I
n
t
e
r
n
at
ion
al
Jou
r
n
al
of
E
lec
t
r
ical
an
d
Com
p
u
t
e
r
E
n
gin
e
e
r
in
g
(
I
JE
CE
)
Vol.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
,
pp.
467
~
479
I
S
S
N:
2088
-
8708
,
DO
I
:
10
.
11591/i
jec
e
.
v
15
i
1
.
pp
4
67
-
479
467
Jou
r
n
al
h
omepage
:
ht
tp:
//
ij
e
c
e
.
iaes
c
or
e
.
c
om
E
n
h
an
c
in
g m
il
li
m
e
t
e
r
-
w
av
e
c
om
m
u
n
ic
at
io
n
:
a t
r
op
i
c
al
p
e
r
s
p
e
c
t
iv
e
o
n
r
ai
n
d
r
op
si
z
e
d
is
t
r
ib
u
t
io
n
a
n
d
si
g
n
al
at
t
e
n
u
at
io
n
Nurul
Naj
wa
M
d
Yus
of
1
,
2
,
Ja
f
r
i
Din
1
,
L
am
Ho
n
g
Yin
3
1
D
e
pa
r
tm
e
nt
of
C
omm
uni
c
a
ti
on E
ngi
ne
e
r
in
g, F
a
c
ul
ty
of
E
le
c
tr
i
c
a
l
E
ngi
ne
e
r
in
g, U
ni
ve
r
s
it
i
T
e
knol
ogi
M
a
la
y
s
ia
,
J
ohor
B
a
hr
u
,
M
a
la
ys
ia
2
D
e
pa
r
tm
e
nt
of
P
ol
yt
e
c
hni
c
a
nd C
omm
uni
ty
C
ol
le
ge
E
duc
a
ti
on
, M
in
is
tr
y of
H
ig
he
r
E
duc
a
ti
on,
P
ut
r
a
ja
ya
,
M
a
la
ys
ia
3
D
e
pa
r
tm
e
nt
of
E
le
c
tr
ic
a
l
E
ngi
ne
e
r
in
g, F
a
c
ul
ty
of
E
ngi
ne
e
r
in
g
T
e
c
hnol
ogy, Unive
r
s
it
i
T
un H
us
s
e
in
O
nn M
a
l
a
ys
ia
,
P
a
r
it
R
a
ja
,
M
a
la
ys
ia
Ar
t
icle
I
n
f
o
AB
S
T
RA
CT
A
r
ti
c
le
h
is
tor
y
:
R
e
c
e
ived
Apr
18,
2024
R
e
vis
e
d
S
e
p
17,
2024
Ac
c
e
pted
Oc
t
1,
2024
T
h
i
s
s
t
u
d
y
t
ac
k
l
e
s
rai
n
at
t
e
n
u
a
t
i
o
n
i
n
mi
l
l
i
met
e
r
-
w
av
e
(mm
-
w
a
v
e)
co
mmu
n
i
ca
t
i
o
n
,
a
cri
t
i
ca
l
co
n
cer
n
fo
r
t
h
e
ad
v
an
c
em
en
t
o
f
5
G
w
i
rel
es
s
t
ech
n
o
l
o
g
y
.
It
e
x
ami
n
es
t
h
e
v
ar
i
ab
i
l
i
t
y
o
f
ra
i
n
d
ro
p
s
i
z
e
d
i
s
t
ri
b
u
t
i
o
n
(
D
SD
)
an
d
i
t
s
i
mp
ac
t
o
n
s
p
e
ci
f
i
c
at
t
en
u
at
i
o
n
,
w
i
t
h
a
fo
cu
s
o
n
t
r
o
p
i
cal
en
v
i
ro
n
men
t
s
s
u
c
h
as
Ma
l
ay
s
i
a.
U
s
i
n
g
t
h
e
J
o
s
s
-
W
a
l
d
v
o
g
el
d
i
s
d
ro
met
er
,
t
h
i
s
s
t
u
d
y
co
l
l
ect
e
d
an
d
an
al
y
zed
ex
t
en
s
i
v
e
D
SD
d
at
a
o
v
er
t
h
ree
y
ears
,
rev
eal
i
n
g
t
h
a
t
t
h
e
h
i
g
h
e
s
t
D
S
D
co
n
cen
t
ra
t
i
o
n
s
d
o
n
o
t
n
ece
s
s
a
ri
l
y
r
es
u
l
t
i
n
t
h
e
g
re
at
e
s
t
s
p
ec
i
fi
c
a
t
t
e
n
u
a
t
i
o
n
.
T
h
i
s
s
t
u
d
y
ad
o
p
t
ed
a
mach
i
n
e
l
earn
i
n
g
ap
p
ro
ac
h
,
s
p
ec
i
fi
ca
l
l
y
s
u
p
erv
i
s
e
d
l
ear
n
i
n
g
w
i
t
h
l
i
n
ear
reg
re
s
s
i
o
n
,
t
o
e
n
h
a
n
ce
t
h
e
accu
racy
o
f
a
t
t
e
n
u
a
t
i
o
n
p
r
ed
i
ct
i
o
n
mo
d
e
l
s
.
A
n
ew
s
e
t
o
f
co
effi
ci
e
n
t
s
fo
r
t
h
e
p
o
w
er
-
l
a
w
mo
d
e
l
o
f
s
p
ec
i
fi
c
at
t
en
u
a
t
i
o
n
w
a
s
d
eri
v
e
d
an
d
b
e
n
ch
mar
k
ed
ag
ai
n
s
t
t
h
e
IT
U
-
R
P.
8
3
8
-
3
s
t
a
n
d
ar
d
an
d
s
i
mi
l
ar
s
t
u
d
i
e
s
i
n
co
m
p
arab
l
e
cl
i
ma
t
es
.
T
h
e
fi
n
d
i
n
g
s
emp
h
as
i
ze
t
h
e
i
m
p
o
r
t
an
ce
o
f
d
ev
el
o
p
i
n
g
reg
i
o
n
-
s
p
ec
i
fi
c
mo
d
e
l
s
t
h
at
c
o
n
s
i
d
er
l
o
cal
me
t
eo
r
o
l
o
g
i
cal
v
ari
at
i
o
n
s
,
p
o
t
e
n
t
i
al
l
y
o
fferi
n
g
s
i
g
n
i
fi
ca
n
t
i
m
p
ro
v
emen
t
s
t
o
t
h
e
rel
i
ab
i
l
i
t
y
a
n
d
d
es
i
g
n
o
f
mm
-
w
a
v
e
co
mmu
n
i
ca
t
i
o
n
s
y
s
t
ems
i
n
t
h
e
fu
t
u
re
.
K
e
y
w
o
r
d
s
:
M
il
li
mete
r
-
wa
ve
R
a
indr
op
s
ize
dis
tr
ibut
ion
S
pe
c
if
ic
a
tt
e
nua
ti
on
S
upe
r
vis
e
d
mac
hine
lea
r
ning
T
r
opica
l
mons
oons
Th
i
s
i
s
a
n
o
p
en
a
c
ces
s
a
r
t
i
c
l
e
u
n
d
e
r
t
h
e
CC
B
Y
-
SA
l
i
ce
n
s
e.
C
or
r
e
s
pon
din
g
A
u
th
or
:
J
a
f
r
i
Din
De
pa
r
tm
e
nt
of
C
omm
unica
ti
on
E
nginee
r
ing
,
F
a
c
ul
ty
of
E
lec
tr
ica
l
E
nginee
r
ing
,
Unive
r
s
it
i
T
e
knologi
M
a
lays
ia
81310
UT
M
J
ohor
B
a
hr
u
,
J
ohor
Da
r
ul
T
a
’
z
im
,
M
a
lays
ia
E
mail:
jaf
r
i@ut
m
.
my
1.
I
NT
RODU
C
T
I
ON
T
he
de
mand
f
or
mor
e
c
omm
unica
ti
on
c
ha
nne
ls
th
a
t
c
a
n
a
c
c
omm
oda
te
f
a
s
ter
da
ta
r
a
tes
a
nd
b
r
oa
de
r
ba
ndwidths
ha
s
led
to
the
r
e
s
e
a
r
c
h
of
the
mi
ll
im
e
ter
-
wa
ve
(
mm
-
wa
v
e
)
s
pe
c
tr
um,
whic
h
typi
c
a
ll
y
ope
r
a
tes
a
bove
20
GH
z
,
with
wa
ve
lengths
les
s
than
10
mm
[
1]
.
How
e
ve
r
,
the
p
r
omi
s
e
of
unpr
e
c
e
de
nted
da
ta
tr
a
ns
mi
s
s
ion
s
pe
e
ds
a
nd
ne
twor
k
r
e
li
a
bil
it
y
is
tempe
r
e
d
by
the
phys
ica
l
li
mi
tations
im
pos
e
d
by
the
a
tm
os
phe
r
e
,
pa
r
ti
c
ular
ly
thr
ough
r
a
in
a
tt
e
nua
ti
on.
R
a
in
a
tt
e
nua
ti
on
plays
a
c
r
uc
ial
r
o
le
in
the
de
ter
io
r
a
ti
on
of
mm
-
wa
ve
s
ignals
[
2]
,
[
3]
by
c
a
us
ing
de
e
p
f
a
de
s
t
ha
t
im
pa
c
t
the
a
va
il
a
bil
it
y
of
the
l
ink
[
4]
,
thus
be
c
omi
ng
a
f
or
mi
da
ble
c
ha
ll
e
nge
to
maintaining
the
int
e
gr
it
y
of
high
-
f
r
e
que
nc
y
s
ignal
tr
a
ns
mi
s
s
ion
s
ove
r
c
on
s
ider
a
ble
dis
tanc
e
s
.
R
a
in
a
tt
e
nua
ti
on
de
pe
nds
he
a
vil
y
on
lo
c
a
l
c
li
matology
[
5
]
.
pr
e
vious
wo
r
ks
ha
ve
highl
ig
hted
the
s
igni
f
ica
nt
im
pa
c
t
of
r
a
indr
op
s
ize
dis
tr
ibut
ion
(
D
S
D)
on
r
a
in
a
tt
e
nua
ti
on.
F
or
e
xa
mpl
e
,
s
e
ve
r
a
l
s
tudi
e
s
ha
ve
e
mphas
ize
d
that
the
e
mpi
r
ica
l
model
pr
ovided
by
I
T
U
-
R
doe
s
not
ha
ve
loca
l
r
a
in
c
ha
r
a
c
ter
is
ti
c
s
[
6]
,
[
7]
.
T
he
I
T
U
-
R
model
may
be
e
f
f
e
c
ti
ve
f
or
lowe
r
f
r
e
que
n
c
ies
,
but
a
s
the
f
r
e
que
nc
y
moves
int
o
the
mm
-
wa
ve
r
a
nge
,
the
I
T
U
-
R
model
be
gins
to
los
e
a
c
c
ur
a
c
y
[
8]
.
T
h
is
li
mi
tation
wa
s
a
c
knowle
dge
d
in
[
9]
that
s
ugge
s
ted
the
I
T
U
-
R
model's
inade
qua
c
y
is
due
to
not
a
c
c
ounti
ng
f
or
loca
l
mete
or
o
logi
c
a
l
va
r
iations
,
ne
c
e
s
s
it
a
ti
ng
a
mor
e
de
tailed
unde
r
s
tanding
of
DSD
[
8]
.
T
he
c
oe
f
f
icie
nt
of
the
powe
r
-
law
in
s
pe
c
if
ic
a
tt
e
nua
ti
on
is
s
igni
f
ica
ntl
y
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
:
467
-
479
468
inf
luenc
e
d
by
the
loca
l
r
a
indr
op
s
ize
dis
tr
ibut
ion
(
DSD)
,
whic
h
c
a
n
di
f
f
e
r
a
c
r
os
s
va
r
ious
r
e
gions
[
10]
.
T
his
im
por
tanc
e
wa
s
c
or
r
obor
a
ted
in
[
11]
whic
h
s
tate
d
that
s
pe
c
if
ic
a
tt
e
nua
ti
on
p
r
ogr
e
s
s
ively
r
e
li
e
s
on
the
DSD
a
t
higher
f
r
e
que
nc
ies
.
M
os
t
s
tudi
e
s
e
xa
mi
ning
r
a
in
a
tt
e
nua
ti
on,
including
thos
e
c
onduc
ted
by
I
T
U
-
R
,
ha
ve
pr
im
a
r
il
y
c
onc
e
ntr
a
ted
on
tempe
r
a
te
r
e
gions
[
12
]
.
How
e
ve
r
,
the
c
ha
r
a
c
ter
is
ti
c
s
a
nd
s
tr
uc
tu
r
e
s
of
r
a
inf
a
ll
e
ve
nts
in
tr
opica
l
a
r
e
a
s
dif
f
e
r
s
igni
f
ica
ntl
y,
lea
ding
to
inac
c
ur
a
te
p
r
e
dictions
f
or
tr
opica
l
c
li
mate
s
[
13]
.
T
he
inf
luenc
e
of
DSD
on
mm
-
wa
ve
pr
opa
ga
ti
on
is
c
r
uc
ial
in
tr
opica
l
r
e
gions
due
to
the
r
a
in’
s
unique
c
ha
r
a
c
ter
is
ti
c
s
,
s
uc
h
a
s
high
r
a
inf
a
ll
r
a
tes
a
nd
int
e
ns
e
pr
e
c
ipi
tation
e
ve
nts
with
lar
ge
r
r
a
ind
r
op
s
,
whic
h
c
a
us
e
gr
e
a
ter
s
c
a
tt
e
r
ing
a
nd
a
bs
or
pti
on
of
e
lec
tr
o
magne
ti
c
wa
ve
s
,
r
e
s
ult
ing
in
higher
a
tt
e
nua
ti
on
[
14]
.
T
his
s
igni
f
ica
ntl
y
de
gr
a
de
s
the
pe
r
f
or
manc
e
of
c
omm
unica
ti
on
s
ys
tems
,
pa
r
ti
c
ula
r
ly
5G
a
nd
6G
ne
two
r
ks
[
15
]
.
B
e
c
a
us
e
I
T
U
-
R
f
r
e
que
ntl
y
unde
r
e
s
ti
mate
s
s
pe
c
if
ic
a
tt
e
nua
ti
on,
s
ome
s
tudi
e
s
ha
ve
highl
ight
e
d
the
im
por
tanc
e
of
r
e
gion
-
s
pe
c
if
ic
a
djus
tm
e
nts
a
nd
mor
e
c
ompl
e
x
modeling
tec
hniques
[
16]
,
[
17]
.
F
or
e
xa
mpl
e
,
the
va
r
iabili
ty
o
f
DSD
in
I
ndia
r
e
ve
a
led
that
the
s
pe
c
if
ic
r
a
in
a
tt
e
nua
ti
on
a
t
100
GH
z
de
c
r
e
a
s
e
s
les
s
with
incr
e
a
s
ing
r
a
in
r
a
tes
than
a
t
60
GH
z
[
18]
.
Anothe
r
s
igni
f
ica
nt
wor
k
by
Alonge
[
19]
a
na
lyze
d
the
e
f
f
e
c
ts
of
r
a
inf
a
ll
a
tt
e
nua
ti
on
a
c
r
os
s
dif
f
e
r
e
nt
r
e
gions
in
A
f
r
i
c
a
,
indi
c
a
ti
ng
the
im
pa
c
t
o
f
c
li
matic
a
nd
ge
ogr
a
phi
c
f
a
c
tor
s
on
DSD
pa
tt
e
r
ns
.
T
he
s
e
s
tudi
e
s
e
mphas
iz
e
d
the
im
por
tanc
e
of
c
ons
ider
ing
r
e
gional
DSD
va
r
i
a
ti
ons
in
a
tt
e
nua
ti
on
modeling
a
nd
im
pr
oving
the
a
c
c
ur
a
c
y
of
r
a
in
f
a
ll
a
tt
e
nua
ti
on
f
or
e
c
a
s
ts
,
ther
e
by
c
ontr
ibut
i
ng
to
the
de
ve
lopm
e
nt
a
nd
opti
mi
z
a
ti
on
of
r
obus
t
c
omm
u
nica
ti
ons
s
ys
tem
s
that
c
a
n
withs
tand
the
s
e
ve
r
e
we
a
ther
c
ondit
ions
typi
c
a
l
of
thes
e
a
r
e
a
s
[
9]
.
T
his
r
e
s
e
a
r
c
h
a
ddr
e
s
s
e
s
the
de
f
icie
nc
ie
s
in
a
c
c
ur
a
tely
pr
e
dicting
r
a
in
a
tt
e
nua
ti
on
due
to
the
ove
r
s
ight
of
r
e
gional
mete
or
ologi
c
a
l
f
luctua
ti
ons
by
e
xis
ti
ng
models
in
tr
opica
l
c
ountr
ies
s
uc
h
a
s
M
a
lays
i
a
.
T
his
is
done
by
a
mas
s
ing
a
nd
e
xt
e
ns
ively
s
c
r
uti
nizing
thr
e
e
ye
a
r
s
of
DSD
da
ta
uti
li
z
ing
the
J
os
s
-
W
a
ldvogel
dis
dr
omete
r
.
B
y
de
ve
lopi
ng
ne
w
c
oe
f
f
icie
nts
f
or
a
powe
r
-
law
model
f
o
r
s
pe
c
if
ic
a
tt
e
nua
ti
on
us
ing
s
upe
r
vis
e
d
mac
hine
lea
r
ning
a
nd
c
ompar
ing
them
with
the
I
T
U
-
R
P
.
838
-
3
s
tanda
r
d,
thi
s
s
tudy
s
tr
ives
to
r
e
f
ine
the
a
tt
e
nua
ti
on
pr
e
diction
model
,
c
a
ptur
ing
s
e
a
s
ona
l
a
nd
we
a
ther
-
r
e
late
d
va
r
iations
.
Additi
ona
ll
y
,
i
t
inv
e
s
ti
ga
tes
the
im
pa
c
t
of
dif
f
e
r
e
nt
r
a
in
types
a
nd
int
e
ns
it
ies
on
s
pe
c
if
ic
a
tt
e
nua
ti
on
a
c
r
os
s
va
r
ious
mm
-
wa
ve
f
r
e
q
ue
nc
ies
.
T
his
pa
pe
r
is
or
ga
nize
d
a
s
f
oll
ows
:
s
e
c
ti
on
2
de
s
c
r
ibes
the
ke
y
f
e
a
tur
e
s
o
f
the
J
os
s
-
W
a
ldvogel
R
D
-
69
dis
dr
omete
r
,
including
it
s
ope
r
a
ti
ona
l
pr
inciple
,
s
e
ns
ing
a
r
e
a
,
a
nd
c
a
li
b
r
a
ti
on
methods
.
S
e
c
ti
on
3
e
xplains
the
c
a
lcula
ti
on
of
s
pe
c
if
ic
a
tt
e
nua
ti
on
(
γ
)
us
ing
th
e
I
T
U
-
R
P
.
838
-
3
s
tanda
r
d
a
nd
the
T
-
matr
ix
metho
d,
whic
h
c
ons
ider
s
c
ompl
e
x
s
c
a
tt
e
r
ing
pa
tt
e
r
ns
o
f
r
a
indr
ops
,
a
s
we
ll
a
s
the
s
pe
c
if
ic
a
tt
e
nua
ti
on
r
e
s
ult
s
f
or
f
r
e
que
nc
ies
in
the
r
a
nge
18
to
38
GH
z
,
highl
ight
ing
the
va
r
ia
ti
ons
be
twe
e
n
dif
f
e
r
e
nt
r
a
in
types
a
nd
c
ompar
ing
the
ne
w
powe
r
-
law
model
c
oe
f
f
icie
nts
with
the
I
T
U
-
R
s
tanda
r
d.
T
his
s
e
c
ti
on
a
ls
o
pr
ovides
c
ompl
e
menta
r
y
c
umul
a
ti
ve
dis
tr
ibut
ion
f
unc
ti
on
(
C
C
DF)
f
o
r
dif
f
e
r
e
nt
r
a
in
e
ve
nts
,
a
c
c
e
nti
ng
the
dif
f
e
r
e
nc
e
s
obs
e
r
ve
d
dur
ing
the
s
outhwe
s
t
(
S
W
)
a
nd
nor
thea
s
t
(
N
E
)
mons
oon
s
.
F
inally,
S
e
c
ti
on
4
c
onc
ludes
the
s
tudy
by
s
um
mar
izing
the
r
e
s
ult
s
a
nd
s
ugge
s
ti
ng
pos
s
ibl
e
im
pr
ove
ments
f
or
mm
-
wa
ve
c
omm
unica
ti
on
s
ys
tems
in
tr
opica
l
r
e
gions
.
2.
M
E
T
HO
D
I
n
thi
s
s
tudy,
D
S
D
wa
s
mea
s
ur
e
d
us
ing
J
os
s
W
a
ldvogel
dis
dr
omete
r
,
pa
r
ti
c
ula
r
ly
the
R
D
-
69
model,
whic
h
wa
s
loca
ted
a
t
Unive
r
s
it
i
T
e
knologi
M
a
l
a
ys
ia
in
Kua
la
L
umpur
(
UT
M
,
KL
)
.
I
t
ope
r
a
te
s
on
the
pr
inciple
o
f
de
tec
ti
ng
dr
op
lets
pa
s
s
ing
thr
ough
a
s
e
ns
ing
a
r
e
a
a
nd
s
or
ti
ng
them
int
o
20
di
f
f
e
r
e
nt
c
ha
nne
ls
ba
s
e
d
on
their
s
ize
a
nd
f
a
ll
ve
locity.
T
he
R
D
-
69
d
is
dr
omete
r
c
ons
is
ts
of
a
c
onica
l
S
tyr
of
oa
m
c
oll
e
c
t
or
that
is
pos
it
ioned
a
bove
a
s
e
n
s
ing
unit
.
T
he
c
oll
e
c
tor
is
i
ntende
d
to
mi
nim
ize
the
e
f
f
e
c
ts
of
wind
a
nd
to
e
n
s
ur
e
that
r
a
indr
ops
f
a
ll
ve
r
ti
c
a
ll
y
onto
the
s
e
ns
or
.
M
e
a
nwhile,
the
s
e
ns
ing
unit
c
ompr
is
e
s
a
c
oil
of
wir
e
that
is
s
us
pe
nde
d
in
a
magne
ti
c
f
ield
[
20]
.
W
he
n
a
r
a
indr
op
s
tr
ikes
the
c
oll
e
c
tor
,
it
e
xe
r
ts
f
o
r
c
e
a
nd
pus
he
s
the
c
oil
.
T
his
moveme
nt
ge
ne
r
a
tes
a
vol
tage
in
the
c
oil
,
wh
ich
is
a
mpl
i
f
ied
a
nd
r
e
c
or
de
d
by
a
da
ta
logger
.
T
h
e
DSD
is
de
r
ived
f
r
om
the
vol
tage
puls
e
s
pr
oduc
e
d
by
the
R
D
-
69
dis
dr
omete
r
.
T
he
puls
e
a
mpl
it
ude
c
or
r
e
s
ponds
to
the
s
ize
of
the
c
a
us
ing
r
a
indr
op.
T
o
c
a
lcula
te
the
DSD,
the
c
ount
of
puls
e
s
f
or
e
a
c
h
s
ize
c
a
tegor
y
is
di
vided
by
the
s
a
mpl
ing
pe
r
iod
,
yielding
the
D
S
D
in
unit
s
of
mm
/h
.
T
he
mea
s
ur
e
ment
s
e
tup
f
or
the
dis
dr
o
mete
r
is
s
hown
in
F
igur
e
1
.
T
he
de
tails
o
f
the
mea
s
ur
e
ment
s
e
tup
c
a
n
be
f
ound
in
[
21]
.
T
he
s
a
mpl
ing
a
r
e
a
f
o
r
the
R
D
-
69
dis
dr
omete
r
is
=
5000
mm
2
wi
th
a
one
-
mi
nute
int
e
gr
a
ti
on
ti
me
of
=
60
s
a
nd
diame
ter
c
las
s
e
s
r
a
nging
be
twe
e
n
0.
3
a
nd
5.
3
mm
.
T
he
r
a
in
r
a
te
(
mm
/h)
given
by
th
e
dis
dr
omete
r
c
a
n
be
c
a
lcula
ted
us
ing
(
1
)
:
=
3600
6
∑
3
20
=
1
(
1)
whe
r
e
is
the
diame
ter
of
the
th
bin
with
the
numbe
r
of
r
a
indr
ops
,
.
T
he
mea
s
ur
e
d
DSD
in
(
m
-
3
mm
-
1
)
c
a
n
be
e
xpr
e
s
s
e
d
by
(
2
)
whe
r
e
(
)
is
the
Gunn
-
Kinz
e
r
ve
locity
of
r
a
indr
ops
in
s
ti
ll
a
ir
[
22]
,
[
23
]
a
nd
∆
is
the
s
pa
n
of
c
las
s
th
bin
de
ter
m
ined
by
the
dis
dr
omete
r
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
E
nhanc
ing
mil
li
me
ter
-
w
av
e
c
omm
unication:
a
tr
o
pical
pe
r
s
pe
c
ti
v
e
on
…
(
N
ur
ul
N
ajw
a
M
d
Y
us
of
)
469
(
)
=
(
)
×
×
×
∆
(
2)
F
igur
e
1.
J
os
s
-
W
a
ldvog
e
l
R
D
-
69
dis
dr
omete
r
mea
s
ur
e
ment
s
e
tup
[
24]
T
his
s
tudy
a
na
lyze
s
36
mont
hs
of
DSD
da
ta
f
r
om
the
R
D
-
69
dis
dr
omete
r
,
whic
h
pr
ovided
r
a
w
da
tas
e
ts
e
nc
ompas
s
ing
61,
384
r
a
iny
mi
nutes
(
1
-
mi
nute
int
e
gr
a
ti
on
ti
me)
.
F
r
om
thes
e
da
ta,
556
r
a
in
e
ve
nts
we
r
e
identif
ied,
with
e
a
c
h
e
ve
nt
s
e
pa
r
a
ted
by
dr
y
p
e
r
iods
of
a
mi
ni
mum
dur
a
ti
on
of
one
hour
,
s
e
pa
r
a
ti
ng
e
a
c
h
e
ve
nt
f
r
om
the
s
ubs
e
que
nt
e
ve
nt.
T
he
e
ve
nts
we
r
e
c
las
s
if
ied
ba
s
e
d
on
their
int
e
ns
it
y
a
nd
dur
a
ti
on
in
to
thr
e
e
c
a
tegor
ies
:
dr
izz
le,
s
howe
r
,
a
nd
thunder
s
tor
m
,
ba
s
e
d
on
p
r
e
vious
s
tudi
e
s
[
25
]
,
[
26
]
.
Dr
izz
le
oc
c
ur
r
e
nc
e
s
a
r
e
c
ha
r
a
c
ter
ize
d
by
r
a
in
r
a
tes
be
low
20
mm
/h
that
las
t
f
or
a
mi
nim
um
of
10
c
ons
e
c
uti
ve
mi
nutes
,
while
s
howe
r
s
e
xhibi
t
r
a
in
r
a
tes
r
a
nging
f
r
om
20
to
70
mm
/h
s
us
taine
d
f
or
10
c
ons
e
c
uti
ve
mi
nutes
.
T
hunde
r
s
tor
m
e
ve
nts
a
r
e
identif
ied
by
r
a
in
r
a
tes
e
xc
e
e
ding
70
mm
/h
s
us
taine
d
f
or
a
t
lea
s
t
5
c
ons
e
c
uti
ve
mi
nutes
.
E
ve
nts
with
r
a
in
r
a
tes
be
low
5
mm
/h
a
t
the
s
tar
t
a
nd
e
nd
we
r
e
not
incl
ude
d.
T
he
b
in
nu
mber
with
it
s
c
o
r
r
e
s
ponding
me
a
n
dr
op
diame
ter
is
p
r
e
s
e
nted
in
T
a
ble
1.
T
he
s
umm
a
r
y
of
a
ve
r
a
ge
mea
s
ur
e
d
dr
op
c
ounts
f
o
r
S
W
a
nd
N
E
m
ons
oons
a
t
3:00
-
6:00
a
m
a
nd
3:00
-
6:00
pm
dur
ing
thunde
r
s
tor
m,
s
howe
r
a
nd
dr
izz
le
e
ve
nts
we
r
e
a
ls
o
give
n
in
the
s
a
me
table
.
T
a
ble
1.
S
umm
a
r
y
of
a
ve
r
a
ge
mea
s
ur
e
d
r
a
indr
ops
dur
ing
thunder
s
tor
m
e
ve
nts
in
the
S
W
a
nd
N
E
mo
ns
oon
B
in
N
umbe
r
M
e
a
n
d
ia
me
te
r
(
mm
)
S
W
T
hunde
r
s
to
r
m
N
E
T
hunde
r
s
to
r
m
S
W
S
how
e
r
N
E
S
how
e
r
S
W
D
r
iz
z
le
N
E
D
r
iz
z
le
AM
PM
AM
PM
AM
PM
AM
PM
AM
PM
AM
PM
1
0.36
0
0
0
0
0
0
0
0
0
0
0
0
2
0.46
1
0
1
0
2
1
0
0
3
1
0
1
3
0.55
11
6
5
6
10
8
11
11
19
5
3
6
4
0.66
39
20
21
18
37
19
65
27
23
17
12
25
5
0.77
53
39
56
39
61
34
105
44
34
29
29
40
6
0.91
151
133
224
132
174
106
249
137
77
116
102
137
7
1.12
250
223
331
228
252
181
348
249
134
202
215
256
8
1.33
196
164
215
174
186
146
229
176
107
156
200
209
9
1.51
141
124
149
128
140
106
163
131
80
109
133
148
10
1.67
131
115
129
112
121
97
139
120
83
94
118
132
11
1.91
214
177
195
163
188
148
230
187
132
136
190
199
12
2.26
156
130
139
120
156
107
178
131
106
101
156
125
13
2.58
75
67
70
63
73
57
65
66
53
55
75
63
14
2.87
48
45
40
43
45
40
42
43
33
39
40
41
15
3.20
34
40
32
38
34
34
34
31
26
33
35
31
16
3.54
14
18
19
20
15
20
13
15
13
19
12
14
17
3.92
9
12
11
14
8
14
9
11
17
15
8
10
18
4.35
2
5
5
7
2
7
4
6
8
9
5
6
19
4.86
1
2
2
3
1
5
1
4
9
5
1
4
20
5.30
0
1
0
0
0
2
0
1
2
2
1
1
2.
1.
De
t
e
r
m
in
a
t
ion
o
f
s
ou
t
h
we
s
t
an
d
n
or
t
h
e
as
t
m
on
s
oon
in
M
alays
ia
M
a
lays
ia,
a
tr
opica
l
c
ountr
y
loca
ted
a
t
4.
2105°
N,
101.
9758°
E
,
c
ompr
is
e
s
a
pe
nins
ular
a
nd
the
is
land
of
B
or
ne
o,
whic
h
include
s
S
a
ba
h
a
nd
S
a
r
a
wa
k.
I
t
e
xpe
r
ienc
e
s
two
main
mons
oon
s
e
a
s
ons
:
the
S
outhwe
s
t
(
S
W
)
mons
oon
f
r
om
late
M
a
y
to
S
e
pte
mber
a
nd
the
Nor
thea
s
t
(
NE
)
mons
oon
f
r
om
Nov
e
mber
to
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
:
467
-
479
470
M
a
r
c
h,
with
a
tr
a
ns
it
ional
int
e
r
-
mons
oon
pe
r
iod
in
be
twe
e
n.
W
ind
s
pe
e
d
a
nd
dir
e
c
ti
on
da
ta,
w
hich
a
r
e
im
por
tant
f
or
identif
ying
thes
e
mons
oons
,
we
r
e
obtaine
d
f
r
om
the
M
a
lays
ian
M
e
teor
ologi
c
a
l
De
pa
r
tm
e
nt
(
M
E
T
)
,
Vis
ua
l
C
r
os
s
ing,
a
nd
the
Na
ti
ona
l
Oc
e
a
nic
a
nd
Atmos
phe
r
ic
Admini
s
tr
a
ti
on
(
NO
AA
)
.
Dur
in
g
the
NE
mons
oon,
the
r
e
gion
e
xpe
r
ienc
e
s
the
s
tr
onge
s
t
winds
,
with
a
ve
r
a
ge
s
pe
e
ds
r
e
a
c
hing
up
to
15.
4
m/
s
[
27
]
whe
r
e
a
s
the
S
W
mons
oon
f
e
a
tur
e
s
lowe
r
wind
s
p
e
e
ds
,
a
ve
r
a
ging
a
r
ound
2
.
0
to
5
.
2
m/
s
,
whic
h
r
e
s
ult
ing
in
c
a
lm
e
r
s
e
a
c
ondit
ions
[
28]
.
De
s
pit
e
the
di
f
f
e
r
e
nc
e
in
wind
s
pe
e
ds
,
a
s
tudy
r
e
por
ted
that
in
P
e
nins
ular
M
a
lays
ia,
wind
s
pe
e
d
doe
s
not
s
igni
f
ica
ntl
y
inf
l
ue
nc
e
wind
dir
e
c
ti
on,
ir
r
e
s
pe
c
ti
ve
of
it
s
or
igi
n
[
29]
.
T
he
s
tudy
a
ls
o
r
e
ve
a
led
that
the
wind
dir
e
c
ti
on
e
xhi
bit
s
gr
e
a
ter
s
tabili
ty
in
the
e
a
s
ter
n
r
e
gion
of
P
e
nins
ular
M
a
lays
ia
c
ompar
e
d
to
the
we
s
ter
n
pa
r
t.
T
his
s
ugge
s
ts
a
tende
nc
y
f
or
wind
dir
e
c
ti
on
to
va
r
y
mo
r
e
d
ur
ing
the
s
outhwe
s
t
S
W
mons
oon
than
the
nor
thea
s
t
N
E
mo
ns
oon.
T
he
J
W
R
D
-
69
dis
dr
omete
r
wa
s
ins
talled
a
t
the
Unive
r
s
it
i
T
e
knologi
M
a
lays
ia
in
Kua
la
L
umpur
(
UT
M
,
KL
)
in
the
s
outhwe
s
t
o
f
P
e
nins
ular
M
a
lays
ia.
S
pe
c
if
ic
wind
s
pe
e
d
a
nd
di
r
e
c
ti
on
da
ta
a
r
e
not
a
va
il
a
ble
f
or
thi
s
loca
ti
on.
T
he
ne
a
r
e
s
t
we
a
ther
s
tations
a
r
e
in
P
e
taling
J
a
ya
a
nd
S
uba
ng,
a
c
c
or
ding
to
M
E
T
a
n
d
NO
AA
r
e
c
or
ds
.
M
e
teor
ologi
c
a
ll
y,
the
wind
dir
e
c
ti
on
is
de
ter
mi
ne
d
via
the
plac
e
of
o
r
igi
n
of
the
wind
[
30]
.
F
igur
e
2
s
hows
a
s
im
pli
f
ied
wind
r
os
e
diagr
a
m
to
he
lp
b
e
tt
e
r
unde
r
s
tand
wind
di
r
e
c
ti
on.
A
de
c
a
de
-
long
s
tatis
ti
c
a
l
a
na
lys
is
f
ound
that
the
wind
dir
e
c
ti
on
dur
ing
the
S
W
mons
oon
in
S
uba
ng
f
r
om
1999
to
2007
va
r
ied
be
twe
e
n
197
a
nd
231
de
gr
e
e
s
,
while
it
va
r
ied
be
twe
e
n
17
0
a
nd
211
de
gr
e
e
s
a
t
Kua
la
L
umpur
I
nter
na
ti
ona
l
Air
por
t
(
KL
I
A)
[
29
]
.
F
u
r
ther
inves
ti
ga
ti
on
r
e
ve
a
led
that
dur
ing
the
NE
mons
oon
a
t
KL
I
A
f
r
om
2014
to
2
019,
the
wind
dir
e
c
ti
on
va
r
ied
be
twe
e
n
11
a
nd
339
de
gr
e
e
s
,
pe
a
king
in
2018,
while
other
ye
a
r
s
e
xpe
r
ienc
e
d
d
ir
e
c
ti
ons
be
twe
e
n
11
a
nd
46
de
gr
e
e
s
[
31
]
.
Dur
ing
the
S
W
mons
oon,
the
di
r
e
c
ti
on
f
luctua
ted
be
twe
e
n
172
a
nd
212
de
gr
e
e
s
.
Ka
mi
s
a
n
e
t
al
.
[
32]
s
howe
d
a
c
i
r
c
ular
dis
t
r
ibut
ion
pa
tt
e
r
n
of
wind
dir
e
c
ti
on
dur
ing
the
S
W
mons
oon
a
t
f
our
loca
ti
ons
in
M
e
laka
,
S
e
na
i,
Alo
r
S
e
tar
a
nd
L
a
ngka
wi
we
r
e
be
twe
e
n
90
a
nd
315
de
gr
e
e
s
.
M
e
laka
,
whic
h
is
the
c
los
e
s
t
to
UT
M
,
KL
,
pa
r
ti
c
ular
ly
s
howe
d
wind
dir
e
c
ti
ons
be
twe
e
n
150
a
nd
285
de
gr
e
e
s
.
T
he
r
e
f
or
e
,
thi
s
s
tudy
a
s
s
umes
that
the
wind
dir
e
c
ti
ons
dur
ing
the
S
W
mons
oon
a
r
e
be
twe
e
n
180
a
nd
270
de
gr
e
e
s
a
nd
be
twe
e
n
0
a
nd
90
de
g
r
e
e
s
dur
ing
the
NE
mons
oon.
F
igur
e
2.
W
ind
r
os
e
diagr
a
m
2.
2
.
S
p
e
c
if
ic
at
t
e
n
u
at
ion
c
alcul
at
io
n
s
T
he
s
pe
c
if
ic
a
tt
e
nua
ti
on,
,
qua
nti
f
ies
the
r
e
duc
ti
on
i
n
s
ignal
s
tr
e
ngth
pe
r
unit
length
c
a
us
e
d
by
r
a
in
,
a
s
s
umi
ng
a
c
ons
tant
r
a
te
of
r
a
inf
a
ll
a
long
the
e
nti
r
e
pa
th
of
s
ignal
pr
opa
ga
ti
on
[
33
]
.
I
t
is
a
c
r
it
ica
l
p
a
r
a
mete
r
in
unde
r
s
tanding
the
im
pa
c
t
of
r
a
inf
a
ll
on
s
igna
l
pr
opa
ga
ti
on.
I
n
thi
s
s
tudy,
wa
s
c
omput
e
d
us
i
ng
two
dis
ti
nc
t
methods
.
F
ir
s
t,
the
I
T
U
-
R
P
.
838
-
3
s
tanda
r
d
[
34
]
,
whic
h
pr
ovides
a
wide
ly
a
c
c
e
pted
f
r
a
mew
or
k
f
o
r
e
s
ti
mating
s
pe
c
if
ic
a
tt
e
nua
ti
on
ba
s
e
d
on
the
r
a
in
r
a
te
a
nd
f
r
e
que
nc
y
us
ing
the
e
xpr
e
s
s
ion
(
3
):
=
(
3)
whe
r
e
a
nd
a
r
e
c
oe
f
f
icie
nts
that
va
r
y
ba
s
e
d
on
pol
a
r
iza
ti
on
a
nd
f
r
e
que
nc
y.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
E
nhanc
ing
mil
li
me
ter
-
w
av
e
c
omm
unication:
a
tr
o
pical
pe
r
s
pe
c
ti
v
e
on
…
(
N
ur
ul
N
ajw
a
M
d
Y
us
of
)
471
B
e
s
ide
s
the
I
T
U
-
R
P
.
838
-
3
model,
s
pe
c
if
ic
a
tt
e
nua
ti
on
c
a
n
be
c
a
lcula
ted
by
incor
por
a
ti
ng
the
f
or
wa
r
d
-
s
c
a
tt
e
r
ing
a
mpl
it
ude
(
F
S
A
)
int
o
the
f
or
mul
a
.
T
he
T
-
matr
ix
method
,
a
n
a
dva
nc
e
d
c
omput
a
ti
ona
l
a
ppr
oa
c
h,
is
e
mpl
oye
d
to
de
ter
mi
ne
the
F
S
A.
T
his
method
a
c
c
ounts
f
o
r
c
ompl
e
x
s
c
a
tt
e
r
ing
pa
tt
e
r
ns
a
nd
pr
ovides
a
de
tailed
e
va
luation
o
f
how
e
lec
tr
om
a
gne
ti
c
wa
ve
s
int
e
r
a
c
t
with
r
a
indr
ops
,
e
s
pe
c
ially
whe
r
e
tr
a
dit
ional
tec
hniques
li
ke
point
matc
hing
mi
ght
f
a
ll
s
hor
t
[
35]
.
T
he
T
-
matr
ix
method
is
us
e
f
ul
f
or
a
na
lyzing
s
c
a
tt
e
r
e
d
wa
ve
s
f
r
om
non
-
s
phe
r
ica
l
r
a
indr
ops
a
nd
is
a
ls
o
r
e
li
a
ble
f
or
va
r
ious
d
r
op
s
ha
pe
s
a
nd
r
a
dio
f
r
e
que
nc
ies
[
36]
,
[
37
]
.
I
n
thi
s
s
tudy,
the
F
S
A
is
c
a
lcula
ted
us
ing
a
P
ython
im
pleme
ntation
f
or
T
-
matr
ix
s
c
a
tt
e
r
ing
c
a
lcula
ti
ons
pr
ovided
by
L
e
inonen
[
38
]
,
a
s
s
umi
ng
r
a
indr
ops
a
r
e
oblate
s
phe
r
oids
wi
th
a
de
f
ined
a
xial
r
a
ti
o.
T
he
f
or
mul
a
f
or
s
pe
c
if
ic
a
tt
e
nua
ti
on
in
dB
/km
is
pr
e
s
e
nted
in
(
4
)
:
,
=
8
.
686
×
10
3
×
λ
×
∑
(
,
(
)
∙
(
)
∙
∆
20
=
1
(
4)
whe
r
e
λ
r
e
pr
e
s
e
nts
the
wa
ve
length
in
mete
r
s
,
(
,
(
)
de
notes
the
im
a
ginar
y
c
omponent
known
a
s
the
F
S
A
in
mete
r
s
,
(
)
is
the
c
ount
of
dr
ops
pe
r
unit
vol
ume
pe
r
unit
diame
ter
of
d
r
op
a
nd
∆
s
pe
c
if
ies
the
int
e
r
va
l
of
dr
op
s
ize
in
mi
ll
i
mete
r
s
.
3.
RE
S
UL
T
S
AN
D
DI
S
CU
S
S
I
ON
T
he
r
e
s
ult
s
a
r
e
pr
e
s
e
nted
in
thi
s
s
e
c
ti
on.
S
ubs
e
c
ti
on
3.
1
pr
e
s
e
nts
c
ompar
is
on
o
f
the
diur
na
l
va
r
iations
be
twe
e
n
S
W
a
nd
NE
mons
oons
f
or
thunder
s
tor
m
e
ve
nts
.
S
ubs
e
c
ti
on
3.
2
pr
ovides
s
pe
c
if
ic
a
tt
e
nua
ti
ons
ove
r
di
f
f
e
r
e
nt
f
r
e
que
nc
ies
r
a
nging
f
r
om
8
to
48
GH
z
f
or
dif
f
e
r
e
nt
types
of
p
r
e
c
ipi
tation
with
c
ompar
is
on
to
the
I
T
U
-
R
P
.
838
-
3.
S
ubs
e
c
ti
on
3
.
3
s
hows
the
de
r
ived
a
nd
us
ing
s
upe
r
vis
e
d
ma
c
hine
lea
r
ning,
a
long
with
the
c
ompl
e
menta
r
y
c
umul
a
ti
ve
dis
tr
ibut
ion
f
unc
ti
on
of
the
s
pe
c
if
ic
a
tt
e
nua
ti
ons
c
omput
e
d
us
ing
the
de
r
ived
c
oe
f
f
icie
nts
.
3.
1.
Diu
r
n
al
var
iat
ion
s
in
r
ain
d
r
op
s
ize
d
is
t
r
ib
u
t
ion
d
u
r
in
g
m
on
s
oon
e
ve
n
t
s
A
c
ompar
is
on
o
f
S
W
a
nd
N
E
mons
oon
thunder
s
t
or
m
e
ve
nts
obs
e
r
ve
d
be
twe
e
n
3:00
a
nd
6:00
pm
loca
l
ti
me
with
r
a
in
r
a
tes
be
twe
e
n
55
a
nd
75
mm
/
h
pr
ovides
int
e
r
e
s
ti
ng
ins
ight
s
int
o
DSD
a
nd
it
s
e
f
f
e
c
ts
on
s
pe
c
if
ic
a
tt
e
nua
ti
on
m
m
-
wa
ve
f
r
e
que
nc
ies
.
F
igur
e
3
s
hows
a
n
a
na
lys
is
of
s
ix
s
a
mpl
e
e
ve
nts
f
r
om
the
S
W
mons
oon
in
F
igur
e
3
(
a
)
a
nd
the
NE
mons
oon
in
F
i
gur
e
3(
b)
,
r
e
ve
a
li
ng
that
the
maximum
DSD
c
onc
e
ntr
a
ti
on
doe
s
not
ne
c
e
s
s
a
r
il
y
r
e
s
ult
in
the
highes
t
s
p
e
c
if
ic
a
tt
e
nua
ti
on.
F
or
both
mons
oons
,
bin
12
c
ons
is
tently
s
howe
d
the
highes
t
s
pe
c
if
ic
a
tt
e
nua
ti
on
with
mea
s
ur
e
ment
s
of
0.
96,
1
.
66,
a
nd
3
.
24
dB
/km
dur
ing
the
S
W
m
ons
oon,
a
nd
s
li
ghtl
y
r
e
duc
e
d
va
lues
of
0.
89
,
1
.
53,
a
nd
2
.
98
dB
/km
dur
ing
the
N
E
mons
oon
a
t
18,
26,
a
nd
38
GH
z
f
r
e
que
nc
ies
.
T
he
c
umul
a
ti
ve
s
pe
c
if
ic
a
tt
e
nua
ti
on
f
or
a
ll
bins
e
xhibi
ted
s
im
il
a
r
t
r
e
nds
with
ove
r
a
ll
v
a
lues
of
6.
33,
11
.
35,
a
nd
17.
44
dB
/km
f
or
the
S
W
mons
o
on
c
ompar
e
d
to
6.
37,
11
.
30,
a
nd
17.
08
dB
/km
f
o
r
the
NE
mons
oon.
T
he
s
e
mi
nor
va
r
iations
a
r
e
qua
nti
f
ied
a
s
0.
04,
0
.
05,
a
nd
0
.
36
dB
/km
f
or
the
f
r
e
que
nc
ies
unde
r
s
tudy.
I
t
wa
s
a
ls
o
obs
e
r
ve
d
that
the
DSD
c
onc
e
n
tr
a
ti
on
in
bin
12
wa
s
lowe
r
du
r
ing
bo
th
mons
oo
ns
,
with
172.
48
m
-
3
mm
-
1
a
nd
158
.
62
m
-
3
mm
-
1
f
or
S
W
a
nd
NE
mons
oons
,
in
s
tar
k
c
ontr
a
s
t
to
the
pe
a
k
c
onc
e
ntr
a
ti
ons
f
ound
in
b
in
7
,
whic
h
r
e
a
c
he
d
735.
81
m
-
3
mm
-
1
a
nd
753.
89
m
-
3
mm
-
1
.
T
he
s
a
me
pa
r
a
mete
r
s
we
r
e
mea
s
ur
e
d
f
or
both
the
S
W
mons
oon
a
nd
the
NE
mons
oon,
a
s
il
lus
tr
a
ted
in
F
igu
r
e
4
.
F
igur
e
4(
a
)
s
hows
the
mea
s
ur
e
ments
f
or
the
S
W
mons
oon
du
r
ing
the
mor
ning
be
twe
e
n
3:00
a
nd
6:00
a
.
m.
,
while
F
igur
e
4
(
b)
pr
e
s
e
nts
the
c
or
r
e
s
ponding
mea
s
ur
e
ments
f
or
the
NE
mons
oon
dur
ing
t
he
s
a
me
pe
r
iod.
How
e
ve
r
,
the
r
e
s
ult
s
we
r
e
li
mi
ted
by
the
r
a
r
it
y
of
the
e
ve
nts
,
r
e
s
ult
ing
in
only
two
s
a
mpl
e
e
ve
nts
be
ing
a
na
lyze
d
f
or
e
a
c
h
mons
oon.
De
s
pit
e
the
li
mi
ted
da
ta,
the
s
tudy
highl
ight
s
the
c
ons
is
tent
obs
e
r
va
ti
on
that
the
de
ns
e
r
DSD
is
not
a
s
s
oc
iate
d
with
gr
e
a
t
e
r
s
pe
c
if
ic
a
tt
e
nua
ti
on.
F
oc
us
ing
on
bin
12,
the
a
na
lys
is
r
e
ve
a
led
that
the
highes
t
s
pe
c
if
ic
a
tt
e
nua
ti
on
va
lues
a
r
e
1
.
15,
1
.
99,
a
nd
3.
88
dB
/km
f
or
the
S
W
m
o
ns
o
o
n
a
nd
1
.
0
3
,
1
.
77
,
a
n
d
3
.
45
dB
/k
m
f
o
r
th
e
N
E
mo
ns
o
on
a
t
1
8
,
2
6
,
a
nd
38
GH
z
,
r
e
s
pe
c
ti
ve
ly
.
T
he
a
g
gr
e
g
a
t
e
s
pe
c
i
f
ic
a
t
ten
ua
ti
on
v
a
l
ue
s
f
o
r
a
ll
b
in
s
we
r
e
6
.
07
,
1
1
.
3
5
,
a
n
d
1
8
.
5
1
dB
/k
m
f
or
th
e
S
W
m
ons
oo
n
,
c
o
mp
a
r
e
d
t
o
6
.
31
,
1
1
.
49
,
a
n
d
1
8
.
3
4
d
B
/
km
f
o
r
t
he
N
E
mo
ns
o
on
.
S
li
gh
t
va
r
iat
i
ons
be
twe
e
n
t
he
s
e
m
ons
oo
n
p
e
r
i
ods
a
r
e
0
.
2
4
,
0
.
1
4
,
a
nd
0
.
17
dB
/
k
m
f
o
r
th
e
t
h
r
e
e
f
r
e
q
ue
n
c
i
e
s
e
xa
mi
n
e
d
.
F
u
r
t
he
r
m
o
r
e
,
i
t
wa
s
n
ot
ice
d
th
a
t
i
n
b
in
1
2
,
t
he
D
S
D
c
on
c
e
n
t
r
a
ti
on
on
ly
r
e
a
c
he
d
20
6
.
6
4
m
-
3
mm
-
1
f
o
r
the
S
W
mo
ns
o
on
a
nd
18
3
.
7
5
m
-
3
mm
-
1
f
o
r
th
e
N
E
m
o
ns
o
on
.
T
h
is
c
o
nt
r
a
s
ts
wi
th
th
e
p
e
a
k
c
o
nc
e
n
t
r
a
t
i
ons
obs
e
r
ve
d
in
bi
n
7
,
wh
e
r
e
v
a
l
ue
s
we
r
e
r
e
c
o
r
de
d
a
t
8
24
.
93
m
-
3
mm
-
1
a
nd
10
92
.
75
m
-
3
mm
-
1
.
T
he
r
e
s
ult
s
indi
c
a
te
that
dur
ing
the
diu
r
na
l
va
r
i
a
ti
ons
f
or
both
mons
oons
,
bins
with
the
highes
t
number
of
r
a
indr
ops
do
not
ne
c
e
s
s
a
r
il
y
lea
d
to
t
he
gr
e
a
tes
t
a
tt
e
nua
ti
on
due
to
incr
e
a
s
e
d
F
S
A
f
r
o
m
lar
ge
r
r
a
indr
op
pa
r
ti
c
les
[
39
]
.
I
n
thi
s
s
tudy,
the
g
r
e
a
tes
t
a
tt
e
nua
ti
on
wa
s
c
a
us
e
d
by
r
a
indr
op
diame
ter
s
r
a
nging
be
twe
e
n
1.
91
mm
a
nd
3.
20
mm
.
S
i
mi
lar
r
e
s
ult
s
we
r
e
r
e
por
ted
in
Du
r
ba
n
whe
r
e
c
r
it
ica
l
r
a
indr
op
diame
ter
s
c
ontr
ibut
ing
mos
t
to
s
pe
c
if
ic
a
tt
e
nua
ti
on
we
r
e
f
ou
nd
to
be
in
the
r
a
nge
of
1
.
0
to
3
.
0
mm
f
or
c
onve
c
t
ive
r
a
in
,
w
h
ich
i
nc
lu
de
s
t
hu
nd
e
r
s
t
or
m
r
a
i
n
[
4
0
]
.
A
mo
r
e
r
e
c
e
n
t
s
t
ud
y
in
t
he
s
a
me
a
r
e
a
ha
d
s
l
ig
ht
ly
d
i
f
f
e
r
e
nt
r
e
s
u
lt
s
w
he
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
:
467
-
479
472
t
he
s
pe
c
i
f
i
c
a
t
ten
ua
ti
on
p
e
a
ke
d
b
e
t
we
e
n
1
.
4
a
nd
2.
7
mm
f
o
r
f
r
e
q
ue
n
c
i
e
s
be
twe
e
n
1
0
to
1
00
G
Hz
[
4
1
]
.
T
he
i
n
f
lue
nc
e
o
f
r
a
i
nd
r
o
p
di
a
m
e
te
r
o
n
a
tt
e
n
ua
ti
on
wa
s
a
ls
o
e
vi
de
n
t
in
t
he
f
in
di
ngs
f
r
om
R
w
a
n
da
whe
r
e
t
he
la
r
ge
s
t
c
on
t
r
i
bu
t
io
ns
t
o
s
p
e
c
if
ic
a
t
te
nua
t
io
n
we
r
e
f
r
om
d
r
o
p
d
ia
me
te
r
s
in
th
e
r
a
n
ge
of
1
.
5
t
o
3
.
5
m
m
[
4
2
]
.
(
a
)
(
b)
F
igur
e
3.
M
e
a
s
ur
e
d
DSD
a
nd
s
pe
c
if
ic
a
tt
e
nua
ti
on
a
t
dif
f
e
r
e
nt
f
r
e
que
nc
ies
(
18
,
26
,
a
nd
38
GH
z
)
du
r
in
g
thunder
s
tor
m
e
ve
nts
obs
e
r
ve
d
be
twe
e
n
3:00
-
6:00
p
m
with
r
a
in
r
a
tes
of
55
-
75
mm
/h
dur
ing
(
a
)
S
W
m
ons
oon
a
nd
(
b)
NE
mons
oon
F
ield
s
tudi
e
s
ha
ve
c
onf
ir
med
that
r
a
ind
r
ops
,
pa
r
ti
c
ular
ly
thos
e
a
r
ound
1
mm
in
diame
ter
,
be
gin
to
os
c
il
late
due
to
e
ddy
s
he
dding,
whic
h
s
igni
f
ica
ntl
y
a
lt
e
r
s
their
s
ha
pe
a
nd
thus
their
s
c
a
tt
e
r
ing
c
ha
r
a
c
ter
is
ti
c
s
.
Obs
e
r
va
ti
ons
of
f
o
r
wa
r
d
s
c
a
tt
e
r
ove
r
long
pa
ths
h
a
ve
de
mons
tr
a
ted
that
r
a
in
c
a
n
s
igni
f
ica
ntl
y
inc
r
e
a
s
e
s
ignal
leve
ls
,
with
lar
ge
r
dr
ops
c
ontr
ibu
ti
ng
mor
e
to
thi
s
e
f
f
e
c
t
due
to
their
gr
e
a
ter
s
c
a
tt
e
r
ing
c
r
os
s
-
s
e
c
ti
ons
[
43]
.
Additi
ona
ll
y,
theo
r
e
ti
c
a
l
a
na
lys
e
s
s
howe
d
that
the
f
or
wa
r
d
s
c
a
tt
e
r
ing
int
e
ns
it
y
is
s
e
ns
it
ive
to
the
pa
r
ti
c
le
diame
ter
,
wi
th
lar
ge
r
pa
r
ti
c
les
e
xhibi
ti
ng
mo
r
e
pr
onounc
e
d
s
c
a
tt
e
r
ing
e
f
f
e
c
ts
[
44
]
.
T
a
ke
n
togeth
e
r
,
thes
e
r
e
s
ult
s
highl
ight
that
r
a
indr
op
diame
ter
plays
a
c
r
it
ica
l
r
ole
in
de
ter
mi
ning
f
o
r
wa
r
d
s
c
a
tt
e
r
ing
a
mpl
it
u
de
,
with
lar
ge
r
dr
ops
c
a
us
ing
s
tr
onge
r
s
c
a
tt
e
r
ing
e
f
f
e
c
ts
d
ue
to
their
lar
ge
r
s
ize
,
os
c
il
lations
,
a
nd
int
e
r
na
l
s
tr
uc
tur
a
l
va
r
iations
.
(
a
)
(
b)
F
igur
e
4.
M
e
a
s
ur
e
d
DSD
a
nd
s
pe
c
if
ic
a
tt
e
nua
ti
on
a
t
dif
f
e
r
e
nt
f
r
e
que
nc
ies
(
18
,
26
,
a
nd
38
GH
z
)
du
r
in
g
thunder
s
tor
m
e
ve
nts
obs
e
r
ve
d
be
twe
e
n
3:00
-
6:00
a
m
with
r
a
in
r
a
tes
of
55
-
75
mm
/h
dur
ing
(
a
)
S
W
m
ons
oon
a
nd
(
b)
NE
mons
oon
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
E
nhanc
ing
mil
li
me
ter
-
w
av
e
c
omm
unication:
a
tr
o
pical
pe
r
s
pe
c
ti
v
e
on
…
(
N
ur
ul
N
ajw
a
M
d
Y
us
of
)
473
3.
2.
S
p
e
c
if
ic
at
t
e
n
u
at
ion
f
or
d
if
f
e
r
e
n
t
r
ain
t
yp
e
s
ac
r
os
s
f
r
e
q
u
e
n
c
ies
F
igur
e
s
5
a
nd
F
igur
e
s
6
highl
ight
s
igni
f
ica
nt
dif
f
e
r
e
nc
e
s
in
s
pe
c
if
ic
a
tt
e
nua
ti
on
va
lues
f
or
va
r
ious
r
a
in
types
dur
ing
the
S
W
mons
oon,
il
lus
tr
a
ted
in
F
igur
e
s
5(
a
)
a
nd
6(
a
)
,
a
nd
the
N
E
mons
oons
,
il
lus
tr
a
ted
in
F
igur
e
s
5(
b)
a
nd
6(
b
)
.
T
he
f
igu
r
e
s
il
lus
tr
a
te
how
the
mea
s
ur
e
d
s
pe
c
if
ic
a
tt
e
nua
ti
on
f
or
dr
izz
le,
s
ho
we
r
,
a
nd
thunder
s
tor
m
de
viate
s
f
r
om
the
I
T
U
-
R
model
a
c
r
os
s
a
f
r
e
que
nc
y
r
a
nge
f
r
om
8
to
48
GH
z
.
M
e
a
s
u
r
e
ments
we
r
e
take
n
be
twe
e
n
3:00
-
6:00
pm
a
nd
3:00
-
6:00
a
m,
with
r
a
inf
a
ll
r
a
tes
r
a
nging
f
r
om
55
to
75
mm
/h.
S
pe
c
if
ic
a
tt
e
nua
ti
on
incr
e
a
s
e
s
with
f
r
e
que
nc
y
f
or
a
ll
c
o
ndit
ions
,
a
c
omm
on
tr
e
nd
due
to
higher
f
r
e
que
nc
y
r
a
in
a
tt
e
nua
ti
on.
(
a
)
(
b)
F
igur
e
5.
C
ompar
is
on
of
s
pe
c
if
ic
a
tt
e
nua
ti
on
f
or
d
i
f
f
e
r
e
nt
r
a
in
types
a
ga
ins
t
I
T
U
-
R
P
.
838
-
3
pr
e
dictio
ns
dur
ing
(
a
)
S
W
mons
oon
a
nd
(
b
)
NE
mons
oon
be
tw
e
e
n
3:00
to
6:00
pm
(
a
)
(
b)
F
igur
e
6.
C
ompar
is
on
of
s
pe
c
if
ic
a
tt
e
nua
ti
on
f
or
d
i
f
f
e
r
e
nt
r
a
in
types
a
ga
ins
t
I
T
U
-
R
P
.
838
-
3
pr
e
dictio
ns
dur
ing
(
a
)
S
W
mons
oon
a
nd
(
b
)
NE
mons
oon
be
tw
e
e
n
3:00
to
6:00
a
m
I
n
the
e
ve
ning,
dur
ing
the
S
W
mons
oon,
dr
izz
l
e
a
nd
s
howe
r
c
los
e
ly
f
oll
ow
the
I
T
U
-
R
model,
s
ugge
s
ti
ng
it
s
r
e
li
a
bil
it
y
f
or
thes
e
r
a
in
types
.
F
or
i
ns
tanc
e
,
a
t
38
GH
z
,
s
pe
c
if
ic
a
tt
e
nua
ti
on
is
15.
6
dB
/km
f
or
dr
izz
le
a
nd
15
.
8
dB
/km
f
o
r
s
howe
r
,
a
li
gning
with
I
T
U
-
R
pr
e
dictions
.
How
e
ve
r
,
thunde
r
s
tor
m
s
hows
s
pe
c
if
ic
a
tt
e
nua
ti
on
a
t
17.
4
dB
/km
whic
h
is
higher
than
the
model
e
s
ti
mate
s
,
li
ke
ly
due
to
lar
ge
r
r
a
ind
r
ops
.
I
n
the
NE
mons
oon,
dis
c
r
e
pa
nc
ies
with
the
I
T
U
-
R
model
be
c
ome
pr
onounc
e
d
a
bove
30
GH
z
,
with
s
pe
c
if
ic
a
tt
e
nua
ti
on
va
lues
f
or
d
r
izz
le,
s
howe
r
,
a
nd
thunder
s
tor
m
a
r
e
17.
5,
17.
3
,
a
nd
17.
1
dB
/km
r
e
s
pe
c
ti
ve
ly,
e
xc
e
e
di
ng
model
pr
e
dictions
.
M
or
ning
s
pe
c
if
ic
a
tt
e
nua
ti
on
va
r
ies
mor
e
s
igni
f
ica
ntl
y
than
in
the
a
f
ter
noon.
Dur
ing
the
S
W
mons
oon,
obs
e
r
ve
d
va
lues
a
r
e
14.
4
dB
/km
f
or
dr
izz
le,
17.
8
dB
/km
f
o
r
s
howe
r
s
,
a
nd
18.
5
d
B
/km
f
or
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
:
467
-
479
474
thunder
s
tor
m,
pr
oving
the
unde
r
e
s
ti
mation
of
t
he
I
T
U
-
R
model.
I
n
the
NE
mons
oon,
the
va
lues
a
r
e
17.
5
dB
/km
f
o
r
d
r
izz
le,
19
.
7
dB
/km
s
howe
r
s
,
a
nd
18.
3
dB
/km
f
or
thunder
s
tor
m,
a
ls
o
s
ur
pa
s
s
ing
model
pr
e
dictions
.
S
ome
o
f
thes
e
dif
f
e
r
e
nc
e
s
a
r
e
lar
ge
r
t
ha
n
2
dB
/km
,
whic
h
c
a
n
s
igni
f
ica
ntl
y
im
pa
c
t
the
de
s
ign
of
wir
e
les
s
c
omm
unica
ti
on
s
y
s
tems
,
r
e
quir
ing
a
djus
tm
e
nts
in
powe
r
mar
gins
or
li
nk
dis
tanc
e
s
to
maintain
s
ignal
qua
li
ty
[
45]
.
T
he
r
e
s
ult
s
r
e
ve
a
l
that
the
NE
mons
oon
e
xhibi
t
s
higher
s
pe
c
if
ic
a
tt
e
nua
ti
on
f
or
a
ll
r
a
in
types
,
whe
r
e
a
s
the
S
W
mons
oon
s
hows
a
mi
xe
d
pa
tt
e
r
n,
with
a
tt
e
nua
ti
on
va
lues
that
a
r
e
s
ometim
e
s
highe
r
,
lowe
r
,
or
c
los
e
ly
f
o
ll
owing
the
I
T
U
-
R
model.
T
his
may
b
e
due
to
the
N
E
mons
oon
c
ontr
ibut
ing
mor
e
r
a
inf
a
ll
a
c
r
os
s
P
e
nins
ular
M
a
lays
ia
than
the
S
W
mons
oon,
c
ou
pled
with
the
f
a
c
t
that
mos
t
a
r
e
a
s
in
P
e
nins
ular
M
a
lays
ia
ha
ve
be
c
ome
dr
ier
dur
ing
the
S
W
mons
oon
ove
r
the
las
t
thr
e
e
de
c
a
de
s
[
46]
.
I
n
c
ompar
is
on,
a
s
tudy
in
Ga
da
nki
[
47]
f
ound
that
the
DSD
dur
ing
the
S
W
mons
oon
ha
s
mor
e
lar
ge
dr
ops
a
nd
f
e
we
r
s
m
a
ll
dr
ops
c
ompar
e
d
to
the
NE
mons
oon,
whic
h
ha
s
mor
e
s
mall
dr
ops
a
nd
f
e
we
r
la
r
ge
dr
ops
,
e
s
pe
c
ially
a
t
low
r
a
in
r
a
tes
.
L
a
r
ge
r
r
a
ind
r
ops
c
a
us
e
higher
a
tt
e
nua
ti
o
n,
r
e
s
ult
ing
in
higher
a
tt
e
nua
ti
on
du
r
ing
S
W
mons
oon
c
ompar
e
d
to
NE
mons
oon,
c
ont
r
a
r
y
to
thi
s
s
tudy’
s
obs
e
r
va
ti
ons
.
T
he
diur
na
l
a
nd
s
e
a
s
ona
l
va
r
iations
in
DSD
a
r
e
a
ls
o
e
vident
in
other
tr
opica
l
loca
ti
ons
li
ke
S
ingapor
e
a
nd
Kototaba
ng,
whe
r
e
loca
l
c
onve
c
ti
o
n
a
nd
oc
e
a
n
-
land
c
ontr
a
s
ts
s
igni
f
ica
ntl
y
in
f
luenc
e
r
a
in
f
a
ll
c
ha
r
a
c
ter
is
ti
c
s
[
48]
.
I
n
S
ingapor
e
,
the
va
r
iations
in
DSD
a
r
e
les
s
pr
onounc
e
d
c
ompa
r
e
d
to
Ga
d
a
nki
a
nd
Kototaba
ng.
T
his
s
tabili
ty
in
DSD
is
a
tt
r
ibut
e
d
to
S
ingapor
e
’
s
unique
pos
it
ion,
inf
luenc
e
d
by
both
land
a
nd
oc
e
a
nic
r
a
inf
a
ll
,
r
e
s
ult
ing
in
mor
e
s
table
r
a
inf
a
l
l
pa
tt
e
r
ns
thr
oughout
the
ye
a
r
.
C
onve
r
s
e
ly,
Kot
otaba
ng
e
xhibi
ts
s
igni
f
ica
nt
diur
na
l
va
r
iations
in
DSD,
inf
luenc
e
d
by
oc
e
a
n
-
land
c
ontr
a
s
t
a
nd
mount
a
in
e
f
f
e
c
ts
that
ge
ne
r
a
te
loca
l
c
onve
c
ti
on,
e
s
pe
c
ially
in
the
a
f
ter
n
oon.
Du
r
ing
bo
th
the
S
W
a
nd
NE
mons
oons
,
Kot
otaba
ng
e
xpe
r
ienc
e
s
notable
c
ha
nge
s
in
DSD,
lea
ding
to
va
r
iations
in
r
a
ind
r
op
s
ize
s
a
nd
r
a
inf
a
l
l
int
e
ns
it
y.
I
t
is
e
xpe
c
ted
that
thi
s
s
tudy
will
s
how
how
t
he
I
T
U
-
R
P
.
838
model
unde
r
e
s
ti
mate
s
s
pe
c
if
ic
a
tt
e
nua
ti
on
in
r
e
gions
with
s
igni
f
ica
nt
diur
na
l
a
nd
s
e
a
s
ona
l
va
r
iations
.
T
he
r
e
s
ult
s
c
onf
ir
m
thi
s
a
nd
s
how
higher
a
tt
e
nua
ti
on
va
lues
than
p
r
e
dicte
d
by
the
I
T
U
-
R
model,
e
s
pe
c
ially
du
r
ing
he
a
vy
r
a
inf
a
ll
e
ve
nt
s
s
uc
h
a
s
thunder
s
tor
ms
a
nd
s
howe
r
s
in
both
the
S
W
a
nd
NE
mons
oons
.
T
he
s
e
obs
e
r
ve
d
va
r
iations
in
D
S
D
a
nd
pr
e
c
ipi
tation
c
ha
r
a
c
ter
is
ti
c
s
highl
ight
the
c
ompl
e
xit
y
a
nd
r
e
gional
s
pe
c
if
icity
of
tr
opica
l
pr
e
c
ipi
ta
ti
on
a
nd
s
ugge
s
t
that
loca
l
model
a
djus
tm
e
nts
may
be
r
e
qu
ir
e
d
to
c
a
ptur
e
the
nua
nc
e
s
of
mons
oon
r
a
inf
a
ll
in
tr
opica
l
r
e
gions
.
3.
3.
De
r
ivat
ion
of
an
d
u
s
in
g
s
u
p
e
r
vis
e
d
m
ac
h
i
n
e
lear
n
in
g
S
upe
r
vis
e
d
mac
hine
lea
r
ning
(
S
M
L
)
s
igni
f
ica
ntl
y
im
pr
ove
s
r
a
in
a
tt
e
nua
ti
on
pr
e
diction
models
by
leve
r
a
ging
his
tor
ica
l
da
ta
to
identif
y
pa
tt
e
r
ns
a
nd
r
e
lations
hips
that
tr
a
dit
ional
s
tatis
ti
c
a
l
mod
e
ls
may
ove
r
look.
As
mentioned
be
f
or
e
,
the
model
r
e
c
om
mende
d
by
the
I
T
U
-
R
of
ten
r
e
li
e
s
on
c
ompl
e
x
mat
he
matica
l
f
or
mul
a
ti
ons
a
nd
e
mpi
r
ica
l
da
ta
that
may
not
ge
n
e
r
a
li
z
e
we
ll
a
c
r
os
s
di
f
f
e
r
e
nt
c
li
matic
z
one
s
,
pa
r
ti
c
ular
ly
in
r
e
gions
with
he
a
vy
r
a
inf
a
ll
li
ke
the
tr
opics
[
49
]
,
[
50]
.
S
M
L
models
c
a
n
be
t
r
a
ined
us
ing
loca
ll
y
mea
s
ur
e
d
da
ta,
incr
e
a
s
ing
their
a
da
ptabili
ty
a
nd
e
f
f
e
c
ti
ve
ne
s
s
f
or
s
pe
c
if
ic
r
e
gions
.
F
or
ins
tanc
e
,
a
mac
hine
lea
r
ning
model
tr
a
ined
on
s
e
ve
n
ye
a
r
s
o
f
r
a
inf
a
ll
da
ta
in
Nige
r
ia
outper
f
o
r
med
the
I
T
U
-
R
model
in
pr
e
dic
ti
ng
r
a
n
a
tt
e
nua
ti
on,
with
a
lowe
r
r
oot
-
mea
n
-
s
qua
r
e
(
R
M
S
E
)
[
49
]
.
S
im
il
a
r
ly
,
a
ne
ur
a
l
ne
twor
k
model
ha
s
pr
o
ve
n
to
be
mor
e
e
f
f
e
c
ti
ve
than
the
e
mpi
r
ica
l
powe
r
-
law
a
ppr
oa
c
h,
lea
ding
to
a
c
ons
ider
a
ble
r
e
duc
ti
on
in
t
he
mea
n
s
qua
r
e
e
r
r
or
[
51]
.
Us
ing
S
M
L
in
r
a
in
a
tt
e
nua
ti
on
p
r
e
diction
is
not
l
im
it
e
d
to
s
im
ple
ne
u
r
a
l
ne
twor
ks
;
mo
r
e
c
ompl
e
x
a
r
c
hit
e
c
tur
e
s
li
ke
c
onvolut
ional
ne
ur
a
l
ne
twor
ks
(
C
NN
s
)
ha
ve
a
ls
o
be
e
n
e
mpl
oye
d.
F
o
r
e
xa
mpl
e
,
a
C
NN
tr
a
ined
on
lar
ge
a
mount
s
of
da
ta
f
r
om
c
omm
e
r
c
ial
mi
c
r
owa
ve
li
nks
c
ould
de
tec
t
r
a
inf
a
ll
-
s
pe
c
if
ic
a
tt
e
nua
ti
on
pa
tt
e
r
ns
with
r
e
a
s
ona
ble
a
c
c
ur
a
c
y
a
nd
r
e
duc
ing
f
a
ls
e
ly
e
s
ti
mate
d
r
a
inf
a
ll
[
52]
.
T
he
a
da
ptabili
ty
of
S
M
L
models
e
na
bles
the
incor
por
a
t
ion
o
f
va
r
ious
input
pa
r
a
mete
r
s
,
s
uc
h
a
s
pa
th
length,
ope
r
a
ti
ng
f
r
e
que
nc
y,
wa
ve
polar
iza
ti
on,
a
nd
r
a
in
r
a
te
dis
tr
ibut
ion.
T
his
f
e
a
tur
e
e
na
bles
them
to
dis
c
ove
r
hidden
c
onne
c
ti
ons
a
nd
e
nha
nc
e
pr
e
diction
without
the
ne
e
d
f
or
c
ompl
e
x
mathe
matica
l
e
xpr
e
s
s
ions
[
53]
,
making
S
M
L
a
va
luable
to
ol
in
the
ongoing
e
f
f
or
t
to
im
pr
ove
r
a
in
a
tt
e
nua
ti
on
pr
e
dictions
a
nd
mi
ti
ga
te
the
a
dve
r
s
e
e
f
f
e
c
ts
of
we
a
ther
on
c
omm
unica
ti
on
s
ys
tems
.
I
n
th
is
s
tudy,
the
S
M
L
a
ppr
oa
c
h
with
li
ne
a
r
r
e
gr
e
s
s
ion
wa
s
us
e
d
to
de
r
ive
a
nd
va
lues
f
or
s
pe
c
if
ic
a
tt
e
nua
ti
on
f
or
dif
f
e
r
e
nt
r
a
in
types
a
t
38
GH
z
f
r
e
que
nc
y
du
r
ing
the
S
W
a
nd
NE
mons
oo
ns
.
A
s
e
t
making
up
80%
of
tr
a
ini
ng
da
ta
a
nd
20%
o
f
tes
t
da
ta
wa
s
us
e
d
to
va
li
da
te
the
mac
hine
lea
r
ning
model's
pe
r
f
or
manc
e
.
T
he
S
M
L
a
lgo
r
it
hm
wa
s
im
pleme
n
ted
us
ing
P
ython,
a
nd
the
c
ode
wa
s
wr
it
ten
in
a
J
upyter
Note
book
f
r
om
Ana
c
onda
Na
vigator
.
T
he
r
e
s
ult
in
g
va
lues
a
r
e
pr
e
s
e
nted
in
T
a
ble
2
a
longs
ide
the
s
tatis
ti
c
a
l
method
f
or
c
ompar
is
on,
togethe
r
with
thei
r
c
o
r
r
e
s
ponding
R
-
s
qua
r
e
d
a
nd
R
M
S
E
va
lues
.
Not
a
bly,
the
dif
f
e
r
e
nc
e
s
be
twe
e
n
S
M
L
a
nd
s
tatis
ti
c
a
l
method
-
de
r
ived
a
nd
a
r
e
r
e
latively
s
mall,
with
h
igh
R
-
s
qua
r
e
d
va
lues
indi
c
a
ti
ng
good
f
it
s
f
or
both
methods
.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
nt
J
E
lec
&
C
omp
E
ng
I
S
S
N:
2088
-
8708
E
nhanc
ing
mil
li
me
ter
-
w
av
e
c
omm
unication:
a
tr
o
pical
pe
r
s
pe
c
ti
v
e
on
…
(
N
ur
ul
N
ajw
a
M
d
Y
us
of
)
475
T
he
s
e
mi
nor
dif
f
e
r
e
nc
e
s
be
twe
e
n
the
S
M
L
a
nd
s
t
a
ti
s
ti
c
a
l
method
de
r
ived
c
oe
f
f
icie
nts
s
ugge
s
t
that
the
li
ne
a
r
r
e
gr
e
s
s
ion
model
ha
s
e
f
f
e
c
ti
ve
ly
c
a
ptur
e
d
the
unde
r
lyi
ng
r
e
lations
hips
withi
n
the
da
ta,
whic
h
the
s
tatis
ti
c
a
l
method
ha
s
tr
a
dit
ionally
a
ppr
oxim
a
ted.
T
his
a
ls
o
indi
c
a
tes
that
the
mac
hine
le
a
r
ning
a
ppr
oa
c
h
doe
s
not
s
igni
f
ica
ntl
y
ove
r
f
it
de
s
pit
e
it
s
c
ompl
e
xit
y
,
a
nd
the
model
is
we
ll
-
tuned
with
a
pr
ope
r
ly
s
ize
d
da
ta
s
e
t
[
45]
.
T
he
c
ons
is
tenc
y
be
twe
e
n
e
s
tablis
he
d
models
is
f
ur
ther
va
li
da
ted
by
c
ompar
ing
the
k
a
nd
α
va
l
ue
s
with
f
indi
ngs
f
r
om
Alhil
a
li
e
t
al
.
[
54
]
a
nd
L
a
m
e
t
al
.
[
55]
,
whos
e
s
tudi
e
s
we
r
e
c
onduc
ted
in
J
ohor
B
a
hr
u
a
nd
Kua
la
L
umpur
,
M
a
lays
ia,
c
or
r
e
s
pondingl
y.
I
t
is
im
por
tant
to
note,
howe
ve
r
,
that
thes
e
s
tudi
e
s
did
not
c
a
tegor
ize
pr
e
c
ipi
tation
by
r
a
in
types
.
F
igur
e
7
dis
plays
the
diur
na
l
va
r
iations
of
the
C
C
DF
to
a
s
s
e
s
s
the
inf
luenc
e
of
r
a
ind
r
op
s
ize
dis
tr
ibut
ion
on
m
m
-
wa
ve
pr
opa
ga
ti
on
a
t
38
GH
z
a
t
s
e
pa
r
a
te
ti
mes
unde
r
dif
f
e
r
e
nt
mons
oon
c
ondit
ions
us
ing
the
s
a
me
s
e
ts
of
s
a
mpl
e
s
a
s
in
s
e
c
ti
on
3.
1.
F
igur
e
s
7(
a
)
a
nd
7(
b
)
pr
e
s
e
nt
the
C
C
DFs
f
or
dr
izz
le
dur
ing
the
e
a
r
ly
mo
r
ning
(
3:
00
to
6:00
a
.
m.
)
a
nd
late
a
f
te
r
noon
(
3:00
to
6:0
0
p.
m.
)
,
r
e
s
pe
c
ti
ve
ly.
F
igur
e
s
7(
c
)
a
nd
7
(
d)
il
lus
tr
a
te
the
r
e
s
ult
s
f
or
s
howe
r
c
ondit
ions
dur
ing
the
s
a
me
ti
me
pe
r
iods
,
while
F
igur
e
s
7(
e
)
a
nd
7
(
f
)
de
pict
the
C
C
DFs
f
or
t
hunde
r
s
tor
ms
,
a
ga
in
c
ompar
ing
the
e
a
r
ly
mor
ning
with
the
late
a
f
ter
noon
,
r
e
s
pe
c
ti
ve
ly.
T
he
s
e
f
igur
e
s
a
ls
o
c
o
mpar
e
pr
e
dictions
f
r
om
I
T
U
-
R
P
.
838
-
3
togethe
r
with
[
54
]
a
nd
[
55]
.
T
he
C
C
DFs
incor
po
r
a
te
da
ta
po
int
s
f
o
r
d
r
izz
le,
s
howe
r
,
a
nd
thunder
s
tor
m
c
ondit
ions
,
us
ing
k
a
nd
α
c
oe
f
f
icie
nts
de
r
ived
us
ing
S
M
L
a
s
outl
ined
in
T
a
bl
e
2.
T
a
ble
2.
C
oe
f
f
icie
nt
o
f
a
nd
a
t
38
GH
z
R
a
in
T
ype
2
R
M
S
E
S
out
hw
e
s
t
mons
oon
S
ta
ti
s
ti
c
a
l
me
th
od
D
r
iz
z
le
0.27933
1.01685
0.99058
0.75917
S
how
e
r
0.30097
0.97304
0.9872
0.96499
T
hunde
r
s
to
r
m
0.29235
0.98425
0.99282
1.19772
S
M
L
D
r
iz
z
le
0.27897
1.01678
0.99193
0.80678
S
how
e
r
0.29961
0.97386
0.98454
0.92712
T
hunde
r
s
to
r
m
0.29152
0.98469
0.99250
1.13874
N
or
th
e
a
s
t
mons
oon
S
ta
ti
s
ti
c
a
l
me
th
od
D
r
iz
z
le
0.27816
1.03249
0.99296
0.72578
S
how
e
r
0.29047
0.00521
0.99214
0.83533
T
hunde
r
s
to
r
m
0.28855
0.9938
0.99306
1.20399
S
M
L
D
r
iz
z
le
0.27765
1.03288
0.99309
0.56209
S
how
e
r
0.29699
0.99543
0.99077
0.95248
T
hunde
r
s
to
r
m
0.28988
0.99279
0.99285
1.16387
A
lh
il
a
li
e
t
al
.
[
54]
-
0.40000
0.93500
0.99700
-
L
a
m
e
t
al
.
[
55]
-
0.35250
0.94820
0.98200
-
T
he
s
e
plot
s
,
juxt
a
pos
e
d
with
the
I
T
U
-
R
P
.
838
-
3
model,
s
how
that
a
ll
r
a
in
types
,
including
thos
e
s
tudi
e
d
in
[
54]
,
[
55]
,
c
ons
is
tently
r
e
gis
ter
a
bove
the
I
T
U
-
R
li
ne
.
T
he
s
e
dis
c
r
e
pa
nc
ies
highl
ight
th
e
I
T
U
-
R
model's
tempe
r
a
te
bias
,
whic
h
may
not
a
c
c
ur
a
tely
c
a
ptur
e
the
mor
e
pr
onounc
e
d
r
a
in
-
induce
d
a
tt
e
nua
ti
on
a
t
higher
f
r
e
que
nc
ies
c
ha
r
a
c
ter
is
ti
c
of
t
r
opica
l
r
e
gio
ns
with
he
a
vy
r
a
inf
a
ll
a
nd
high
-
wa
ter
c
ontent
c
lo
uds
[
50]
,
[
56]
.
T
he
phys
ica
l
ba
s
is
of
the
I
T
U
-
R
model,
whic
h
a
im
s
to
maximi
z
e
a
c
c
or
da
nc
e
with
e
xpe
r
im
e
n
tal
da
ta,
a
ls
o
f
a
c
e
s
c
ha
ll
e
nge
s
in
tr
opica
l
r
e
gions
be
c
a
us
e
of
the
unique
a
tm
os
phe
r
ic
c
ondit
ions
,
whic
h
a
r
e
not
f
ull
y
a
c
c
ounted
f
or
in
the
model
[
57
]
.
Additi
ona
ll
y
,
the
f
r
e
que
nt
high
winds
in
tr
opica
l
a
r
e
a
s
may
not
be
a
de
qua
tely
a
ddr
e
s
s
e
d
by
the
model
,
lea
ding
to
p
otential
unde
r
e
s
ti
mation
of
a
tt
e
nua
ti
on
[
15
]
.
S
tud
ies
f
r
om
M
a
lays
ia
a
nd
Ota
,
Nige
r
ia,
ha
ve
high
li
ghted
th
e
inade
qua
c
y
of
I
T
U
-
R
models
in
thes
e
r
e
gion
s
,
whe
r
e
e
xis
ti
ng
models
unde
r
e
s
ti
mate
d
the
a
tt
e
nua
ti
on
mar
gins
r
e
quir
e
d
f
or
r
e
li
a
ble
c
omm
unica
ti
on
[
5
8]
,
[
59]
.
T
he
s
e
s
tudi
e
s
e
mphas
ize
the
ne
e
d
f
or
ne
w
mode
ls
that
be
tt
e
r
p
r
e
dict
r
a
in
a
tt
e
nua
ti
on
in
tr
opica
l
c
li
mate
s
.
E
xpe
r
im
e
ntal
s
tudi
e
s
in
tr
opica
l
loca
ti
ons
a
ls
o
hi
ghli
ghted
the
c
ombi
ne
d
e
f
f
e
c
ts
o
f
r
a
in
a
nd
wind
on
s
ignal
a
tt
e
nua
ti
on,
s
ugge
s
ti
ng
that
the
I
T
U
-
R
model
c
ould
be
s
uppleme
nted
with
a
ddit
ional
models
to
a
c
c
ount
f
or
thes
e
f
a
c
tor
s
[
15]
.
T
he
de
r
ived
k
a
nd
α
c
oe
f
f
icie
nts
in
thi
s
s
tudy
a
r
e
tailor
e
d
to
the
s
pe
c
if
ic
c
ondit
ions
mea
s
ur
e
d,
potentially
pr
ovid
ing
im
pr
ove
d
a
c
c
ur
a
c
y
of
s
pe
c
if
ic
a
tt
e
nua
ti
on
than
the
s
tanda
r
d
I
T
U
-
R
model
.
S
i
gnif
ica
nt
dif
f
e
r
e
nc
e
s
be
twe
e
n
thes
e
mac
hine
lea
r
ning
-
ba
s
e
d
pr
e
dictions
a
nd
the
I
T
U
-
R
model
may
indi
c
a
te
the
ne
e
d
f
or
r
e
gional
o
r
c
ondit
ion
-
s
pe
c
if
ic
e
nha
nc
e
ments
to
r
a
in
a
tt
e
nua
ti
on
pr
e
dictions
.
S
pe
c
if
ic
a
tt
e
nua
ti
on,
e
s
pe
c
ially
in
the
mm
-
wa
ve
f
r
e
que
nc
ies
,
ha
s
s
ubs
t
a
nti
a
l
c
ons
e
que
nc
e
s
f
or
s
ignal
pr
opa
ga
ti
on.
Ove
r
e
s
ti
mation
c
a
n
lea
d
to
c
ons
e
r
va
ti
ve
s
ys
tem
de
s
igns
,
whe
r
e
a
s
unde
r
e
s
ti
mation
may
r
e
s
ult
in
f
a
il
u
r
e
s
unde
r
a
dve
r
s
e
c
ondit
ions
[
60]
.
T
he
obs
e
r
va
ti
on
of
dB
/km
va
r
iations
withi
n
the
C
C
DF
c
ur
ve
s
plays
a
r
ole
in
s
ha
ping
the
s
y
s
tem'
s
li
nk
mar
gin
.
T
his
dir
e
c
tl
y
im
pa
c
ts
ke
y
de
s
ign
c
hoice
s
r
e
ga
r
ding
t
r
a
ns
mi
tt
e
r
powe
r
s
e
tt
ings
,
a
nte
nna
ga
in
s
e
lec
ti
ons
,
a
nd
the
e
s
tablis
hment
of
s
uit
a
ble
f
a
de
mar
gins
.
I
n
m
il
li
mete
r
-
wa
ve
pr
opa
ga
ti
on,
e
ve
n
mi
nor
dif
f
e
r
e
nc
e
s
c
a
n
ha
ve
a
noti
c
e
a
ble
im
pa
c
t
on
li
n
k
budge
t
de
s
ign
a
nd
s
ys
tem
r
e
li
a
bil
it
y
[
15]
.
T
h
e
dB
/km
dis
c
r
e
pa
nc
ies
in
the
C
C
DF
c
ur
ve
s
a
r
e
c
r
it
ica
l
f
or
de
ter
mi
ning
li
nk
mar
gins
a
nd
de
f
in
ing
s
ys
tem
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
S
N
:
2088
-
8708
I
nt
J
E
lec
&
C
omp
E
ng
,
Vol
.
15
,
No.
1
,
F
e
br
ua
r
y
20
25
:
467
-
479
476
c
ha
r
a
c
ter
is
ti
c
s
,
s
uc
h
a
s
tr
a
ns
mi
tt
e
r
powe
r
a
nd
f
a
de
mar
gins
.
T
h
is
is
r
e
leva
nt
f
o
r
c
omm
unica
ti
on
s
ys
tems
ope
r
a
ti
ng
withi
n
f
r
e
que
nc
ies
s
us
c
e
pti
ble
to
r
a
in
a
t
tenua
ti
on,
r
e
qui
r
ing
the
us
e
of
a
da
pti
ve
methods
t
o
e
ns
ur
e
c
ons
is
tent
s
e
r
vice
qua
li
ty.
(
a
)
(
b)
(
c
)
(
d)
(
e
)
(f)
F
igur
e
7.
T
im
e
pe
r
c
e
ntage
oc
c
ur
r
e
nc
e
s
a
t
38
GH
z
,
r
e
pr
e
s
e
nted
by
s
pe
c
if
ic
a
tt
e
nua
ti
on
(
γ
)
va
lues
a
c
r
os
s
dif
f
e
r
e
nt
r
a
in
types
a
nd
ti
mes
:
(
a
)
Dr
izz
le,
3:00
to
6:00
a
m,
(
b)
Dr
izz
le,
3:00
to
6:00
p
m,
(
c
)
S
howe
r
,
3:00
to
6:00
a
m,
(
d)
S
howe
r
,
3:00
to
6
:00
pm
,
(
e
)
T
hunde
r
s
tor
m,
3:00
to
6:00
a
m
,
a
nd
(
f
)
T
hunde
r
s
tor
m
,
3:00
to
6
:00
pm
Evaluation Warning : The document was created with Spire.PDF for Python.