
ronm
ent
s
”
.
I
EE
E
Tr
ans.
Geo
sci.
Re
m.S
ens
.
4
8(4),
2015
–
2023
(2010).
[21]
PH
Chen,
MC
S
hastr
y
,
CP
La
i
,
RM
Nara
y
an
an.
“
A
porta
ble
real
-
ti
m
e
digi
t
al
noi
se
rad
ar
s
y
st
em
for
through
-
the
-
wall
imaging
”
.
I
EE
E
Tr
ans.
Geo
sci.
Re
m.
Sens
.
P
P(99),
1
–
12
(201
2).
[22]
Fangm
in
Li
,
Ch
ao
Yang,
Yuqin
g
Xia,
Xiaolin
Ma,
T
ao
Zh
ang
and
Zhou
Zhou.
“
An
Adap
ti
ve
S
-
Method
to
An
aly
z
e
Micro
-
Doppler
Signal
s for
Hum
an
Act
ivi
t
y
Cl
assific
a
ti
on
”
.
S
ensors
2017,
17,
27
69.
[23]
Dus
ti
n
P.
Fair
ch
il
d,
Ram
M.
Nar
a
y
ana
n
,
Eri
c
R.
Bec
ke
l,
W
a
i
K.
Luk,
and
Ger
ard
A.
Ga
et
a
.
“
Cha
pte
r
5
:
Through
-
the
-
W
al
l
Mi
cro
-
Doppler
Signat
u
res.
IE
T
Ra
da
r
”
,
Sonar
and
Navigation
Seri
es
34,
Radar
Mic
ro
-
Doppler
Signat
ur
e
Proce
ss
ing
and
appli
cations
,
Ed
it
ed
b
y
Vi
ct
or
C
.
Chen,
Dav
id
Tahm
oush,
and
W
il
li
am
J.
Mic
eli,
Printe
d
in
the
U
K
b
y
CP
I
Group (
UK
)
Lt
d,
Cro
y
d
on,
IE
T
2014
.
[24]
R.
W
il
son.
“
Pr
opaga
t
ion
losses
thr
ough
common
buil
ding
m
at
eri
a
ls:
2.
4
GH
z
vs.
5
GH
z
”
.
Te
chnica
l
Repor
t,
Univer
sit
y
of
Southern
C
al
iforn
i
a,
Augus
t
2002.
Evaluation Warning : The document was created with Spire.PDF for Python.