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.
10
,
No.
4
,
A
ugus
t
2020
,
pp. 360
5
~3
614
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
4
.
pp3605
-
36
14
3605
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
Adaptiv
e quanti
zation fo
r s
p
ec
t
ru
m exch
ange inf
or
m
atio
n in
mobil
e c
og
niti
ve radi
o networks
Arief
M
arwan
to
1
,
Shari
f
ah
Ka
mi
la
h Syed
Y
us
of
2
, Muh
am
mad H
aika
l
Sa
t
ri
a
3
1
Depa
rtment of
El
e
ct
ri
ca
l
Eng
in
ee
ring
,
Univ
ersitas Isla
m
Sulta
n
Agung (UNIS
S
ULA),
Indone
si
a
2
Depa
rtment of
El
e
ct
ri
ca
l
Eng
in
ee
ring
,
Fa
cul
t
y
o
f
Engi
n
ee
ring
,
Univer
sit
y
Te
kno
l
ogi
Mal
a
y
s
ia,
M
al
a
y
si
a
3
Depa
rtment of
Biom
edi
cal
Eng
i
nee
ring
,
Univ
ersity
Te
knolog
i
M
al
a
y
si
a, Ma
lay
si
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
N
ov
6,
2018
Re
vised Ja
n 1
6, 20
20
Accepte
d
Ja
n 2
9,
2020
To
red
uce
th
e
det
ection
failu
re
of
the
exc
h
angi
ng
signal
power
onto
the
OF
DM
sub
ca
rri
er
signal
a
t
uniform
quant
iz
ation,
d
y
namic
subca
rrier
m
appi
ng
is
appl
ie
d.
Moreov
er,
t
o
addr
essing
low
SN
R’
s
wall
le
ss
tha
n
pre
-
det
ermine
thre
s
hold,
non
-
unifor
m
quant
iz
a
ti
on
or
ada
pt
ive
qu
a
nti
z
at
ion
fo
r
the
signal
quan
ti
z
at
ion
siz
e
pa
ramet
e
r
is
proposed.
μ
-
la
w
is
adopt
ed
for
ada
pt
ive
quan
tization
subca
rri
e
r
m
appi
ng
which
is
depl
o
y
e
d
in
m
obil
i
t
y
envi
ronm
ent
,
su
ch
as
Doppler
Eff
ect
and
Ra
yle
igh
Fading
pr
opaga
t
ion.
In
thi
s
works
,
sensing
node
rec
ei
v
ed
signa
l
power
the
n
sam
ple
d
int
o
a
diffe
ren
t
pola
r
ity
posi
ti
ve
and
nega
t
ive
in
μ
-
law
quant
iz
a
ti
on
and
divi
de
d
int
o
seve
r
al
seg
m
ent
at
ion
le
v
el
s
.
E
ac
h
segm
enta
ti
on
l
eve
ls
are
d
ivi
ded
int
o
seve
ral
sub
-
seg
m
ent
has
rep
re
senti
ng
one
ton
e
signal
subc
ar
rie
r
num
ber
OF
DM
which
has
the
num
ber
of
quantization
le
ve
l
and
th
e
w
idt
h
power
.
The
resul
ts
show
tha
t
b
y
usin
g
both
m
et
hods,
a
signi
ficant
diffe
ren
ce
is
obta
in
ed
aro
u
nd
8
dB
compar
ed
t
o
those
no
t
using
the a
d
apt
iv
e
m
e
thod.
Ke
yw
or
d
s
:
Ad
a
ptive
quant
iz
at
ion
Cognit
ive ra
di
o
Dynam
ic
su
bc
arr
ie
r
m
app
ing
Mu
-
la
w
Sp
ect
r
um
ex
ch
ang
e
in
form
at
i
on
Copyright
©
202
0
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
:
Ar
ie
f
Mar
wan
t
o,
Dep
a
rtm
ent o
f El
ect
rical
En
gi
neer
i
ng,
Un
i
ver
sit
y Isla
m
Su
lt
an
Agu
ng
(UNISS
ULA),
Kali
gaw
e
Stree
t,
KM
04, Sem
arang,
Central
Java,
Indo
nesi
a
.
Em
a
il
:
arief@u
niss
ula.ac.i
d
1.
INTROD
U
CTION
Mob
il
it
y
m
akes CR no
de dete
ct
s the p
ower c
hange at any ti
m
e b
ecause eve
ry
m
ov
em
ent
fo
ll
owe
d
in
the
tra
vel
dist
ance.
A
cc
ordi
ng
to
[1
–
3]
c
onduct
ing
a
r
andom
m
ov
e
m
ent
act
ivit
y
in
the
CR
N
will
do
the
un
-
co
ndit
ion
al
unif
or
m
quantiz
at
ion
.
W
he
rein
t
he
le
vel o
f
qua
ntiza
ti
on
an
d
width
is
t
he
sam
e,
j
us
t
powe
r
s
in
detect
io
n
ar
e
cha
ng
e
d
to
t
he
distance.
As
sta
te
d
in
[
4]
qu
antiz
ing
t
he
lo
cal
energy
us
i
ng
num
ero
us
sym
bo
ls
reg
i
on
al
s
ensi
ng
dete
rm
inatio
n.
T
he
m
eth
od
has
inte
nd
e
d
f
or
l
ow
SN
R
sig
nal
detect
io
n
p
r
ob
le
m
s.
An
i
nteresti
ng
propose
d
m
et
hod
has
s
how
n
in
[5
-
12]
,
the
local
dec
isi
on
sta
ti
sti
cs
in
the
fo
rm
of
log
-
li
kelihoo
d
rati
o
(LLR)
from
ind
ivid
ua
l
detect
or
s
ar
e
quanti
zed
a
nd
s
ent
t
o
the
fusio
n
ce
n
tre
(F
C)
.
T
he
sta
ti
sti
cal
prop
e
rtie
s
of
the
decisi
on
sta
ti
sti
cs
in
the
pr
ese
nce
o
f
qu
a
ntiza
ti
on
a
re
est
abl
ished
.
The
qu
a
ntize
d
decisi
on
sta
ti
sti
cs
are
sent
t
hro
ugh
a
c
hannel
that
m
ay
c
ause
er
r
or
s
.
T
he
ef
fect
of
c
hannel
error
s
is
inco
rpor
at
e
d
in
the
analy
sis
through
Bi
t
Err
or
P
robab
il
it
y
(BEP).
The
decisi
on
of
hy
po
t
hesis
te
sti
ng
from
ind
ividu
a
l sensin
g
no
de dete
ct
or
s i
n
the
f
orm
o
f
log
-
li
kelihoo
d
rati
o are qua
ntize
d
a
nd
se
nt to
t
he m
ast
er
node.
T
he
re
sul
t
of
quantiz
ed
decisi
on
will
be
se
nt
via
cha
nn
el
m
igh
t
cau
ses
er
ror
on
re
cepti
on
at
the
m
ast
e
r
node. T
he
er
r
ors
on the c
ha
nnel
can be a
naly
sis t
hro
ugh bit
error p
roba
bili
ty
(
BEP)
.
To
op
ti
m
iz
e
s
econda
ry
trans
m
it
po
wer
al
locat
ion
sp
ect
r
um
sh
arin
g,
A
cco
r
ding
to
[13
–
18]
ha
s
inv
est
igate
s
th
e
ge
ner
al
iz
e
d
Lloyds
-
t
ype
al
gorithm
(G
L
A
)
[19
–
22]
that
util
iz
ing
qua
ntize
d
c
ha
nn
el
sta
te
inf
or
m
at
ion
(C
SI
)
.
A
m
od
ifie
d
GL
A
is
co
ns
t
ru
ct
e
d
for
obt
ai
nin
g
m
axi
m
al
powe
r
c
od
e
bo
ok,
t
hat
pr
ov
e
n
to
be
gen
e
rall
y
m
er
ged
an
d
factua
ll
y
ste
ady.
Th
e
res
ults
desc
r
ibe
that
by
us
i
ng
3
-
4
bits
of
com
pen
sat
io
n,
f
ully
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.
10
, No
.
4
,
A
ugus
t
2020
:
3
6
0
5
-
3
6
1
4
3606
channel
sta
te
i
nfor
m
at
ion
c
ould
ac
hie
ve
by
m
od
ifie
d
GL
A
al
go
rithm
it
m
akes
ergo
dic
capaci
ty
ve
r
y
cl
os
e
and
use
s
on
ly
4
bits
of
feedb
ack.
H
ow
e
ve
r,
in
[
23]
has
sho
wn
that
a
go
od
-
co
ns
e
r
ve
qua
nt
ise
d
w
hich
di
m
inish
m
ean
squar
e
e
rror
of
in
div
i
dual
obse
rv
at
i
on
data'
s
LLR
a
t
qu
a
ntised
out
com
e
is
gaine
d
by
de
pl
oyin
g
Ll
oyd
schem
e.
The
quantise
d
is
des
ign
e
d
by
util
izing
P
DF
of
LL
R
of
the
se
ns
i
ng
inf
orm
ation
.
Nu
m
erical
ou
t
com
e
sh
ows
that
t
he
GL
A
al
gorith
m
giv
en
e
qu
al
sensi
ng
ac
hievem
ent
against
the
i
niti
al
LLR
te
st
by
us
in
g
a
f
e
w
qu
a
ntiza
ti
on
bi
ts.
To
a
ddressi
ng
ba
ndwi
dth
const
raints,
[23
–
25]
has
in
ve
st
igate
d
op
ti
m
i
zi
ng
quantiz
at
ion
of
LLR
sensing
data
in
cy
cl
os
ta
ti
on
ary
sig
na
l
detect
ion
us
i
ng
Ll
oyd
-
Ma
x
al
go
rithm
.
The
num
erical
resu
lt
s
pr
ese
nt
that
re
com
m
end
ed
de
sign
us
es
only
le
ss
quantiz
at
ion
bits
wh
ic
h
perform
relat
ively
the
sam
e
s
ensin
g
perform
ance
w
it
h
coarse
se
nsi
ng
obser
vatio
n,
it
’s
e
xcee
de
d
the
a
rr
a
ngem
ents
with
unif
orm
qu
antise
d.
Since
un
i
form
qu
ant
iz
at
ion
ad
opte
d
by
[2
,
23
,
26]
a
n
in
div
i
dual
sen
sin
g
does
n’
t
im
pact
to
re
duce
ba
ndwi
dth
const
raints
a
nd
rem
ai
ns
in
ov
e
r
head
i
n
con
t
ro
l
c
hannel.
Wh
e
reas,
[23]
has
s
how
n
the
re
duct
ion
of
qu
a
ntiza
ti
on
bi
ts
fo
r
sa
ving
the
ba
ndwidt
h.
Those
are
propose
d
m
et
ho
ds
is
not
co
nsi
der
in
g
non
-
un
ifor
m
qu
a
ntiza
ti
on
wh
ic
h
le
ads
t
o
ada
ptive
t
o
t
he
cha
ng
e
of
si
gn
al
fluc
tuati
on
receiv
ed
f
or
i
nd
i
vidual
SN.
More
ov
e
r,
m
os
t
of
the
pr
opose
d
m
et
ho
ds
ar
e
no
t
co
ns
i
der
i
ng
sam
pling
si
gn
al
a
nd
proce
ed
in
sta
ti
c
m
ob
il
it
y,
this is bec
om
e a ch
al
le
nge
w
he
n
se
ns
in
g n
ode is g
oing t
o
m
ob
il
e.
2.
RESEA
R
CH
SY
STE
M MO
DEL
Ex
plainin
g
r
es
earch
ch
ron
ologica
l,
incl
ud
i
ng
r
esearc
h
des
ign
,
resea
rc
h
proce
dure
(in
the
f
or
m
of
al
gorithm
s,
Pseu
doc
od
e
or
ot
her),
how
to
te
st
and
data
acq
uisit
ion
[1
,
3
,
27]
.
T
he
descr
i
ption
of
the
co
ur
se
of
researc
h
sho
uld
be
sup
port
ed
ref
e
ren
ce
s,
s
o
the
exp
la
nat
ion
can
be
acc
epted
sci
entifi
cal
ly
[4
,
27]
.
I
n
this
works,
a
sec
ondar
y
m
ob
il
e
w
irel
ess
netw
orks
(
SN)
re
pr
e
s
ent
a
sensi
ng
node
wh
ic
h
se
ns
e
d
the
P
U
s
pectral
ho
le
s
.
At
the
SN
,
the
detect
ed
powe
r
le
ve
l
is
exch
anged
(q
ua
ntize
d)
in
to
on
e
sel
ect
e
d
subcar
rier
num
ber
wh
ic
h rep
rese
nt
s the
P
U
s
pect
ru
m
inf
orm
at
io
n on S
N.
As
s
how
n
in
Figure
1
,
a
ce
ntrali
zed
net
w
ork
is
ass
um
e
d
util
iz
ed
on
the
c
ogniti
ve
r
adio
net
wor
k
topolo
gy.
The
nodes
a
re
m
ov
ed
in
ra
ndom
sp
ee
d
w
hich
t
he
vel
ocity
is
360
km
/h
aver
aged.
The
node
s
al
s
o
m
ov
e
in
ran
do
m
path;
theref
or
e
,
the
ph
a
se
ang
le
is
def
i
ne
d
to
be
ra
ndom.
The
distance
is
set
to
0.
1
km
to
10
km
wh
ic
h
the
SN
nodes
m
oving
from
the
ori
gin
place
to
wards
t
he
P
U
locat
ion
a
s
a
de
sti
nation.
The
PU
is
sta
ti
on
ary.
Sin
ce
the
m
ob
il
it
y,
path
,
a
nd
phase
a
ngle
is
r
andom
,
the
di
r
ect
ion
al
c
ou
l
d
be
rand
om
l
y
m
ov
ing
towa
rd
the
PU
or
f
arthe
r
fro
m
PU
.
Like
wise,
S
N
nodes
t
rav
el
le
d
ove
r
t
he
distances
ca
n
be
cl
os
er
or
far
the
r
from
MN lo
cat
ion
,
whe
re th
e
MN is stat
io
na
ry.
M
o
b
i
l
e
S
N
1
D
i
s
t
a
n
c
e
1
(
d
1
-
P
U
)
D
i
s
t
a
n
c
e
2
(
d
2
-
P
U
)
M
o
v
i
n
g
P
a
t
h
o
f
S
N
1
D
i
s
t
a
n
c
e
m
(
d
m
-
P
U
)
D
i
s
t
a
n
c
e
n
(
d
n
-
P
U
)
M
o
v
i
n
g
P
a
t
h
o
f
S
N
x
M
o
b
i
l
e
S
N
1
M
o
b
i
l
e
S
N
x
M
o
b
i
l
e
S
N
x
v
=
m
/
s
v
=
m
/
s
v
=
m
/
s
D
i
s
t
a
n
c
e
n
(
d
n
-
P
U
)
D
i
s
t
a
n
c
e
z
(
d
z
-
P
U
)
D
i
s
t
a
n
c
e
n
-
M
N
(
d
n
-
M
N
)
D
i
s
t
a
n
c
e
2
(
d
2
-
M
N
)
D
i
s
t
a
n
c
e
1
-
M
N
(
d
1
-
M
N
)
P
r
i
m
a
r
y
U
s
e
r
M
a
s
t
e
r
N
o
d
e
N
L
o
S
N
o
n
L
i
n
e
-
o
f
S
i
g
h
t
(
N
L
o
S
)
L
o
S
Figure
1
.
A sce
nar
i
o of
t
he
m
ob
il
e
sp
ect
r
um
ex
cha
nge in
f
orm
at
ion
[
28
]
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
Ad
ap
ti
ve
quanti
za
ti
on for spe
ct
ru
m
exch
ange
infor
m
ation
…
(
Arie
f Mar
wan
t
o
)
3607
3.
μ
-
L
AW ADA
PTIVE Q
U
A
NTIZ
ATION
ALGO
RITH
M MO
DEL
To
perform
quantiz
at
ion
of
t
he
detect
ion
pow
e
r
le
vel
,
ea
ch
S
N
nee
ds
t
o
sam
ple
the
r
ecei
ved
sig
nal
of
t
he
ra
dio
w
a
velen
gth
e
nerg
y
into
a
disc
rete
-
tim
e
sign
al
. Mo
re
ov
e
r,
a se
qu
e
nce o
f
sam
pling
could d
e
plo
y
to
sam
ple
the
el
ect
ro
m
agn
et
ic
wav
e
s.
The
m
ai
n
f
unct
ion
of
se
quence
sa
m
ple
is
to
re
duct
ion
of
co
n
t
inuous
sign
al
w
hich
va
lue
is
set
at
a
po
i
nt
in
tim
e
and
/
or
sp
ace
in
vo
l
ves
ti
m
e
per
io
dical
ly
sensing
withi
n
it
sa
m
pler.
An
e
xtracti
on
of
co
ntin
uous
sign
al
is
prov
i
ding
a
qu
a
ntiza
ti
on
le
vel
an
d
quantiz
at
ion
interval
in
sensing
process
.
T
her
e
fore,
i
n
one
pa
rtic
ular
se
ns
i
ng
proce
ss,
t
he
sam
pling
is
de
pendin
g
on
the
sam
ple
rate
of
the
sig
nal.
Mor
eov
e
r,
t
he
sam
pling
fr
e
quenc
y
or
sam
pling
rate,
f
s
,
is
the
aver
a
ge
nu
m
ber
of
sam
ples
obta
ine
d
in one
seco
nd (sam
ples p
er s
e
cond
).
Pr
e
vious
m
et
ho
d
has
s
how
n
un
-
co
ndit
i
onal
unif
or
m
qu
a
ntiza
ti
on
of
the
sp
ect
rum
exch
ang
e
inf
or
m
at
ion
.
N
ow,
to
ad
dressi
ng
the
li
m
it
a
tio
n
in
previ
ou
s
works
has
s
ho
wn
the
dynam
i
c
subcar
rier
m
app
i
ng
wh
ic
h
able
t
o
m
anag
e
intel
li
gen
tl
y
subcar
ri
er
wi
dth
acc
ordan
ce
with
t
he
conditi
ons
of
the
detect
io
n
powe
r
le
vel.
The
refo
r
e,
the
qu
a
ntiza
ti
on
res
ults
ha
ve
a
diff
e
ren
t
width
i
n
acc
ordan
ce
with
t
he
powe
r
detect
ion
f
or
each
node
.
Fig
ur
e
2
s
hows
th
e
m
echan
ism
s
of
ada
ptive
qu
antiz
at
ion
f
or
sp
ect
r
um
qu
an
ti
zat
ion
inform
at
ion
into
O
FD
M
s
ub
ca
rr
ie
r
num
ber.
μ
-
la
w
i
s
adopted
f
or
a
dap
ti
ve
quanti
zat
ion
s
ub
ca
rr
i
er
m
app
ing
ba
sed
on
the d
et
ect
e
d power
level.
D
e
t
e
r
m
i
n
e
d
I
n
p
u
t
S
i
g
n
a
l
P
a
r
a
m
e
t
e
r
s
P
o
w
e
r
M
a
x
P
o
w
e
r
M
i
n
M
a
x
I
n
t
e
r
v
a
l
R
a
n
g
e
o
f
I
n
p
u
t
S
i
g
n
a
l
I
n
t
e
r
v
a
l
L
e
n
g
t
h
(
N
c
)
R
a
n
g
e
o
f
B
o
u
n
d
a
r
i
e
s
S
i
g
n
a
l
O
b
s
e
r
v
e
d
P
U
S
i
g
n
a
l
S
N
R
D
o
p
p
l
e
r
S
h
i
f
t
e
d
V
e
l
o
c
i
t
y
/
M
o
b
i
l
i
t
y
F
r
e
q
u
e
n
c
y
D
i
s
t
a
n
c
e
W
a
v
e
l
e
n
g
t
h
,
λ
(
i
)
S
a
m
p
l
i
n
g
S
i
g
n
a
l
s
,
M
(
i
)
i
t
h
,
S
e
n
s
i
n
g
N
o
d
e
s
D
e
t
e
c
t
e
d
P
o
w
e
r
L
e
v
e
l
P
r
(
i
)
E
a
c
h
p
o
l
a
r
i
t
y
d
i
v
i
d
e
d
i
n
t
o
s
e
v
e
r
a
l
s
e
g
m
e
n
t
l
e
v
e
l
o
f
s
i
g
n
a
l
E
a
c
h
s
e
g
m
e
n
t
d
i
v
i
d
e
d
i
n
t
o
s
e
v
e
r
a
l
s
u
b
s
e
g
m
e
n
t
o
f
i
n
t
e
r
v
a
l
s
i
g
n
a
l
S
i
g
n
a
l
L
e
v
e
l
&
I
n
t
e
r
v
a
l
o
f
P
r
(
i
)
S
i
g
n
a
l
Q
u
a
n
t
i
z
a
t
i
o
n
S
i
z
e
(
μ
)
S
i
g
n
a
l
P
o
l
a
r
i
t
y
(
s
g
n
(
P
r
(
i
)
)
A
s
s
i
g
n
e
d
b
i
t
s
f
o
r
S
e
g
m
e
n
t
S
i
g
n
a
l
L
e
v
e
l
A
s
s
i
g
n
e
d
b
i
t
s
f
o
r
S
u
b
-
S
e
g
m
e
n
t
S
i
g
n
a
l
I
n
t
e
r
v
a
l
D
e
t
e
r
m
i
n
e
d
A
d
a
p
t
i
v
e
Q
u
a
n
t
i
z
a
t
i
o
n
P
a
r
a
m
e
t
e
r
s
P
r
≠
γ
C
a
l
c
u
l
a
t
e
f
o
r
C
o
m
p
r
e
s
s
a
n
d
E
x
p
a
n
d
e
r
o
f
S
e
g
m
e
n
t
L
e
v
e
l
a
n
d
S
u
b
-
S
e
g
m
e
n
t
I
n
t
e
r
v
a
l
S
e
n
d
t
o
M
N
C
o
o
p
e
r
a
t
i
v
e
D
e
c
i
s
i
o
n
D
e
t
e
r
m
i
n
e
d
S
i
g
n
a
l
P
o
l
a
r
i
t
y
P
r
(
i
)
Figure
2. A
da
pt
ive quan
ti
zat
ion al
gorithm
sp
ect
r
um
ex
cha
ng
e
in
form
at
io
n
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.
10
, No
.
4
,
A
ugus
t
2020
:
3
6
0
5
-
3
6
1
4
3608
At
the
first
st
age,
in
put
sig
nal
par
am
et
ers
sh
ould
be
det
erm
i
ned
.
Thes
e
par
am
et
ers
i
nclu
ding
of
the
m
axi
m
u
m
m
ini
m
u
m
ran
ge
of
the
detect
ion
powe
r
le
vel
(
P
max
and
P
min
),
m
axi
m
u
m
n
um
ber
of
the
i
nter
val
le
ng
th
(
IL
)
,
quantiz
at
ion
inte
rv
al
s
range
(
I
)
,
an
d
t
he
ra
ng
e
of
the
quant
iz
at
ion
inter
va
ls
range
of
the
input
sign
al
.
In
t
his
work,
the
pr
i
nciple
of
qu
a
nt
iz
at
ion
requir
es
−∞
=
P
0
≤
P
1
≤
...
≤
P
+1
≤
P
=
∞
wh
ic
h
determ
ined
ra
nges
of
detect
ion
powe
r
le
vel
in
on
e
pa
rtic
ul
ar
se
ns
in
g
pro
cess.
T
he
ra
nge
of
detect
ion
powe
r
le
vel is g
i
ven by
:
=
10
−
5
=
10
−
2
(1)
is
the
pr
e
-
de
te
rm
ined
m
ini
m
u
m
of
the
detect
ion
pow
er
ra
nge
of
t
he
in
put
sig
na
l
and
is
the m
axi
m
u
m
of
t
he
detect
io
n
po
wer
range
of
t
he
in
pu
t si
gnal
. T
he
segm
entat
ion
of
eac
h si
gn
al
is re
quired by
determ
ined
th
e
m
agn
it
ud
e
of the inte
rv
al
le
ngth
(IL) i
n
a
dvance. The
num
ber o
f
inte
rv
al
le
ng
th
is
giv
e
n by:
=
512
(2)
In
t
his
te
rm
,
the
interval
le
vel
is
rep
r
esente
d
as
subcar
rier num
ber
OFDM.
The
inter
val
le
ng
t
h
(
IL)
is
al
so
util
iz
ed
to
perform
the
quantiz
at
io
n
interval
(
I
)
for
se
gm
entat
ion
of
t
he
rece
ived
sig
nal
powe
r,
wh
ic
h give
n b
y:
=
(
+
)
(3)
=
(
10
−
5
+
10
−
2
)
512
=
1
.
9551
−
5
(4)
Ther
e
f
or
e,
the
inter
val
(
I
)
ra
ng
e
that
giv
e
for
bounda
ries
valu
e
am
on
g
qu
a
ntiza
ti
on
re
s
olu
ti
on
i
nter
va
l
of
the
detect
io
n
s
ign
al
powe
r.
The
range
of
the
in
put
sig
na
l
is
segm
ented
by
the
quanti
zat
ion
inte
rv
al
valu
e
wh
ic
h give
n b
y:
∶
∶
(5)
At
the
sec
ond
sta
ges,
the
sen
sing
res
ul
t
is
m
easur
ed
base
d
on
the
a
ve
ra
ge
po
wer
of
e
ach
S
N
no
de
us
in
g wh
ic
h
is
sam
pled
into
non u
nif
or
m
v
ar
ia
ble coder
. T
he
r
ecei
ve
d p
ower is
giv
e
n by:
−
(
)
=
+
+
(
)
−
20
10
(
4
∗
∗
(
)
∆
)
)
+
(
)
(
)
(6)
The
m
axi
m
u
m
detect
ion
pow
er
le
vel
is
sam
pled
base
d
on
t
he
qua
ntiza
ti
on
res
olu
ti
on
i
nterv
al
sig
na
l
value
ba
sed
on
the
inter
val
le
ng
th
(I
L
)
va
lue.
Eve
ry
bit
the
resu
lt
s
of
the
sa
m
pling
sign
al
is
placed
on
the
po
sit
ive
a
nd
ne
gative
pola
rity
.
Ever
y
pola
rity
bits
are
div
ide
d
into
512
se
gm
ents
or
qu
a
ntiza
ti
on
sign
al
le
vel, which
is
giv
e
n by:
(
)
=
=
⌊
(
−
min
)
⌋
(7)
=
⌊
(
10
−
2
−
10
−
5
)
1
.
9551
−
5
⌋
≅
512
(8)
A
segm
entat
ion
of
the
input
sign
al
is
def
i
ned
as
the
value
does
not
overla
p
in
the
set
of
real
nu
m
ber
s.
T
o
m
ake
the
segm
entat
ion
as
t
he
crit
erion
sta
nd
ard,
the
li
st
of
end
point
has
diff
e
re
nt
res
olu
ti
on
i
n
the v
ect
or
.
I
f
n
is t
he
le
ngth
of
the se
gm
entat
i
on
,
the
n
t
he
i
nd
ex
of
vecto
r
k
th
entry is:
0
if
sig (
k
)
≤
se
gm
e
nta
ti
on
(
1)
m
if
seg
men
t
ation
(m
)
< sig(
k
)
≤
se
gm
e
nta
ti
on
(
m+1
)
n
if
seg
me
nta
ti
on
(
n) < si
g
(
k
)
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
Ad
ap
ti
ve
quanti
za
ti
on for spe
ct
ru
m
exch
ange
infor
m
ation
…
(
Arie
f Mar
wan
t
o
)
3609
The
res
olu
ti
on
intervals
of
the
segm
entat
i
on
vecto
r
are
on
e
le
ss
tha
n
a
m
ou
nt
of
the
qu
a
ntiza
ti
on
interval.
Com
m
on
ly
,
an
-
degree
qua
ntize
r
(
)
of
a
n
input
value
i
s
denoted
by
a
set
of
qu
a
nt
iz
at
ion
degrees
=
{
1,
...,
}
a
nd
qu
a
ntiza
ti
on
bounda
ry
P
=
{P0,
.
..,
P
}.
At
the
thir
d
sta
ge,
the
se
gm
e
ntati
on
le
vels are di
vide
d
int
o 501
96 s
ub
-
se
gm
ents.
−
ℎ
=
=
(
m
ax
powe
r
∗
(
2
∗
min
power
)
)
(9)
=
(
10
−
2
∗
(
2
∗
10
−
5
)
)
1
.
9551
−
5
=
0
.
0102
(10)
Ther
e
f
or
e,
in o
ne parti
cula
r
se
ns
in
g
cy
cl
e
of
t
he dete
ct
ion p
ower
level i
s c
om
pr
ise
d
of:
=
=
0
.
01
02
(11)
=
512
0
.
0102
=
50196
segments
(12)
In
s
uch
a
wa
y,
su
b
-
segm
en
ts
interval
width
co
uld
repr
esent
subcar
rier
m
app
ing
pa
ram
et
er
and
lim
it
ed
by
se
gm
entat
ion
le
ve
l
value
as
bounda
ries.
Ba
s
ed
on
t
he
s
ub
-
se
gm
ents
interva
l
width
cal
cul
at
ion,
the
adap
ti
ve
quantiz
at
io
n
for
the
detect
ed
powe
r
le
vel
is
per
f
or
m
ed.
The
qu
a
ntiza
ti
on
of
the
detect
ion
sign
al
into s
ubcar
rier
nu
m
ber
is
giv
e
n by
:
(
)
=
(
(
)
)
.
.
∆
⌊
|
(
)
|
∆
+
1
2
⌋
(13)
The
r
esult
is
re
tur
ned
as
thir
d
ou
t
pu
t
p
aram
eter
.
Qu
a
ntiz
f
unct
ion
gen
e
rates
qu
a
ntiza
ti
on
l
evel
in
de
x
and
the
quanti
zed
ou
t
pu
t
val
ue.
This
functi
on
ex
presse
d
a
s
ind
e
x
of
the
input
sig
nal
pa
ram
et
ers
to
r
est
or
e
the
qu
a
ntiza
ti
on
le
vels
in
real
vector
sig
nal
(sig)
us
in
g
the
segm
entat
ion
par
am
et
ers.
Wh
il
e,
segm
entation
is
a
real
vecto
r
si
gn
al
that
’s
ins
erted
int
o
sp
ec
ifie
d
orde
r.
Q
ua
ntiza
ti
on
is
done
t
o
obta
ine
d
the
m
app
in
g
sign
al
y=
f(
x)
of
ar
eal
sign
al
int
o
a
di
screte
functi
on
val
ued.
T
he
basic
pri
nciple
is
to
m
ake
the
reso
l
utions
ste
ps
that
var
y
a
cco
r
ding
to
acce
ptabl
e
conditi
ons
in
the
sig
nal.
I
n
this
w
orks
,
each
sam
pling
adopted
dif
fe
ren
ce
reso
l
ution
t
hat
m
ai
ntains
quantiz
at
ion
is
use
d
to
perfor
m
qu
antiz
in
g
for
in
div
id
ual
SN
us
ers
.
The
ind
e
x
qu
a
ntiza
ti
on fo
r
se
gm
entat
ion
level (
IQL) is
giv
e
n by
:
=
(
(
)
)
.
.
∆
⌊
|
(
)
|
∆
+
1
2
⌋
(
⌊
10
(
20
)
⌋
,
,
)
(14)
w
he
re
SL
is se
gm
entat
ion
lev
el
ind
e
x,
SSW
i
s sub
-
segm
entat
ion
inte
rv
al
sign
al
.
The
ne
xt
sta
ge
is
determ
ined
adap
ti
ve
qua
nt
iz
at
ion
pa
ram
e
te
rs.
T
he
par
a
m
et
e
rs
are
qu
a
ntiza
ti
on
siz
e
m
app
in
g
pa
ra
m
et
ers
fo
r
m
easur
em
ent
val
ue
of
the
i
nd
e
x
sig
nal
le
vel
and
i
nter
val
w
idth
of
rec
ei
ve
d
po
wer,
sign
al
po
la
rity
of
receive
d
power,
ass
ign
e
d
bit
for
segm
entat
ion
le
vel,
a
nd
an
assi
gn
e
d
bit
for
su
b
-
se
gm
entat
i
on
le
vel
.
The
a
i
m
’s
is
to
m
a
ke
the
dy
nam
i
c
range
(
diff
e
r
ent
width
of
s
egm
entat
ion
le
vel
of
the sig
nals)
of
the in
pu
t
sig
na
l powe
r.
The
pro
cess
of
com
pan
ding
(
com
pr
essio
n
and
e
xp
a
ndin
g)
is
us
ed
to
ov
e
r
com
e
po
or
of
SN
R
le
vel
or
to
com
pr
ess
the
si
gn
al
m
ore
tha
n
pr
e
-
det
erm
ined
thres
ho
l
d.
The
co
m
pan
din
g
is
aim
ed
to
disti
ngui
s
h
the
de
ns
it
y
of
the
qua
ntiza
ti
on
sig
nal
ada
ptively
,
if
the
det
ect
ion
S
NR
is
poor,
qu
a
ntiza
ti
on
le
vels
at
lowe
r
sign
al
powe
r
m
us
t
be
den
se
r
tha
n
the
sig
nal
po
wer
at
a
higher
le
vel.
If
sig
nals
det
ect
ion
is
high
er
tha
n
thres
ho
l
d,
the
n
the
qua
ntiza
ti
on
le
vel
at
hig
he
r
le
vel
sh
ould
be
de
ns
er
tha
n
lo
w
le
vel
sign
a
l
po
we
r
.
The q
uan
ti
zat
ion o
f
t
he
se
gme
ntati
on
sig
nal
le
vel and i
nter
val is a
dap
ti
vel
y chan
ges
b
y
:
(
)
=
(
(
)
)
=
(
(
)
)
(
ln
(
1
+
|
(
)
|
)
ln
(
1
+
)
)
(15)
w
he
re,
0
≤
(
)
≤
1
;
(
(
)
)
is
-
1
if
(
)
ne
gative
a
nd
1
ot
herwise,
µ
is
512
bits
w
hich
represe
nt
as
qu
a
ntiza
ti
on
m
easur
em
ent
siz
e/
width
par
am
et
er
for
eac
h
s
egm
entat
ion
le
vel
an
d
is
detect
ed
powe
r
le
vel
each S
N n
od
e
.
The
c
om
pan
di
ng sig
nal
value
(
CSV
)
r
e
fer
e
nc
e sig
nal is g
i
ve
n by:
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.
10
, No
.
4
,
A
ugus
t
2020
:
3
6
0
5
-
3
6
1
4
3610
(
)
=
(
(
)
)
(
ln
(
1
+
|
(
)
|
)
ln
(
1
+
)
)
(
⌊
10
(
20
)
⌋
,
,
)
(16)
By
find
i
ng
no
n
-
ze
ro
el
em
ent
value
in
row
and
col
um
n
within
in
de
xing
value,
the
qua
ntiza
ti
on
int
o
su
bc
ar
rier
nu
m
ber in
dex g
i
ve
n by:
=
=
∑
=
1
=
(
1
+
)
2
>
(
(
)
)
(17)
(
ln
(
1
+
|
(
)
|
)
ln
(
1
+
)
)
(
⌊
10
(
20
)
⌋
,
,
)
w
he
re
Nc
is
th
e
subca
rr
ie
r
in
dex
of
the
ref
e
ren
ce
sig
nal;
IL
is
inte
rv
al
le
ng
t
h
by
giv
e
n
512;
P
m
ax
is
m
axim
u
m
detect
ion PU
pow
e
r,
10
-
2
.
4.
MOBILE
S
P
ECTR
UM E
XC
H
ANGE I
N
FOR
M
A
TIO
N
BA
SED
CO
OPER
ATIVE
SENSIN
G
All
the
com
po
nen
ts
par
am
et
e
r
has
be
en
m
od
el
ed
as
sta
te
d
in
pr
e
vious
subcha
pter
.
I
n
this
par
t,
e
very
m
ob
il
e SN
’s n
od
e
s w
il
l recei
ves
s
om
e d
ist
ribu
te
d powe
r
f
r
om
p
rim
ary us
ers (PU
), w
hic
h
is
giv
e
n by:
−
(
)
=
+
+
(
)
−
20
10
(
4
∗
∗
(
)
∆
)
)
+
(
)
(
)
(18)
w
he
re t
he
(
)
is R
ay
le
igh
f
a
ding
base
d on su
m
m
ing
sin
usoids
w
it
h Jakes
m
od
el
[1
1].
(
)
=
1
√
∑
cos
(
2
+
)
=
1
(19)
a
nd
(
)
=
1
√
∑
sin
(
2
+
)
=
1
(20)
a
nd
(
)
=
(
)
+
(
)
(21)
is
the
D
oppler
sh
ift,
=
is
the
am
pl
it
ud
e
of
th
e
sign
al
a
nd
N
is
m
ulti
path
co
m
po
nen
ts
wit
h
ang
le
of
a
rr
i
va
l
of
the
node
s.
The
detect
ed
pr
im
ary
sign
al
po
wer
the
n
e
xc
hange
d
i
nto
s
ub
ca
rr
ie
r
num
ber
of
OFDM
know
as
qua
ntiza
ti
on
powe
r
pro
cess.
T
he
c
on
ven
ti
onal
quan
ti
zat
ion
m
app
ing
of
t
he
s
pec
trum
exch
a
nge in
f
orm
at
ion
at
ℎ
SN
giv
e
n by
[
1]
.
(
)
=
⌊
(
)
∗
⌋
(22)
k
is
ass
um
ed
is
the
f
unct
ion
of
po
wer
qua
ntiza
ti
on
at
the
f
requ
ency
car
rier
,
.
I
f
a
ssu
m
ing
that
the
=
;
is t
he fu
nc
ti
on
of freq
ue
ncy in
D
oppler
Effect
which
gi
ven
by:
=
+
∆
(23)
If
detect
ed
power
is
a
func
ti
on
of
f
re
qu
e
ncy
then
t
he
k
is
a
functi
on
of
f
re
qu
e
nc
y,
howe
ver,
a
m
ov
em
ent
and
im
pairm
ent
of
t
he
cha
nne
l
cou
l
d
rise
duri
ng
the
se
ns
in
g
pr
ocess,
t
herefo
re,
t
he
rece
ive
d
fr
e
qu
e
ncy
of
new
s
ubcar
rier
1
′
is
sh
ifte
d
duri
ng
m
ob
il
it
y
in
∆
sp
eed
,
w
hich
re
su
lt
ed
i
nto
∆
area.
Ther
e
f
or
e,
′
is g
iven by:
′
=
(
+
∆
)
−
=
1
±
[
∆
]
(24)
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
Ad
ap
ti
ve
quanti
za
ti
on for spe
ct
ru
m
exch
ange
infor
m
ation
…
(
Arie
f Mar
wan
t
o
)
3611
w
he
re
is
sub
carrier
width
fr
e
qu
e
ncy
i
n
Her
tz
;
H
ow
e
ve
r,
due
t
o
sig
nal
is
non
-
li
ne
ar
t
her
e
fore
is
disti
nction i
n v
aries,
hen
ce:
′
=
[
1
±
∆
]
∗
(25)
w
he
re
the m
axi
m
u
m
r
ece
ived
f
re
quency
D
opple
r
s
hifted
at the
n
th
se
ns
in
g n
od
e
s
[
29]
,
if
≫
∆
then
′
=
(26)
and
∆
f
R
(
i
)
=
f
R
−
f
c
(27)
5.
RESU
LT
S
AND A
N
ALYSIS
In
this
te
rm
,
m
ob
il
e
sp
ect
rum
exch
an
ge
(
qu
a
ntiza
ti
on)
i
s
eval
uated
wi
th
tw
o
dif
fer
e
nt
m
et
ho
ds
.
The
first
m
et
ho
d
is
sp
ect
rum
exch
ang
e
inf
or
m
at
ion
in
m
ob
il
i
ty
en
vir
on
m
ent
base
d
on
[
1,
30]
pr
opos
e
d
m
et
ho
d.
The
s
econd
m
et
ho
d
is
m
ob
il
e
sp
ec
trum
exch
an
ge
inf
or
m
at
ion
ut
il
iz
ing
Dop
pler
E
f
fect
pa
ra
m
et
er.
Th
os
e
m
et
ho
ds
will
inv
est
igate
in
te
rm
of
a
da
ptive
quantiz
at
io
n
as
util
iz
ed.
The
aim
s
are
to
exp
l
ore
the
pe
rfor
m
ances
of
the
c
ooper
at
ive
decisi
on
pro
ba
bili
ty
at
the
gath
eri
ng
i
nfor
m
at
ion
node.
I
n
this
work,
the
s
ub
ca
rr
ie
r
m
app
in
g
par
a
m
et
er
in
te
rm
is
fix
ed
,
wh
e
re
the
s
ub
ca
rr
ie
r
m
app
in
g
is
gi
ve
n
by
α
=2
.
T
he
refor
e
,
the
equat
ion
(
22)
sho
uld
util
iz
e
to
per
f
or
m
the
qu
a
ntiza
ti
on
detect
ion
powe
r
le
vel
into
subcar
rier
num
ber
.
In
a
ddit
ion
,
t
he
m
et
ho
d
that
util
iz
ed
Dop
pl
er
Eff
ect
is
ut
il
iz
ed
in
(2
5),
wh
e
re
α
is
sta
ti
on
ary.
In
te
rm
of
adap
ti
ve
qu
a
nt
iz
at
ion
,
the
subcar
rier
pa
ram
et
er
is
based
on
the
in
(22)
f
or
co
nventio
na
l
m
e
tho
d;
an
d
in
(25
)
for
the
m
et
ho
d
that
util
iz
ed
D
oppler
E
ff
ect
.
Exce
pt
at
the
la
st
su
b
-
cha
pter
,
the
ada
ptive
qu
a
ntiza
ti
on
m
et
hod
is
evaluated
by
us
in
g
dy
nam
i
c
su
bca
rr
ie
r
m
app
i
ng
for
c
om
par
at
ive
study
between
c
on
ven
ti
onal
m
et
h
od
s
t
hat
util
iz
ed
dy
nam
ic
subcar
rier m
app
i
ng.
5.1.
An
analysis
of
spect
r
um e
xc
ha
n
ge
in
f
orm
at
i
on
u
sing
ad
ap
ti
ve qu
an
tiz
at
i
on
Figure
3
sho
w
s
the
pe
rfor
m
ance
of
detect
ion
pro
ba
bili
ty
i
n
m
ob
il
e
sp
ect
ru
m
exch
a
ng
e
inf
or
m
at
ion
.
In
this
sta
ge,
two
m
et
ho
ds
a
r
e
propose
d
to
be
eval
uated.
I
n
[
30
]
m
et
ho
d
is
assum
ed
that
Dopp
le
r
E
ffec
t
is
zero
(
Ol
d
K
O
HT
A
)
an
d
wh
i
le
the
oth
e
r
m
et
hod
is
bas
ed
on
t
he
perfor
m
ance
Oh
ta
m
od
el
us
in
g
Dopp
le
r
Eff
ect
(
D
opple
r
K
)
.
Both
m
e
t
hods
al
s
o
are
e
valuated
by
usi
ng
a
dap
ti
ve
quantiz
at
io
n
(
M
u
Ol
d
K
OHT
A
and
Mu
D
oppler
K
).
T
her
e
fore,
f
our
m
et
ho
ds
a
r
e
analy
sis
to
pe
rfor
m
cooper
at
ive
detect
ion
in
MN.
T
he
ve
locit
y
of the
nodes
is
set
s for
360 k
m
/h w
hic
h
th
e
directi
onal
p
at
h
a
nd m
ov
ing i
s r
a
ndom
iz
ed.
Fr
om
the
gr
ap
h
it
is
cl
ear
tha
t
the
evaluati
on
analy
sis
for
the
first
m
et
ho
d
is
m
ark
ed
by
blu
e
li
ne
is
aver
a
ge
d
at
the
po
int
on
0.99.
Howe
ver,
at
hig
h
le
vel
of
s
ubcarrier
th
res
ho
l
d
val
ue,
the
de
te
ct
ion
is
decr
e
ase
d
to
0.9
08.
Wh
il
e
util
iz
ing
D
op
pler
E
ff
ect
m
eth
ods
w
hich
m
ark
e
d
by
re
d
c
ircl
e
li
ne
are
a
l
m
os
t
the
sam
e
with
the
pe
rfor
m
ance
of
t
he
m
et
ho
d
that
unutil
iz
ed
D
opple
r
Ef
fe
ct
.
As
ca
n
be
s
een
al
so
on
the
gr
a
ph,
th
e
m
eth
od
s
wh
ic
h
are
util
iz
ing
both
the
Dop
pler
Ef
fect
and
the
ada
ptive
quantiz
at
io
n
m
e
tho
d
hav
e
per
f
or
m
ed
the
sa
m
e
le
vel
of
detect
ion
prob
a
bili
ty
.
It’s
s
how
n
by
the
gr
ee
n
li
ne
an
d
re
d
ci
r
cl
e
li
ne
on
th
e
gr
a
ph.
H
ow
ever,
us
es
b
ot
h
m
eth
ods
wer
e
sli
ghtl
y
bette
r
pe
r
form
ance
oth
e
r
tha
n
un
util
ized
ada
ptive
qu
antiz
at
ion
,
w
he
nev
e
r
the d
et
ect
io
n o
f
s
ub
ca
rr
ie
r
th
r
esh
old
value
gr
eat
er th
a
n 8 dB
.
In
c
om
par
ison
per
f
orm
ance,
can
cl
early
be
seen
betwee
n
[1]
m
et
ho
d
w
hich
util
iz
ed
or
unutil
iz
e
d
the
ada
ptive
quantiz
at
io
n,
th
e
detect
ion
prob
a
bili
ty
was
sli
gh
tl
y
lowe
r
than
m
et
ho
d
t
hat
util
iz
ed
D
opple
r
Eff
ect
a
nd
a
da
ptive
qua
ntiza
ti
on
.
It
is
co
rr
es
ponde
nt
to
the
aver
a
ge
of
the
sel
ect
ed
subca
rr
ie
r
powe
r
thres
ho
l
d,
the
m
et
ho
d
that
util
iz
ed
the
Dop
pler
E
ff
ect
an
d
adap
ti
ve
quan
ti
zat
ion
has
de
te
ct
ed
belo
w
the
pre
-
determ
ined
the
detect
ion
th
re
sh
ol
d
r
ules.
T
her
e
fore,
t
he
pro
ba
bili
ty
can
perform
bette
r
than
unutil
iz
ed
both
m
et
ho
ds.
As
il
lustrate
d
by
the
grap
h,
F
igure
4
s
how
s
the
pro
ba
bili
ty
of
false
al
arm
that
evaluate
s
the
sp
ect
r
um
e
xch
a
nge
inf
or
m
at
ion
is
pr
es
ented.
As
ca
n
be
seen
from
t
he
grap
h,
t
he
conve
ntion
al
m
et
ho
d
(
Ol
d
K
O
HT
A
)
has
the
sam
e
pro
bab
il
it
y
of
f
al
se
al
arm
per
f
or
m
ance
com
par
e
to
the
m
et
ho
d
w
hich
is
util
iz
ed
Dop
pler
E
ff
ect
.
H
ow
e
ver,
at
the
high
le
vel
su
bc
ar
rier
dete
ct
ion
th
reshold
value
sta
rt
f
r
om
8
dB,
the
false
al
arm
beco
m
e
r
os
e
to
0.0
92
.
O
n
oth
e
r
ha
nd,
the
m
et
ho
d
that
util
iz
ed
the
Dop
pler
Eff
ect
a
nd
a
da
ptive
qu
a
ntiza
ti
on
pe
rfor
m
lowest
oth
e
r
tha
n
co
nventio
nal
m
et
ho
ds.
T
he
sam
e
thing
is
s
how
n
in
the
grap
h,
whe
n
the
subca
rr
ie
r
detect
ion
t
hr
es
ho
l
d
le
vel
at
po
sit
io
n
8
dB
,
it
was
increa
sed
false
al
a
r
m
,
bu
t
s
m
al
ler
tha
n
conve
ntion
al
m
et
ho
d, le
ss th
an 0.
08.
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.
10
, No
.
4
,
A
ugus
t
2020
:
3
6
0
5
-
3
6
1
4
3612
Figure
3.
A
n
a
naly
sis
of
detect
ion
pro
bab
il
it
y from
the m
ob
il
e sp
e
ct
ru
m
ex
cha
nge inform
at
ion
t
hat
util
iz
e o
r u
nu
ti
li
zed ad
a
ptive
qu
a
ntiza
ti
on
Figure
4
.
A co
m
par
at
ive an
al
ysi
s stud
y
of f
a
lse
alarm
pro
bab
il
it
y for
sp
ect
r
um
ex
ch
ang
e
in
form
at
i
on
util
iz
ing
or
un
util
iz
ed
ada
ptive
q
ua
ntiza
ti
on
6.
CONCL
US
I
O
N
It
can
be
s
how
n
that
the
pe
rfo
rm
ance
of
ada
ptive
quantiz
at
ion
f
or
sp
ect
rum
exch
an
ge
in
form
ation
i
n
OFDM
CR
networks
is
pr
ese
nted.
A
com
par
at
ive
stud
y
al
so
prese
nted
by
co
m
par
e
the
per
f
orm
ance
of
tw
o
adap
ti
ve
m
et
ho
ds
in
sp
ect
r
um
ex
chan
ge
in
form
ation
.
B
ot
h
of
a
da
ptive
m
et
ho
ds
base
d
on
the
detect
ion
po
wer
le
vel
wh
ic
h
i
s
aim
to
ob
t
ai
n
dif
fer
e
nt
width
of
qua
ntiza
ti
on
sig
na
l
le
vel.
As
pr
e
viously
m
e
ntion
e
d,
the
adap
ti
ve
quantiz
at
io
n
is
able
to
perf
orm
well
rather
than
co
nventi
onal
sp
ect
r
um
e
xch
a
nge
inf
orm
at
ion
.
W
he
reas,
util
iz
ing
dynam
ic
su
bc
ar
rier
m
a
pp
i
ng
ha
ve
th
e
sam
e
per
for
m
ance
with
a
dap
ti
ve
qua
ntiza
ti
on
.
More
ov
e
r,
the
com
bin
ing
of
dynam
ic
su
bcar
rier
m
app
in
g
and
a
da
ptive
quantiz
at
io
n
ha
ve
obta
ine
d
th
e
best
perform
ance
i
n
sp
ect
r
um
exch
an
ge
in
form
at
ion
rathe
r
th
an
if
only
dynam
ic
su
bcarri
er
m
app
ing
or
on
ly
adap
ti
ve
qua
ntiza
ti
on
.
F
or
th
e
fu
t
ur
e
w
ork
s,
the
c
om
bin
ing
of
a
dap
ti
ve
m
e
tho
ds
co
uld
be
de
plo
y
ed
i
n
com
plexity
mo
del
to
in
vestigat
ed
sync
hro
nizat
ion
delay
,
tim
e
delay
an
d
subcar
rier
bi
t
synchron
iz
at
ion
t
o
com
ba
t
the
de
la
y
pr
opa
gatio
n
ef
fect
withi
n
re
portin
g
c
ha
nn
el
.
I
n
a
f
ut
ur
e
,
an
e
xperi
m
ent
and
te
sti
ng
a
re
require
d
to
obt
ai
n
the a
ct
ual
pe
rfor
m
ance tha
t can
be
a
pp
li
e
d
in
the
real
w
or
l
d.
ACKN
OWLE
DGE
MENTS
This
re
searc
h
was
s
upporte
d
by
Mi
nistry
of
Highe
r
E
duc
at
ion
of
I
ndonesi
a
(
Kem
enRist
ekD
ikti
)
unde
r
Wo
rl
d
Cl
ass
Pr
ofess
or
(20
18)
P
r
ogram
m
e
20
18,
Un
i
ver
sit
y
Is
lam
Su
lt
an
A
gung
(
U
NI
S
S
ULA),
Un
i
ver
sit
y Te
knol
og
i M
al
ay
sia
(UTM).
REFERE
NCE
S
[1]
M.
Ohta,
T
.
Fuji
i,
K.
M
ura
ok
a,
a
nd
M.
Ari
y
oshi
,
“
An
O
FD
M
bas
ed
sensing
infor
m
at
ion
exc
h
ang
e
for
coope
r
ativ
e
sensing
in cogni
t
ive
r
adi
o
s
y
st
ems
,
”
Proc
.
SD
R20
08
,
2008
.
[2]
M.
Ohta,
T
.
Fuji
i,
K.
Muraoka
,
a
nd
M.
Ari
y
oshi,
“
A
Novel
Pow
er
Controlled
Se
nsing
Inform
at
io
n
Gathe
r
ing
for
C
ooper
ative
Sen
sing
on
Share
d
Spect
rum
with
Prim
ar
y
Spect
ru
m
,
”
in
Proce
ed
i
ngs
of
the
6th
Inte
rnational
ICS
T
Confe
renc
e
on
C
ognit
ive Radio
Or
ie
nte
d
W
irele
ss
Net
works and Com
municat
ions
,
pp
.
111
–
115
,
2
011
.
[3]
H.
Uchi
y
ama
et
al.
,
“
Stud
y
on
soft
dec
ision
b
ase
d
coope
ra
ti
ve
sen
sing
for
cogni
ti
v
e
rad
io
n
et
works
,
”
IEI
C
E
Tr
ans.
Comm
un.
,
vol
.
9
1,
no
.
1
,
pp
.
95
–
101,
2008
.
[4]
L.
Chen
,
J.
W
ang,
S.
Li,
“
Coopera
ti
v
e
spec
trum
sensing
with
m
ult
i
-
bi
ts
loc
a
l
sen
sing
dec
isions
in
cogni
ti
v
e
rad
io
cont
ex
t,
”
in
Wir
el
ess Comm
unic
ati
ons and
N
et
w
orking
Conf
ere
n
ce
,
2008.
WCNC
2008.
IEEE
,
pp.
570
–
575
,
2008
.
[5]
A.
Marwant
o,
S.
K.
S
y
ed
Y
usof,
and
M.
H.
Satri
a
,
“
Orthogonal
Freque
nc
y
-
Div
ision
Multi
pl
exi
ng
-
B
ase
d
Coopera
t
ive
Spect
rum
Sensing
for
Cognit
ive
Radi
o
Networks,”
TEL
KOMNIKA
(
Tele
c
omm
unic
ati
on
Comput
er
El
e
ct
ron
ic
s and
Control
)
,
vol. 12
,
no
.
1
,
p
p
.
143
-
152
,
Mar
.
2014
.
[6]
H.
A.
Surawe
er
a,
P.
J.
Sm
it
h,
a
nd
M.
Shafi
,
“
C
apa
c
ity
li
m
it
s
a
nd
per
form
ance
anal
y
s
is
of
cog
nit
ive
rad
io
wi
th
imperfe
ct c
h
ann
el
knowl
edge,”
I
EE
E
Tr
ans.
Ve
h
.
Technol
.
,
vol
.
5
9,
no
.
4
,
pp
.
181
1
–
1822,
2010
.
[7]
A.
Sahai
,
N.
Hoven,
and
R.
T
andr
a,
“
Som
e
f
undamenta
l
l
imits
on
cogni
ti
ve
rad
io,”
in
Al
l
ert
on
Confe
renc
e
on
Comm
unic
ati
on,
Control, and
Co
mputing
,
pp
.
166
2
–
1671
,
2004
.
[8]
Y.
Ta
ni
and
T.
Saba,
“
Quanti
z
a
ti
on
sche
m
e
for
ene
rg
y
de
te
c
tor
of
soft
dec
ision
coope
ra
ti
ve
spe
c
trum
sensing
in
cogni
ti
v
e
r
adi
o
,
”
in
2010
IE
EE Globe
com
Worksh
ops
,
pp.
69
–
73
,
2010
.
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
Ad
ap
ti
ve
quanti
za
ti
on for spe
ct
ru
m
exch
ange
infor
m
ation
…
(
Arie
f Mar
wan
t
o
)
3613
[9]
K.
Huang
and
R.
Zha
ng,
“
Coope
rat
iv
e
fee
db
ac
k
for
m
ult
ia
nte
nn
a
cogni
ti
v
e
rad
io
net
works
,
”
IE
E
E
Tr
ans.
Signa
l
Proce
s
s.
,
vol. 59
,
no
.
2
,
pp
.
747
–
758,
2010
.
[10]
Z.
T
ia
n
,
E. Bl
asc
h,
W
.
Li,
G.
Che
n,
X.
Li,
“
Perfor
m
anc
e
ev
al
ua
ti
o
n
of
distri
but
ed
c
om
pre
ss
ed
wideba
nd
sensing
for
cogni
ti
v
e
r
adi
o
n
et
works
,
”
in
200
8
11th
In
te
rnatio
nal
Conf
ere
nce
on
Information Fusion
,
pp
.
1
–
8
,
2
008
.
[11]
A.
Ta
h
erp
our,
M.
Nasiri
-
Kena
ri,
A.
Jam
shidi,
“
Eff
icien
t
co
oper
ative
spe
ct
r
um
sensing
in
cogni
t
ive
r
adio
net
works
,
”
in
20
07
IEEE
18th
In
te
rnational
Sym
posium
on
Pe
rs
onal,
Indoor
an
d
Mobile
Rad
io
Comm
unic
ati
ons
,
pp.
1
–
6
,
2007
.
[12]
S.
Chaudha
r
i
an
d
V.
Koivunen
,
“
Eff
ect
of
qu
a
nti
z
at
ion
and
ch
anne
l
err
ors
on
col
l
abor
ative
sp
ec
trum
sensing
,
”
in
2009
Confe
r
enc
e
Record
of
the
Fort
y
-
Thir
d
Asil
omar
Confe
renc
e
on
Si
gnals,
Syste
ms
and
Computers
,
pp.
528
–
533
,
20
09
.
[13]
X.
W
ang,
Z.
Li,
P.
Xu,
Y.
Xu,
X.
Gao,
and
H
.
-
H.
Chen,
“
Spect
rum
sharing
in
cogni
ti
ve
rad
i
o
net
works
—
An
auc
t
ion
-
base
d
ap
proa
ch,”
I
EEE
T
rans
.
Syst. Man,
Cybe
rn. Part
B
,
vol.
40
,
no
.
3
,
pp
.
587
–
596
,
2009
.
[14]
D.
Ni
y
a
to
and
E.
Hos
sain,
“
A
game
-
the
ore
t
ic
appr
oac
h
to
c
om
pet
it
ive
spe
c
trum
sharing
i
n
cogni
ti
v
e
rad
i
o
net
works
,
”
in
20
07
IEEE
W
irel
es
s Comm
unic
ati
o
ns and
Net
work
i
ng
Confe
ren
ce
,
pp.
16
–
20
,
2007
.
[15]
X.
Kang,
Y.
-
C
.
Li
ang
,
H.
K.
Ga
rg,
and
L.
Zha
n
g,
“
Sensing
-
base
d
spec
trum
shari
ng
in
cognitive
r
adi
o
ne
tworks,”
IEE
E
Tr
ans.
Ve
h
.
Techno
l.
,
vo
l.
5
8,
no
.
8
,
pp
.
464
9
–
4654,
2009
.
[16]
D.
Ni
y
at
o
and
E
.
Hos
sain,
“
Compe
ti
t
ive
pri
ci
ng
f
or
spec
trum
sharing
in
cognitive
rad
io
net
works
:
D
y
namic
g
ame
,
ine
ffi
ci
en
c
y
of
n
ash
equilibri
um
,
and
co
ll
usion,
”
I
EE
E
J
.
Se
l. area
s Comm
un.
,
vol
.
26,
no
.
1
,
pp
.
19
2
–
202,
2008
.
[17]
D.
Ni
y
at
o
and
E
.
Hos
sain,
“
Compe
ti
t
ive
spe
ct
ru
m
sharing
in
cog
nit
ive
rad
io
net
w
orks:
a
d
y
namic
game
appr
oa
ch,”
IEE
E
Tr
ans.
Wir
el
.
Comm
un.
,
vo
l.
7
,
no
.
7
,
pp
.
26
51
–
2660,
2008
.
[18]
R.
Zha
ng
and
Y.
-
C.
Liang,
“
Expl
oiting
m
ult
i
-
antenna
s
for
opportun
isti
c
spe
ct
rum
sharing
in
cogni
ti
v
e
rad
i
o
net
works
,
”
IEEE
J. Sel. Top.
S
ign
al
Proc
ess.
,
vo
l.
2,
no
.
1
,
pp
.
88
–
102,
2008
.
[19]
C.
Senth
amara
i
and
N.
Ma
lmurugan,
“
Optimal
Pow
er
Allocati
on
for
Outage
Anal
y
sis
Cogni
ti
ve
Full
Dupl
ex
Mult
i
-
Channe
l
R
el
a
y
Network
S
y
stems
Us
ing
Multi
ple
Objec
t
ive
Opti
m
iz
at
ion
Te
chn
i
que,
”
J.
Comput
.
Theor.
Nanosci
.
,
vol.
15
,
no
.
3
,
pp
.
1004
–
1011
,
20
18.
[20]
Y.
Y.
He,
“
Topi
cs
in
resourc
e
opti
m
iz
ation
in
wire
l
ess
net
works
with
li
m
it
ed
feedbac
k,
”
PhD
the
sis,
Engi
nee
r
ing
,
El
e
ct
ri
ca
l
and
E
l
ec
tron
ic
Engi
n
eering,
The Unive
r
sit
y
of
Me
lbourn
e
,
2011
.
[21]
J.
C.
F.
Li,
W
.
Z
hang,
and
J.
Yu
a
n,
“
Opportunistic
spec
trum
shari
ng
in
cogn
it
iv
e
r
adi
o
n
et
works
b
ase
d
on
pr
imar
y
li
m
it
ed
fee
db
ac
k
,
”
I
EEE
Tr
ans. Com
mun.
,
vol
.
5
9,
no
.
12
,
pp
.
32
72
–
3277,
2011
.
[
22]
Y.
He
and
S.
De
y
,
“
Pow
er
al
loc
a
ti
on
in
spe
c
trum
sharing
cogni
ti
ve
r
adi
o
n
e
tworks
with
quant
i
ze
d
ch
anne
l
informati
on,
”
I
E
EE
Tr
ans.
Com
mun.
,
vo
l. 59, no
.
6
,
pp
.
1644
–
16
56,
2011
.
[23]
N.
Ngu
y
en
-
Thanh
and
I.
Koo,
“
Log
-
li
kelihoo
d
rat
io
opt
imal
quant
i
ze
r
for
coope
ra
ti
ve
spe
ct
rum
sensing
in
cogni
ti
v
e
r
adi
o
,
”
IEEE
Comm
un.
Lett
.
,
vol
.
15,
no
.
3
,
pp
.
317
–
319
,
2011.
[24]
S.
Bokharaie
e
,
H.
H.
Ngu
y
en
,
and
E
.
Shw
ed
y
k
,
“
Bli
nd
spe
c
trum
sensing
for
OF
DM
-
base
d
cogni
t
ive
rad
i
o
s
y
stems
,
”
I
EEE
Tr
ans.
Ve
h.
Tec
hnol.
,
vol
.
60
,
no
.
3
,
pp
.
858
–
871
,
2011.
[25]
N.
Ngu
y
en
-
Tha
n
h
and
I.
Koo,
“
A
cl
uster
-
b
ase
d
se
le
c
ti
ve
coope
r
at
i
ve
spec
trum
sen
sing
sche
m
e
in
c
ognit
ive
rad
io
,
”
EURA
SIP
J
.
W
irel
.
Comm
un.
Ne
t
w.
,
vo
l. 2013, no
.
1
,
pp
.
1
–
9
,
201
3.
[26]
S.
Chaudha
ri
,
J.
Lundé
n,
V.
Koi
vunen,
and
H.
V.
Poor,
“
Coop
era
t
ive
sensing
with
imperfe
c
t
rep
orti
ng
ch
annel
s:
Hard
de
ci
sions o
r
soft
de
ci
sions
?
,
”
IE
EE Tr
ans.
Si
gnal
Proc
ess.
,
2
012.
[27]
M.
Ohta,
T
.
Fuji
i,
K.
Muraoka
,
a
nd
M.
Ari
y
oshi,
“
A
Novel
Pow
er
Controlled
Se
nsing
Inform
at
io
n
Gathe
r
ing
for
Coo
per
ative
Sen
sing
on
Share
d
Spect
rum
with
Prim
ar
y
Spect
ru
m
,
”
2011
6th
Inte
rnational
ICS
T
Confe
renc
e
o
n
Cognit
ive Radio
Or
ie
nte
d
W
irele
ss
Net
works and Com
municat
ions (
CROWNCOM
)
,
pp
.
111
–
115
,
2
011.
[28]
A.
Marwant
o,
S.K.
S
y
ed
Y
usof,
and
M.
H
.
Satria,
“
Orthogon
al
Frequ
en
c
y
-
Div
ision
M
ult
iplexing
-
B
ase
d
Coopera
t
ive
Spect
rum
Sensing
for
Cognit
ive
Radi
o
Networks,”
TEL
KO
MNIKA
(
Tele
communic
ati
on
Compute
r
El
e
ct
ron
ic
s and
Control
)
,
vol. 12
,
no
.
1
,
p
p
.
143
,
Mar.
2014
.
[29]
F.
Xiong
and
M
.
Andro,
“
The
Eff
ec
t
of
Doppler
Fre
quency
Shif
t,
Freque
nc
y
Off
set
of
the
Loc
al
Os
ci
ll
at
o
rs,
and
Phase
Noise
on
t
he
Perform
ance of Cohe
ren
t
OF
DM
Rec
ei
v
ers,
”
NASA
S
ci
.
Tech
.
Inf.
,
vol
.
2
,
no
.
2
,
pp
.
1
–
4
,
2001
.
[30]
M.
Ohta,
T.
Fu
ji
i,
K.
Muraok
a,
and
M.
Ari
y
oshi,
“
A
Coopera
t
ive
Sensing
Area
Ex
pansion
Us
ing
Orthogonal
Freque
nc
y
Base
d
Inform
at
ion
Coll
e
ct
ion
Metho
d,
”
2009
4th
Int
e
rnational
Confe
r
enc
e
on
Cogni
ti
v
e
Radi
o
Or
ie
nt
e
d
Wirel
ess Networ
ks
and
Comm
uni
cat
ions
,
2009.
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
Arie
f
Ma
r
w
ant
o
,
recei
ved
h
is
Master
and
PhD
in
E
le
c
tri
c
al
Engi
ne
eri
ng
fro
m
Univer
siti
Te
knologi
Mal
a
y
sia
(UTM).
H
e
be
ca
m
e
L
ec
t
ure
r
a
t
Depa
r
t
m
ent
of
Tele
c
o
m
m
unic
at
ion
Engi
ne
eri
ng,
Un
ive
rsit
y
Islam
Sulta
n
Agung
(UN
ISS
ULA)
Indo
nesia
in
2001
.
Dr
Arief
h
as
publi
shed
over
50
art
i
cles
in
jo
urna
ls
and
conf
ere
nc
e
pap
ers
a
nd
book
cha
p
te
r
.
His
rese
arc
h
int
er
est
in
cl
udes
the
areas
of
wir
el
ess
comm
unicati
ons,
in
p
art
i
cu
la
r
are
as
of
Cog
nit
ive
Radi
o
,
Inte
rne
t
of
Thi
n
gs
(IoT
)
,
Software
Defi
ned
Radio
(SD
R),
Softw
are
Defi
ned
Net
work
(SDN)
,
Biom
edi
cal
Eng
i
nee
ring
and
Appli
c
at
ion
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.
10
, No
.
4
,
A
ugus
t
2020
:
3
6
0
5
-
3
6
1
4
3614
Sh
ari
fah K
.
Sy
e
d Y
us
of
,
rec
ei
ve
d
her
BS
c
in
El
ec
trica
l
Eng
ine
er
i
ng
from
George
W
ashingt
on
Univer
sit
y
,
in
1
988.
She
becam
e
As
sistant
Lect
ure
r
at
the
Dep
art
m
ent
of
Com
m
unic
at
ion
Engi
ne
eri
ng,
Univer
siti
T
eknol
o
gi
Malay
si
a
(U
TM)
where
she
pursued
her
Ma
ster
and
PhD
degr
ee
s
at
UT
M
respe
ct
ivel
y
.
Her
rese
arc
h
int
ere
st
in
cl
u
des
the
are
as
of
wire
le
ss
comm
unic
at
ions
,
in
p
articu
l
ar
a
rea
s
of
OF
DM
,
Cognit
iv
e
Rad
i
o,
W
ire
l
ess
Sensor
Network
(W
SN
),
Networ
k
Coding,
Software
Defi
n
ed
Rad
io
(SD
R),
Software
Defi
n
ed
Net
work
(SD
N)
and
Engi
ne
eri
ng
Educ
ation.
She
has
publi
shed
over
150
art
icle
s
in
journa
ls
and
conf
ere
nc
e
pape
rs.
She
is
a
co
-
ed
it
or
o
f
a
book
entit
le
d
Outcome
-
B
ase
d
Scie
n
ce,
Te
chno
log
y
,
Engi
ne
eri
ng,
an
d
Mathe
m
atics
E
duca
t
ion:
Innov
a
ti
ve
Prac
t
ices a
n
d
seve
r
a
l
book
c
hapt
ers.
Muhammad
Hai
kal
Satria
,
r
e
ce
iv
ed
his
Mast
er
in
Com
pute
r
Engi
ne
eri
ng
fro
m
Univer
sita
et
Duisburg
Essen
Germ
an
y
and
PhD
in
in
El
ec
t
ric
a
l
Engi
n
ee
rin
g
from
Univer
siti
Te
kno
logi
Malay
s
ia
(UT
M)
.
His
rese
ar
ch
intere
st
ar
e
as
of
Software
Engi
ne
eri
ng
,
QoS
W
ire
le
ss
Networks
,
Medi
ca
l
Inform
atics
,
Hea
lt
h
Inform
at
i
cs
,
Hea
lt
hc
ar
e
Inform
at
ic
s
,
Te
l
emedic
in
e
,
Hea
lt
h
Inform
ati
on
Mana
gement
,
Medic
a
l
Inform
at
i
cs
Com
puti
ng
.
He
has
publi
sh
ed
over
50
art
i
cl
es
in
journ
al
s
and
conf
erence
pape
rs
and
book
cha
pte
r
.
Curre
ntly
he
joi
n
ed
as
Senior
Le
c
ture
r
at
Dep
art
m
ent
of
Cli
n
ic
a
l
Studie
s
Facul
t
y
of
Biom
edica
l
Engi
n
ee
r
ing,
Univer
siti
Te
knologi Mal
a
y
sia
(UTM).
Evaluation Warning : The document was created with Spire.PDF for Python.