TELKOM
NIKA
, Vol.14, No
.1, March 2
0
1
6
, pp. 101~1
0
9
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v14i1.2442
101
Re
cei
v
ed Au
gust 16, 20
15
; Revi
sed
De
cem
ber 1
0
, 2015; Accepte
d
De
cem
ber
28, 2015
Resear
ch on Sensor Network Spectrum Detection
Technology based on Cognitive Radio Network
Wen
z
hun
Hu
ang*, Xinxin Xie, Yuting Zhang, Shan
w
e
n
Zhan
g
Dep
a
rtment of Electron
ic Information En
gi
ne
er
in
g, Xi
jin
g Un
iversit
y
,
Xi
’a
n 7
101
23, Ch
ina,
No. 1, Xi
jin
g R
oad, Ch
an
g An
District, Xi'
an,
Shaa
n
x
i Provi
n
ce, Chin
a, (86)
029
84
188
25
7
Corresp
on
din
g
author, e-mai
l
: huan
g
w
e
n
zh
u
n
@
x
ij
ing.
edu.c
n
A
b
st
r
a
ct
W
i
th the
bursti
ng
dev
elo
p
m
e
n
t of c
o
mputer
scie
n
ce
an
d t
he
har
dw
are t
e
chn
o
lo
gy, Int
e
rnet
of
T
h
ings
a
nd w
i
r
e
less
sens
or
n
e
tw
orks has
b
een
p
opu
larly
studie
d
i
n
th
e
community
of
eng
ine
e
ri
ng. U
nde
r
the e
n
viro
n
m
e
n
t of Internet
o
f
T
h
ings, w
e
c
a
rry out
th
eore
t
ical a
nalys
is a
nd n
u
m
eric
al s
i
mulati
on
on t
h
e
sensor
netw
o
r
k
spectru
m
d
e
tection t
e
chn
o
lo
gy b
a
sed
on co
gn
itive r
adi
o n
e
tw
ork. As a
me
ans
of
infor
m
ation and
intelli
gence,
infor
m
ation s
e
r
v
ice syst
em
is
an
important
researc
h
hotspot in the field
of
Internet of t
h
in
gs.
W
i
reless
s
ensor
netw
o
rk
is co
mp
ose
d
o
f
a l
a
rge
n
u
mb
er of
micr
o s
e
nsor
nod
es, w
h
ic
h
have th
e functi
on of infor
m
ati
on col
l
ecti
on, d
a
ta pr
ocess
i
n
g
, and w
i
rel
e
ss communic
a
tio
n
, character
i
z
e
d
by
the i
n
tegrati
o
n
of w
i
reless
se
lf-orga
n
i
z
a
t
i
on.
How
e
ve
r,
mo
st of the
met
h
odo
log
i
es
pro
p
o
sed
by th
e ot
her
institutes ar
e
suffering for
m
the hi
gh c
o
mp
lexity w
h
il
e w
i
th the h
i
g
h
ti
me-cons
u
m
in
g
w
hen proc
essi
ng
infor
m
ati
on. T
herefor
e, this
study
is to
ass
e
ss the
eco
n
o
m
ic f
easi
b
il
it
y
of usin
g th
e o
p
timi
z
e
d
multi
pat
h
protoco
l
ava
ila
bility a
nd th
e i
n
creas
ed b
a
n
d
w
idth and s
e
v
e
ral
mo
bi
le o
p
e
rators thro
ug
h the us
e of c
o
st-
ben
efit a
nalys
i
s
, sing
le
path
sel
e
ction
mo
del
is to
dev
elo
p
mor
e
p
a
t
h
a
g
ree
m
ent t
o
ac
hiev
e
bet
ter
perfor
m
a
n
ce.
T
o
test the rob
u
stness, w
e
co
mp
are
our
met
hod w
i
th th
e ot
her state-
of-th
e
-art ap
pro
a
ch
in
the si
mulati
on
section
a
nd
pr
ove th
e
effectiveness
of
our
meth
od
olo
g
y.
T
he ex
peri
m
en
tal res
u
lt refl
ec
ted
that our
ap
pro
a
ch c
oul
d ac
hi
eve
hig
her
ac
curacy w
i
th l
o
w
time-co
n
su
mi
ng w
h
en
de
alin
g w
i
th c
o
mplex
sources of info
rmati
on.
Ke
y
w
ords
: Se
nsor N
e
tw
ork; Cog
n
itive
Ra
di
o Netw
ork; C
o
l
l
ab
oratio
n
for Distribut
ed Det
e
ction;
S
pectru
m
Detectio
n T
e
ch
nol
ogy; T
opo
lo
gy Optimi
z
a
tio
n
.
Copy
right
©
2016 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Wirel
e
ss se
n
s
or
net
wo
rk con
s
i
s
ts of
a
la
rg
e n
u
mb
er
of ra
ndo
m
dist
ribution,
energy
con
s
um
ption
and re
so
urce
con
s
traint
s o
f
sensor n
o
d
e
s, it has the
perceptio
n, calcul
ation abil
i
ty
and co
mmu
nicatio
n
abili
ty,
through
self-o
rg
ani
zat
i
on way co
n
s
titute the wirele
ss n
e
twork,
colla
boration
to re
al-time
acquisitio
n
and p
r
o
c
e
s
si
ng of the
ph
ysical
wo
rld
informatio
n,
can
reali
z
e th
e functio
n
of I
O
T comp
reh
ensive
pe
rc
e
p
tion. If the
reali
z
ation
of
co
gnitive ra
dio
techn
o
logy in
wirele
ss
se
n
s
or net
work,
not only
can
alleviate the
con
g
e
s
tion of
publi
c
fre
que
ncy
band, an
d re
duce the work freq
uen
cy o
f
the wirel
e
ss sen
s
o
r
net
work
also ma
kes the n
ode
after
singl
e jump
coverag
e
is
wi
de and
cha
n
g
e
, which grea
tly simplifies the network to
pology. But the
cog
n
itive radi
o technol
ogy
ha
s
bro
ught
many
ch
a
lle
nge
s
to
wi
rel
e
ss sen
s
o
r
n
e
tworks
[1
-2].
Wirel
e
ss
sen
s
or
network requi
re
s ne
w algo
rithm
s
and p
r
oto
c
ols to a
c
hie
v
e the two
key
techni
que
s:
cog
n
itive ra
d
i
o spe
c
trum
sen
s
in
g a
n
d
dynami
c
ch
annel
switchi
ng. Fo
r e
n
e
r
g
y
con
s
trai
ned
wirel
e
ss
se
nsor n
e
two
r
k, to
be
su
cce
ssfu
l
introd
uctio
n
of co
gnitive radio te
ch
nolo
g
y,
not only to
achi
eve the
s
e two te
chni
que
s, more t
o
re
du
ce the
ene
rgy con
s
umptio
n in
the
pro
c
e
ss of im
plementatio
n. Spectru
m
se
nsin
g to
sho
r
t
en the sle
ep
time of sen
s
o
r
node
s which
increa
se th
e
burd
en
of the
nod
es to
mo
nitor
cha
nnel
for
the exch
a
nge of
test re
sults
an
d
traff
i
c.
Inevitable det
ect errors
and will ca
use t
he sensor
net
work
confli
ct
with major customers, furt
her
increa
se the
addition
al en
ergy co
nsum
ption. In addition, freque
nt
cha
nnel switch also con
s
u
m
es
a lot of energy. Becau
s
e
often exist multi
path fad
i
ng wirele
ss
comm
uni
cati
on enviro
n
m
ent,
hidde
n terminal a
nd time
-varying
characteri
stics,
in
this
enviro
n
ment, the in
dividual i
s
ra
the
r
difficult to ca
n reali
z
e a
ccurate
spe
c
tru
m
sen
s
in
g. The use of sp
ace dive
rsity cha
r
a
c
teri
stics,
adopt the m
e
thod of multi
p
le u
s
er
co
o
peratio
n
a
w
a
r
ene
ss can e
ffectively overco
me the
ab
ove
probl
em
s impact on the reliability of spectrum sensi
ng.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 1, March 2
016 : 101 – 1
0
9
102
As a result, of this study is
to assess
the economi
c
feas
ibility of using the m
u
ltipath
proto
c
ol avail
ability and in
cre
a
sed ba
nd
width an
d
se
veral mobil
e
operators.
Th
roug
h the use of
co
st-be
nefit
analysi
s
, si
n
g
le path
sel
e
cti
on
model
is to d
e
vel
op mo
re p
a
t
h agreeme
n
t.
Demo
nst
r
ate
the use of th
e model, there are two
po
tential mobile
application scen
ario
s, fro
m
increasing av
ailability or bandwi
d
th. In the following
sections, we will
discuss the i
s
sues in detai
l.
2. Our Propo
sed Me
thod
olog
y
and Approac
h
2.2. The Sensor Ne
t
w
o
r
k
Spectr
u
m Detec
t
ion
With the
de
velopment
of wi
rele
ss
co
mm
uni
cation,
wi
rele
ss
sp
ectru
m
h
a
s
become
increa
singly
scarce, but some of the freque
ncy s
p
e
c
trum utilization is not filling urn. For m
o
re
effective to i
m
prove
sp
ectrum
utilization, nee
d to
further impr
ove the tra
d
i
t
ional spe
c
trum
manag
eme
n
t method, therefore, whi
c
h
use
d
in co
gni
tive radio sp
ectru
m
analy
s
is
strategy h
a
s
become a
hot
resea
r
ch topi
c [3]. Cog
n
itive radi
o spe
c
trum ca
n be
de
tected a
nd
ca
n be a
d
ju
sted
according to
the use of th
e co
rrespon
d
i
ng an
d
ne
w
techni
que i
s
an effective u
s
e of
spe
c
tru
m
.
Automatic
sp
ectru
m
dete
c
tion the b
a
si
s of a va
riet
y of cog
n
itive wi
rele
ss sensor
network
coexi
s
tence, but how to m
a
ke the
spect
r
um detection is more reli
ab
le is
still a research hot spot.
In fading
cha
nnel, the
mult
ipath effe
ct, shado
w
effect
and l
o
cal inte
rfere
n
ce
will
cause the
sign
al
to noise ratio
range i
s
lower than the thre
shol
d, tests are not co
mplete so a
singl
e wirele
ss
sen
s
in
g tech
nology is
unreliable. But d
ue to differ
e
n
t
position
sig
nal inten
s
ity in the network, if
colla
boration
for di
strib
u
te
d dete
c
tion, t
h
rou
gh
th
e n
e
twork
ca
n a
v
oid dete
c
tio
n
could
not
be
compl
e
ted. S
uppo
rt ve
ctor
machi
ne i
s
a
kind
of
b
a
sed
on stati
s
tical
learni
ng mod
e
l
an
d stru
ctu
r
e
risk minimu
m
princi
ple of machi
ne lea
r
ning metho
d
, is widely u
s
ed in statisti
cal cla
ssifi
cati
on,
appli
c
able
to
linear
sepa
ra
ble a
nd li
nea
r insepa
rabl
e
with u
nde
r th
e small t
r
aini
ng
sam
p
le
s
can
obtain the
ch
ara
c
teri
stics
of the go
od
cla
ssifi
cati
on
effect. The S
V
M cla
s
sifier ca
n be
u
s
ed
to
training th
e time-d
omain
si
gnal spe
c
tru
m
. At the
same time after using
com
p
ression p
e
rce
p
tion
spe
c
tru
m
det
ection p
r
o
b
le
m can b
e
tre
a
ted as li
nea
r insep
a
rable
ca
se cl
assification problem
[4].
In the spe
c
trum sen
s
ing
algorith
m
ba
sed
on com
p
re
ssi
on perceptio
n
an
d energy
test, the
comp
utationa
l complexity of syst
em are
mainly con
c
entrated in th
e recon
s
tru
c
ti
on of the sig
nal,
so we
try
to avoid
the co
mpre
ssed sig
nal
re
co
nst
r
u
c
tion a
nd di
rect u
s
e o
b
se
rvation sequ
e
n
ce
test spe
c
trum
. In the formula one, we d
e
fine the obje
c
t
i
ve function.
02
,:
,
,
,
N
x
zx
R
x
s
s
k
s
R
(1)
Observation
matrix is
Ga
ussian
ra
ndo
m matr
ix, an
d the
spa
r
se
matrix is o
r
thogo
nal
matrix, the two irrel
e
vant, and
com
p
re
ssed
sen
s
in
g
matrix with i
s
omet
ric
co
n
d
itions fo
r la
rge
probability constrai
nts.
22
2
22
2
11
s
ss
(2)
Here we only
care abo
ut the cla
s
sificati
on of the sa
mples
and n
o
t to reco
nst
r
uct the
origin
al time
domain
sig
n
a
l
, so do
es
not
need to
kn
o
w
the
spa
r
se
matrix. So in
the time dom
ain
sign
al in a transfo
rm do
m
a
in with
kno
w
n spa
r
se feature
s
, dire
ct
obse
r
vation
matrix is u
s
e
d
to
cha
nge the t
i
me domai
n sign
al is ma
pped to a d
o
main, an
d in the ob
serv
ational do
ma
in
cla
ssif
i
e
r
t
r
ai
n
i
ng an
d s
a
mp
le cla
s
sif
i
cat
i
on,
it
s
pe
rformance an
d di
rectly
to c
l
as
sify
performanc
e
clo
s
e to the origin
al time domain
sign
a
l
. In the
archi
t
ecture
of wireless sen
s
or
netwo
rk, a la
rge
numbe
r of
sensor nod
es are ra
ndom
ly
distri
buted
i
n
the m
onito
r
ed area,
to test
the nee
d
of
physi
cal qu
a
n
tities, these
sen
s
o
r
nod
es in a
c
cord
ance with th
e self-org
ani
zation m
ode
of
netwo
rk, with
the method
of multiple ho
ps to tran
smi
t
data to monitoring cente
r
or the Intern
et,
use
r
s
can m
onitor cente
r
for wirele
ss sen
s
o
r
net
wo
rks or the Int
e
rnet [8]. As a result of the
comm
on
sensor nodes communi
cation
ability is weak
,
often
in wireless sensor networks based
on regio
n
divi
ded i
n
to ma
n
y
clu
s
ters, e
a
c
h
clu
s
ter to
set u
p
a
com
m
unication
a
b
ility stron
g
n
ode
as the
clu
s
te
r hea
d, clu
s
t
e
r he
ad is
re
spo
n
si
ble for colle
cting int
e
re
st within t
h
is
cluste
r n
o
de,
and then thro
ugh multiple
hop
s way to tran
sfer
the in
formation to the gathe
ring
node, the no
de
is re
spo
n
si
ble
for the excha
nge of inform
ation with the
outside
worl
d [9].
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Re
sea
r
ch on
Senso
r
Network Sp
ect
r
um
Detectio
n Te
chn
o
log
y
ba
sed on… (We
n
zh
un Huan
g
)
103
Wirel
e
ss sen
s
or network and cellula
r mobile
tele
p
hone
n
e
two
r
k,
wirele
ss
l
o
cal are
a
netwo
rk an
d
other t
r
aditio
n
a
l wi
rele
ss n
e
t
work u
s
in
g
t
he wirel
e
ss si
gnal com
m
un
ication, but
th
ey
have differen
t
feature
s
a
n
d
de
sig
n
go
a
l
s [10]. In th
e
wirele
ss sen
s
or net
wo
rk,
in ad
dition to
a
small
numb
e
r of nod
es ne
ed to m
o
ve
beyond, m
o
st
node
s
are stationary, an
d gen
erally in
a
state of m
obil
e
cellula
r mo
bile telep
hon
e net
work
users,
ho
w to
e
n
su
re th
at th
e u
s
er call qu
ality
in mo
bile
ca
se, at the
sam
e
time th
e u
s
e of th
e m
a
ximum
save
b
and
width, i
s
a cellula
r m
o
bile
telepho
ne n
e
t
work n
eed
to solve the
p
r
oble
m
. S
upp
ose
in
a na
rrow
ban
d cog
n
itive in
sen
s
or
netwo
rk hav
e a
prima
r
y
use
r
, it allo
ws
se
con
dar
y
use
r
s u
s
e th
e spe
c
trum
i
n
spe
c
trum
free
circum
stan
ce
s. In orde
r to make full use of sp
e
c
tru
m
resource
s
and do
es not
affect the main
use
r
s,
seco
n
dary u
s
e
r
s
must
be
abl
e to q
u
ic
kly
and
accu
rate
ly perceive t
he p
r
e
s
en
ce
or
absen
ce of u
s
er
sig
nal. In orde
r to imp
r
ove c
ognitiv
e
perfo
rma
n
ce, ordin
a
ry seco
nda
ry use
r
's
perceptio
n of
the lo
cal
inf
o
rmatio
n tra
n
s
mitted to
th
e fusi
on
ce
nter, a
c
cording
to the fu
sio
n
cente
r
re
ceiv
es the info
rm
ation of ea
ch
node in the j
udgme
n
t [11]. Each sen
s
o
r
nod
e sp
ect
r
um
perceptio
n ca
n be reg
a
rde
d
as a
bin
a
ry hypothe
sis te
st probl
em.
0
1
,
,
k
k
kk
nH
xn
s
nn
H
(3)
Traditio
nal e
nergy dete
c
ti
on method first
re
ceived
after the sa
mpling si
gnal
modulu
s
squ
a
re, the
n
the ene
rgy a
c
cumul
a
tion b
y
using
statist
i
cs,
comp
ari
n
g with the
de
cisi
on thresh
old
finally and make
s a ruli
ng
whi
c
h is
sho
w
n in the formula four.
1
0
:
:
CED
C
ED
CED
C
ED
HT
HT
(4)
Con
c
lu
sion
b
y
statistical knowl
edge: if t
he n i
nde
pen
dent rand
om
variable
s
are
subj
ect
to Gau
ssi
an
distrib
u
tion, i
s
that
the
su
m of the squ
a
re
s of the
N
rand
om vari
a
b
les to
deg
re
es of
freedo
m for
N card
dist
rib
u
tion [12]. When the
average
N ra
ndo
m variabl
es
n
o
t ze
ro, they
pose
of the sum
of the sq
uares
of
the chi
-
sq
uare
ra
ndom
variable
s
a
r
e
subj
ect to th
e ce
nter. Wh
en
the ene
rgy a
c
cumulatio
n
points
whe
n
N is la
r
ge e
n
ough, the u
s
e of central limit theorem,
chi-
squ
a
re di
stri
b
u
tion ca
n be
approximate to Ga
u
ssi
an di
stributio
n den
oted as the fo
llowing.
2
22
1
24
0
:~
,
2
:~
,
2
CED
CED
H
TN
P
N
P
HT
N
N
(5)
Becau
s
e
of v
a
riou
s cogniti
ve nod
e di
stri
bution
s
in
different
ge
ogra
phical lo
catio
n
, sig
nal
sampli
ng, the
perceptio
n o
f
the starting
time is
not n
e
ce
ssarily
synch
r
on
ou
s. At the same ti
me,
the pe
rceptio
n to the fu
sio
n
center data
tran
smi
ssi
on
between
no
des w
hen
asked
delay i
s
not
the sam
e
, these factors will seri
ously affect the sy
nchronous collaboratio
n sensory perception
perfo
rman
ce
of the algorith
m
. In addition, the si
gnal in
a certain p
e
ri
od of when to
appea
r, whe
n
to exit in th
e
spe
c
tru
m
i
s
u
npre
d
icta
ble,
this
m
a
y re
su
lt in the
ph
ase spe
c
trum
p
e
rception
sig
nal
s
had al
re
ady
disa
ppe
are
d
. Data fu
sio
n
cent
er
usin
g Bayesi
an
optimal d
e
ci
sion rule
s of
the
likeliho
od
rati
o as the
wei
ght of ea
ch
soft deci
s
i
on,
and a
c
co
rdin
g to the
stati
s
tical fe
ature
s
of
authori
z
e
d
spectrum o
c
cupied mo
del,
the sen
s
o
r
y information
from the asy
n
ch
ron
o
u
s
d
a
ta
fusion,
reali
z
e the
sign
al i
s
ab
out the
existen
c
e
of
a ce
rtain
sp
e
c
trum
soft d
e
c
isi
on, so a
s
to
compl
e
te the
function of
coop
erative spe
c
trum
sensi
ng. In the formula
6, we defin
e the
comp
re
ssion sampli
ng
eq
u
a
tion.
1
N
ii
i
x
so
r
x
s
(6)
The ob
se
rvation se
que
nce
,
resultin
g fro
m
the com
p
ressed
sen
s
in
g is num
eri
c
al data,
and i
n
the
p
r
o
c
e
s
s of
cla
ssi
fier trai
ning
n
u
meri
cal
larg
er
data i
n
the
data
set
play
s a
lea
d
ing
ro
le,
influen
ce of training
to g
e
t
the
p
e
rfo
r
mance
of th
e cl
assifier.
Signal o
b
servation sequ
e
n
ce
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 1, March 2
016 : 101 – 1
0
9
104
norm
a
lization
pre
p
rocessin
g to inte
rval,
avoid bi
g dat
a in th
e d
a
ta
set pl
ays
a le
ading
rol
e
a
n
d
at
the sam
e
time re
du
ce the computati
onal
comp
l
e
xity. Collaboration with ex
isting
sequ
e
n
tial
detectio
n
often only does
seq
uential de
tection
in FC,
and CSDS
method in ea
ch no
de to do a
first ord
e
r pe
netration te
st. This is because for a
certain pe
riod
of logarithmi
c
likelihoo
d ra
tio,
con
c
rete dat
a may be ab
solute valu
e is very sm
all,
if need to directly uplo
a
d
the logarith
m
ic
likelihood rat
i
o to F
C
, for the impact
of judgm
e
n
t
is ve
ry
sm
all, but it
consume
s
m
o
re
colla
boration
overhe
ad [13
-
14].
At the beginn
ing of each ti
me slot, all node
s to che
c
k wh
ether th
ere is d
a
ta transfe
r: if
not, the network
within the
time slot to
enter
a do
rm
ant state; Spectr
u
m
dete
c
tion. Otherwi
se,
the
net
work bega
n
a sig
n
i
ficant cha
r
a
c
teristics of
wi
rele
ss sen
s
or network i
s
t
o
have
a la
rge
numbe
r of i
n
tensive di
st
ributi
on
of sensor n
ode
s and this to
pology is ve
ry suitabl
e for
coo
perative spe
c
tru
m
det
ection.
Coop
erative spe
c
trum dete
c
tio
n
than the
non-co
ope
rati
ve
spe
c
tru
m
de
tection ha
s
highe
r reli
ab
ility, es
pecial
l
y in a multipath fading
and shad
o
w
comm
uni
cati
on environm
ent [15]. Assume that all
s
e
ns
or
no
d
e
s
in
w
i
re
les
s
s
e
ns
or
ne
two
r
k
coo
perative spectrum dete
c
tion on a ch
annel at
the same time. T
he sche
dulin
g node an
d the
test re
sult
s o
f
each
nod
e
in the fusi
on
and d
e
ci
sio
n
, get the
whole n
e
two
r
k of the final
test
results. All
da
ta in th
e p
r
o
c
ess of
testin
g
thro
ugh
pu
bli
c
cont
rol
ch
a
nnel t
r
an
smi
s
sion
of
wirele
ss
sen
s
o
r
net
wo
rk. If the dete
c
ted a m
a
in u
s
er
ch
ann
el is idle, ea
ch
sensor n
ode in
the letter on
to
sched
uling
n
ode to t
r
an
smit data. Th
roug
h t
he
re
ply to co
nfirm wh
en the
frame i
n
d
a
ta
transmissio
n,
the sen
s
o
r
n
ode and data
tran
smi
ssi
on
sche
duling
n
ode
s a
r
e
kn
o
w
n
re
sults. If
the
sen
s
o
r
netwo
rk tra
n
smi
s
si
on to the main use
r
ch
ann
el is busy sta
t
e, in the current time slot for
the re
st of time to enter
a dorm
ant st
ate, wait
until
the next time slot ra
n test spect
r
um. T
h
e
Figure 1 illust
rates thi
s
.
Figure 1.
The Depl
oymen
t
of the
Sensor Net
w
o
r
k S
pectrum Dete
ction
2.3. The Cog
n
itiv
e Radio Net
w
o
r
k
Wirel
e
ss se
n
s
or
net
wo
rk
is comp
osed of
a
set
com
posed
of a la
rge
num
ber o
f
sen
s
o
r
s
deploye
d
in monitori
ng area of the wireless
network and it can
be used for
colla
boration
to
perceive, col
l
ection a
nd
pro
c
e
ssi
ng n
e
twork
coverage of the
area
of perceptio
n obje
c
t
informatio
n, and sent to the co
ntrolle
r.
Senso
r
nod
es st
ru
cture
will be de
sig
ned a
c
cordi
n
g to
different
ap
plication scena
rios,
the stru
cture of
the wirel
e
ss sen
s
or
network node i
s
used
to
descri
be
the different
pe
ri
ods and different
types
of sen
s
o
r
no
de
s. The
role
o
f
powe
r
supp
ly
module i
s
to
provide
ene
rgy to all sen
s
or no
de
s, typically the initial ene
rgy of sen
s
o
r
no
de
s is
equal, but in
orde
r to gua
rantee the tra
n
smi
ssi
on
task, gateway and the initial value of the base
station is l
a
rg
er. Informatio
n pro
c
e
s
sing
module
i
s
co
mposed of two part
s
of the
comp
uting a
nd
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Re
sea
r
ch on
Senso
r
Network Sp
ect
r
um
Detectio
n Te
chn
o
log
y
ba
sed on… (We
n
zh
un Huan
g
)
105
stora
ge,
the sensor node
itsel
f, the
colle
cted d
a
ta a
r
e
stored in
me
mory. Clu
s
te
r head
and
ba
se
station is u
s
e
d
to calculate
function, use
d
to
allocate
and man
age
the informatio
n tran
smissio
n
and
will a
c
ce
pt the info
rm
ation fusi
on
p
r
ocessin
g
. Se
nso
r
mo
dule
main d
u
ty is t
o
of all
kind
s
o
f
informatio
n collectio
n and
monitorin
g
environ
m
ent.
After the commissio
ning
of the wirel
e
ss
comm
uni
cati
on mod
u
le,
all se
nsor n
o
des
ca
n a
c
complish the
comm
uni
cati
on bet
wee
n
each
other, wo
rk together a
nd m
onitorin
g
tasks.
The features
of the cog
n
itiv
e radio n
e
twork
coul
d b
e
summ
ari
z
e
d
as the follo
wing. (1
)
Se
n
s
or
no
d
e
s
ar
e
us
ua
lly d
i
s
t
r
i
bu
te
d in
ve
r
y
ext
ensive
areas, in ord
e
r t
o
co
mplete
the
perceptio
n of the real wo
rld, high den
sity, sens
or n
ode depl
oym
ent is also v
e
ry intense. The
whol
e wirele
ss
sen
s
o
r
n
e
twork
syste
m
reliabilit
y and qu
ality of work rely
on larg
e-scale,
redu
nda
ncy
of emb
edd
ed
devices to
work togeth
e
r
to
implem
e
n
t and
imp
r
o
v
e. In this way,
throug
h pe
rceived inform
a
t
ion in the view of the di
fferent spa
c
e
sig
nal-to
-
noi
se ratio is bigg
er.
In
a large
amo
unt of info
rmation, di
stri
buted
pr
o
c
e
s
sing
metho
d
ca
n effe
ctively improve
the
accuracy of t
he monito
ring
, and lower p
r
eci
s
io
n
re
qui
reme
nts for
a
single
se
nso
r
nod
e. Due t
o
the pre
s
e
n
ce
of large am
o
unts of re
dun
dant nod
es,
t
he syste
m
fault toleran
c
e.
(2) Ea
ch
sen
s
or
node in the wirel
e
ss sen
s
or network is random di
vi
sion in monit
o
ring a
r
ea, th
e location of the
node ca
n't accurate
lo
cation
in adv
ance,
t
he n
e
ighb
or
relat
i
onship b
e
tween n
ode
s
are
unpredi
ctable
,
this requi
re
s that e
a
ch
sen
s
o
r
n
ode
itself ha
s
hi
gh ability of
self-o
rg
ani
zat
i
on
whi
c
h ca
n a
u
tomatically
config
ure a
n
d
automat
ic
manag
eme
n
t, through the
netwo
rk p
r
ot
ocol
and topol
ogy
control me
chani
sm auto
m
atically form forwa
r
di
ng
monitorin
g
d
a
ta more
wireless
netwo
rk.
(3)
Comm
on ca
ble co
mmuni
cation di
stan
ce of sen
s
o
r
node
s, in a
few meters to
hund
red
s
of
meters. Nod
e
s
can comm
unicate with
i
n
your comm
unication wit
h
in the sco
p
e
of
the adjacent node
s. But if
you need to communi
cate
with more tha
n
its commu
n
i
cation rang
e of
the othe
r n
o
des, yo
u mu
st pa
ss by
other
nod
es
fo
rwa
r
di
ng
rout
ing. The
mult
i-hop
routing
of
wirel
e
ss
sen
s
or net
wo
rk i
s
the com
m
on
netwo
rk
nod
e coll
abo
ratio
n
, that is to say, sometime
s
sen
s
o
r
nod
es nee
d to
a
c
t
as th
e
sp
on
sor
of the
info
rmation, fo
rwardin
g
and
th
e reci
pient th
e
three
role
s [1
6]. (4) Sen
s
o
r
netwo
rk i
s
di
fferent fr
om t
he add
re
ss centere
d
to th
e Internet, it is a
task-ba
s
e
d
n
e
twork a
nd it
s aim i
s
to o
b
tain the d
a
ta inform
ation.
Senso
r
n
ode
s in a
network of
all use seria
l
numbe
r id
entification, t
he
net
work
node
numb
e
r
is u
n
ified
depe
nd
s on
the
comm
uni
cati
on p
r
oto
c
ol
o
f
the
system.
Una
b
le to
det
ermin
e
the position of the sen
s
o
r
nod
es in
advan
ce, the
se
rial n
u
mb
er of
sen
s
o
r
node
s a
r
e
de
pendi
ng o
n
t
he sy
stem
dynamic allo
cat
i
on,
namely nod
e
numbe
r and
the node p
o
s
ition is n
o
n
e
ce
ssary lin
k betwee
n
the
two. (5) Se
n
s
or
energy
suppl
y in the
wi
rel
e
ss
sen
s
o
r
n
e
twork is
u
s
ually carry from its o
w
n
b
a
ttery, on
ce t
he
battery run
s
o
u
t, that the se
nso
r
no
de will
not
be able t
o
contin
ue to
work. Extend
workin
g ho
urs
of all nod
es,
therefo
r
e, tha
t
mean
s the
whol
e pr
olon
g the
workin
g life of the
wirel
e
ss
sen
s
o
r
netwo
rk.
The
node
in the
data tra
n
smission
pha
se
is
the con
s
um
ption of e
nergy, this requ
est
s
us in th
e ca
se of
compl
e
te the task doe
s not
affec
t
as
far
as
poss
ible
to reduc
e
data
transmissio
n, so as to a
c
hieve the pu
rpose of
save
energy and
prolo
ng the li
fe of the wh
ole
netwo
rk [17].
In the Intern
et of thing
s
,
an a
w
a
r
en
ess of
sen
s
in
g
node
s
will e
n
vironm
ent, so ofte
n
appe
ar this
ki
nd of conditio
n
, namely pe
rce
p
tion no
de
s in a netwo
rk, sign
al is al
ways le
ss than
spe
c
tru
m
allo
cation
and it
greatly redu
ces the
benefi
t
of a single
use
r
pe
rcepti
on [18]. Nod
e
s
and pe
rceptio
n of the Internet of things i
s
num
ero
u
s,
so the la
st of the
algorith
m
is propo
se
d in
th
e
n
e
x
t c
h
ap
te
r
w
e
co
ns
id
er
s
p
ec
ific
s
p
e
c
tr
um
u
s
ers pe
rcepti
on d
e
man
d
of the o
b
je
ctive
function, go
a
l
is to minimi
ze the p
e
rce
p
tion of
the failed to satisfy the user b
enefits. In the
Figure 2,
we
sho
w
th
e flo
w
chart
of the
propo
se
d method. Each
use
r
perce
ption expre
s
sed
with
a nod
e, if two users p
e
rception
o
r
con
f
licts bet
wee
n
the interfe
r
e
n
ce
and th
en
these
two
users
can
not u
s
e
th
e same
freq
u
ency
at the
same time,
th
e ne
ed to
a
d
d
a
ca
ble
bet
wee
n
two n
o
des.
Clou
d
comp
u
t
ing is the
core of th
e Internet of
thin
gs i
n
telligent i
n
fo
rmation
an
alysis.
The
u
s
e
of
cloud comput
ing technology, make
the real-time dynamic manage
ment of hundreds of millions
of various types of obje
c
ts.
With the developme
n
t of
the Internet a
pplicati
on, the increa
se of the
numbe
r
of th
e termi
nal,
can p
r
o
c
e
s
s
mass info
rm
ation, with
th
e aid
of
clou
d computin
g
for
auxiliary deci
sion-making
whi
c
h
will promote the in
formation processing ab
ility. Therefore, as
a
kind of virtu
a
lizatio
n, clo
ud com
putin
g soluti
o
n
of hard
w
a
r
e a
nd software
operation, it can
provide th
e Internet of thi
ngs
with efficient co
m
putat
ion, storage
cap
a
city
, pro
v
ide extensiv
e in
the link of the
Internet of things n
e
two
r
k
engin
e
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 1, March 2
016 : 101 – 1
0
9
106
Figure 2. The
Flocha
rt of the Propo
se
d Algorithm
From
the
poi
nt of the Int
e
rnet
of thin
gs
appli
c
atio
ns, thei
r
con
s
tru
c
tion i
n
d
u
strie
s
are o
f
the
system, is
no
t convenie
n
t to a variety of busin
es
s ex
pan
sion, con
s
tru
c
tion
stan
dard, if there
is
no unifie
d
sta
ndard Interne
t
access, inte
gration
ma
n
a
gement pl
atform, the Intern
et of things
will
be be
ca
use
of differen
c
e
s
in vario
u
s in
dustri
e
s
can't
pro
d
u
c
e
sca
l
e effect, in
crease the
use
of
compl
e
xity and
co
st. At the
same
time there i
s
ev
en a
coll
ecti
on an
d a
col
l
ection i
s
already
assign
ed fre
q
uen
cy ban
d, said i
n
a
c
cord with t
he
co
nstrai
nt condi
tions a
nd the
distri
bution
of
target un
der t
he premise of
a colle
ction
of freque
nci
e
s assig
ned to
the use
r
pe
rceptio
n. Another
colle
ction
can
be descri
bed
using
spe
c
trum
. The form
ula 7~8 defin
es the p
r
ocess.
11
Mk
s
idle
q
q
qq
NM
N
N
i
j
(7)
1
/
k
bs
s
b
ii
j
j
PP
(8)
Due to th
e u
s
er at the
sam
e
time, in the
posse
ssive spectrum
sp
a
c
e perce
ption
can
not
prod
uce inte
rfere
n
ce to prima
r
y use
r
s, and pe
rce
p
tion of use
r
s and the
main locatio
n
is
con
s
tantly ch
angin
g
, so th
e perce
ption
of differ
ent u
s
ers, its
ca
n
use f
r
equ
en
ci
es a
r
e
gene
rally
not the same.
Network laye
r acco
rd
i
ng to
the perceive
d
extensi
on o
f
busin
es
s ch
ara
c
t
e
ri
st
ic
s,
t
o
optimize the
netwo
rk fea
t
ures, b
e
tter realiz
e the
conte
n
t and
the comm
uni
cation b
e
twe
en,
conte
n
t and comm
uni
cati
on amon
g the peopl
e a
s
well a
s
interpe
r
son
a
l communi
catio
n
, it
requi
re
s m
u
st establi
s
h
an
end
-to-e
nd
global
netwo
rk. The
r
e
are
a lot of e
quip
m
ent a
c
cess
in
the Intern
et o
f
things, i
s
a
ubiquito
us acce
ss,
heterog
eneo
us a
c
ce
ss. E
a
ch
u
s
e
r
pe
rceptio
n
has
the two corresp
ondi
ng collectio
ns, n
a
mely ea
ch
node h
a
s
two sets a
s
so
ciated
with it.
Perception
of users in
a
certain time
a
nd
spa
c
e
ca
n
be u
s
e
d
o
n
t
he
spe
c
tru
m
of a colle
ctio
n of
resou
r
ces, i
s
the availabl
e
spe
c
tru
m
mat
r
ix. Sens
in
g spectrum
se
ction in the
u
s
e
r
a
c
cess to th
e
colle
ction,
wil
l
not affect th
e main
users or ot
h
e
r
perceptio
n. Spectrum is
usuall
y
divided into
a
seri
es of orth
ogon
al sub b
and, the main use
r
s ta
ke up one of the band, to a certai
n po
wer to
comm
uni
cate
and there i
s
no interfe
r
en
ce betwe
en fre
quen
cy and freque
ncy ba
n
d
[19-20].
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Re
sea
r
ch on
Senso
r
Network Sp
ect
r
um
Detectio
n Te
chn
o
log
y
ba
sed on… (We
n
zh
un Huan
g
)
107
3. Experimental An
aly
s
is and Simulation
In this sectio
n, we
cond
uct nume
r
ical
si
mu
lation
on
the p
r
o
p
o
s
ed
methodol
ogy
to test
the feasibility
of our approach. The sim
u
lation pl
atform is: PC wit
h
2.43GHz
CPU, 8GB RA
M,
Wind
ows 7
o
peratin
g
syst
em an
d MAT
L
AB R20
13a
spe
c
ial
editi
on. Du
e to t
he inte
rfere
n
c
e
phen
omen
on
of wirele
ss, within the
clu
s
ter afte
r
clu
s
ter h
ead, if the dista
n
ce
betwe
en the
m
is
less than the
cluste
r ra
diu
s
, can p
r
od
u
c
e inte
rfe
r
e
n
c
e bet
wee
n
clusters. Seco
ndary the cl
o
s
er
the di
stan
ce
between th
e cl
uste
r h
e
ad, then t
h
e
interfe
r
en
ce
between th
e cl
uste
rs wi
ll be
stron
g
e
r
. It
may even
cause m
e
ssa
ge
retra
n
sm
i
ssi
on, le
ad t
o
ad
ditional
energy
co
sts. So,
setting up seco
nda
ry clu
s
terin
g
linea
r wirele
ss
se
nso
r
network protocols in
each subn
et
irrig
a
tion are
a
, to compre
hen
sive co
nsideratio
n of cluster
comm
u
n
icatio
n ene
rgy con
s
umpti
o
n
and comm
un
ication dista
n
c
e and clu
s
t
e
r comm
uni
cation
covera
ge
ove
r
lap b
e
twee
n
the
t
w
o
comm
uni
cati
on interfe
r
en
ce p
r
obl
ems.
First of all, the forme
r
in
itial cluste
r n
u
mbe
r
is 1, t
he
average which
is
0.5,
effectivel
y reduc
e
the c
h
oice scope of
c
l
us
ter heads
.
Mak
e
a c
l
us
ter head
in a
rea
s
on
a
b
le rang
e. Beca
use of th
e improvem
ent after
c
l
us
tering algorithm in the initial
clu
s
terin
g
i
s
i
n
trodu
ce
d
co
nne
ctivity, so cho
o
se
a
re
aso
nabl
e initi
a
l clu
s
te
r he
a
d
s. To
o mu
ch of
the o
r
iginal
al
gorithm
to fo
rm cl
uste
r
nu
mber an
d
clu
s
ter st
ruct
ure
is mo
re f
r
ag
mented,
and
the
improve
d
alg
o
rithm of clu
s
ter numb
e
r i
s
6, the fo
rmation of clu
s
ter
stru
ct
ure si
ze
is rea
s
o
nabl
e,
good net
wo
rk conne
ctivity; Second in th
e final cluste
r head ele
c
tio
n
sea
s
on 4 times pe
r clu
s
t
e
r
head
ele
c
tion
pro
bability in
cre
a
sed. Alth
ough
co
mpa
r
ed with
the o
r
iginal al
gorith
m
improves t
he
electio
n
e
r
ror proba
bility in
cre
a
se
2 time
s. But d
u
e
to
the introdu
cti
on of
conne
ct
ivity in advan
ce
to en
su
re th
e final
clu
s
te
r h
ead
ele
c
ti
on i
s
i
n
a
re
aso
nabl
e
sco
pe, so th
e a
c
cura
cy
will
not
redu
ce, an
d the algo
rithm
can p
a
ss fewer iteratio
n
el
ects the final
clu
s
ter he
ad
s. In the Figure 3,
we
sh
ow the
experi
m
enta
l
com
p
a
r
ison
of the
average tran
smi
s
sion
po
we
r p
e
rform
a
n
c
e, t
he
corre
s
p
ondin
g
cu
rve
s
a
r
e
the metho
d
s pro
p
o
s
ed
by
us
and
by the literature
s
of [4-7]. In t
h
e
Figure 4, we
sho
w
the experim
ental co
mpari
s
o
n
of the sum rate and the addit
i
onal refe
ren
c
e
method i
s
p
r
opo
sed
by [9] as th
e mo
st
the rig
h
t si
d
e
of the curve.
To test th
e ro
bustn
ess of t
he
algorith
m
, we
dem
on
strate
anoth
e
r set
of expe
riment
as the
Figu
re 5
and
6,
re
spe
c
tively. From
the expe
rime
ntal re
sult,
we could
ea
sily
co
ncl
ude
tha
t
our metho
d
perfo
rms
well
. Com
pared
with
the othe
r stat
e-of-th
e
-a
rt al
gor
ithm
s, ou
r method
ology
achi
eves th
e
better p
e
rfo
r
mance a
nd t
h
e
stron
g
e
r
rob
u
s
tne
ss. As for the accu
ra
cy
, our
method
ology enha
nces the traditio
nal one
s by the
degree
of the
23%. As for t
he time
-con
suming,
our
al
gorithm
redu
ces th
e ove
r
all
time
con
s
um
ed
in the extent of the 21%. The simu
alatio
n result is sati
sfacto
ry.
Figure 3. The
Experimental
Result on Averag
e
Tran
smi
ssi
on
P
o
wer
Figure 4. The
Experimental
Result on the Sum
Rate
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 1, March 2
016 : 101 – 1
0
9
108
Figure 5. Re
sult on Averag
e Tran
smi
ssi
on
Power Set Two
Figure 6. The
Experimental
Result on the Sum
Rate Set Two
4. Conclusio
n
In this pap
er
,
we cond
uct
theoreti
c
al a
n
a
lysis a
nd n
u
m
eri
c
al
simul
a
tion on the
sen
s
o
r
netwo
rk
spe
c
trum detectio
n
technol
ogy based on
co
g
n
itive radio network und
er
the environm
ent
of Internet of
Thing
s
.
As a
nother
revolu
tion of
inform
ation indu
stry
, wirele
ss
se
nso
r
network
the
logical information world
and the o
b
j
ective phy
si
cal
worl
d to
gethe
r
,
ch
an
ged the
way
of
intera
ction be
tween hu
man
and nature, the impa
ct on human a
r
e compa
r
abl
e to those obtain
e
d
with the em
e
r
gen
ce
of the Internet.
Al
ong
with the
increa
singly
wide a
ppli
c
ation of wi
rel
e
ss
sen
s
o
r
net
wo
rk, its
se
cu
rity proble
m
s a
r
e b
r
ou
ght
to the attention
of the peopl
e, the se
cu
rity
probl
em i
s
n
o
t only rel
a
te
d to the n
e
twork th
e
con
f
identiality of informatio
n, and i
s
of g
r
eat
signifi
can
c
e f
o
r the
stabl
e
ope
ration
of the net
work
itself.
T
i
me
synchro
n
ization p
r
oto
c
ol i
s
wirel
e
ss se
n
s
or net
wo
rk
is an import
ant part of
the many applicatio
ns of wirel
e
ss se
n
s
or
netwo
rk, info
rmation qu
ery
,
for example,
target
tra
c
kin
g
, node p
o
siti
oning, rely on
accurate time
synchro
n
ization, time synchroni
zation
proto
c
ol is al
so the f
ound
ation of many communi
ca
tion
proto
c
ol
run.
Therefore, th
e safety of time sy
n
c
h
r
oni
zation
proto
c
ol is the
nee
d of the wi
rel
e
ss
sen
s
o
r
net
work
and
sta
b
le ope
ratio
n
whi
c
h i
s
also the f
ound
ation of
multi-ch
ann
el
comm
uni
cati
on. Our pro
p
o
se meth
odol
ogy solves th
e issu
es well whi
c
h co
uld be refle
c
ted from
the experime
n
tal result. In the near future, we
plan to use the network optimi
z
ation theo
ry to
modify the cu
rre
nt method
to obtain better re
sult.
Ackn
o
w
l
e
dg
ements
This wo
rk is financially sup
porte
d by t
he scientific rese
arch
program of Shaanxi
Provinci
al Educatio
n De
pa
rtm
ent (Program No. 14
JK
2156
).
Referen
ces
[1]
T
Ali, P Siddi
q
ua, MA Mati
n.
Performanc
e e
v
alu
a
tion
of d
i
fferent mo
dul
ati
on sch
emes fo
r ultra
w
i
d
e
ban
d s
y
stems.
Journ
a
l of Elec
trical Eng
i
ne
eri
n
g
. 201
4; 65(3)
: 184-18
8.
[2]
H W
ang, J Wang.
An effecti
v
e imag
e repr
esentati
on
met
hod us
in
g ker
nel cl
assificati
on
. T
he 26th
Internatio
na
l C
onfere
n
ce o
n
T
ools
w
i
t
h
Artifi
cial Intel
lig
enc
e. Limasso
l (CY). 2014: 85
3-85
8.
[3]
A Massouri, L
Card
oso, B Guillo
n, et al.
Cor
t
eXlab: An o
p
e
n
FPGA-based
facility for testing SD
R &
cogn
itive rad
i
o
netw
o
rks in
a repro
duci
b
l
e
enviro
n
m
ent
.
2014 IEEE
Confer
ence
o
n
Comp
uter
Commun
i
cati
o
n
s W
o
rkshops
(INF
OCOM
WKSH
PS).
T
o
ronto, Cana
da. 2
014: 10
3-1
04.
[4]
D Z
hang, Q Liu, W
Xu. A
ccess Conv
er
genc
e of Mec
han
ical E
qui
p
m
ents to Het
e
rog
ene
ous
Net
w
orki
ng E
n
vironme
n
t in In
ternet of T
h
ing
s
for Manufact
u
rin
g
.
Journ
a
l
of Conv
erge
nc
e Informatio
n
T
e
chno
logy
. 2
013.
[5]
KE Ale
x
and
er. Barriers
an
d
ena
blers
to d
e
l
i
ver
y
of th
e h
e
a
lth
y
ki
ds ch
ec
k: an a
n
a
l
ysis
informe
d b
y
the theoretic
al
doma
i
ns fr
ame
w
o
r
k an
d com-
b mode
l.
Im
plem
entation Science
. 201
4; 60(
6): 396-4
15.
[6]
Z
Guo,
X W
a
n
g
, J T
ang,
X
Li
u, Z
Xu, M W
u
, Z
Z
hang.
R2:
An
app
lic
ation
-
level
ker
nel
fo
r record
a
n
d
repl
ay
.
T
he 8th USENIX c
o
n
f
erence o
n
Operatin
g s
y
stem
s desig
n an
d impleme
n
tatio
n
.
San Dieg
o
,
CA (US). 2008:
193-2
08.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Re
sea
r
ch on
Senso
r
Network Sp
ect
r
um
Detectio
n Te
chn
o
log
y
ba
sed on… (We
n
zh
un Huan
g
)
109
[7]
H Skin
nemo
e
n
. Introducti
o
n
spec
ia
l iss
ue o
n
dv
b-r
cs2.
Internati
ona
l Jo
urna
l
of Satel
lite
Co
mmun
icati
o
ns & Netw
orkin
g
. 2013; 3
1
(5): 199-
200.
[8]
AM Shah, S Z
hang, C Mapl.
Cog
n
itive
ra
dio
netw
o
rks for
Internet
of T
h
i
n
gs: App
licati
o
n
s
, chal
len
g
e
s
and fut
u
re
. T
he 19th
Intern
ation
a
l C
onfer
en
ce on
Au
tomat
i
on
an
d Com
p
uting (ICA
C). L
ond
on (GB).
201
3: 1-6.
[9]
M Gechev, S.
Kasabov
a, A
D
Mihovska,
et al.
N
ode
d
i
s
c
overy and
i
n
terpretati
on in
unstructure
d
resourc
e
-const
r
ain
ed e
n
vir
o
n
m
e
n
ts
. T
he 4th Internati
o
n
a
l
Conf. on W
i
r
e
less
Commu
n., Vehicu
la
r
T
e
chnolog
y,
In
formation
T
heor
y a
nd A
e
ros
pace
& El
ectro
n
ic S
y
stems
(
V
IT
AE). Aalbor
g(DK). 2
014:
1-5.
[10]
F
Antóni
o, L
Migu
el, O R
o
dolfo,
et a
l
.
Cha
nne
l Av
ail
abil
i
t
y
Assess
ment for
Cog
n
itive
Ra
dios
.
T
e
chno
log
i
ca
l Innov
atio
n for
the Internet of T
h
in
gs.
Sprin
ger
Berlin H
e
id
elb
e
rg. 201
3.
[11]
F Antóni
o, L
Mi
gue
l, O R
odo
lfo,
D
Rui,
B L
u
i
s
. Cha
n
n
e
l Av
a
ilab
ilit
y
Assess
ment for
Co
gni
tive R
adi
os
.
T
e
chno
log
i
ca
l Innov
atio
n for the Internet of T
h
in
gs
. Sprin
ger
Berlin H
e
id
elb
e
rg. 201
3.
[12]
A Aijaz, M
T
s
hang
ini, M Nak
hai, et al. Ener
g
y
-E
ffici
ent Up
l
i
nk Res
ource
Alloc
a
tion i
n
L
T
E Net
w
orks
W
i
th M2M/H2
H Co-E
xisten
ce Un
der St
atistical QoS
Guarant
ees.
Co
mmu
n
ic
ations IEE
E
T
r
ansactio
n
s o
n
. 2014; 6
2
(7): 235
3-23
65.
[13]
J Le
ithon,
JL T
eng, S
Su
n.
E
nergy
exch
an
g
e
a
m
ong
b
a
se
stations
in
a
C
e
llu
lar
Netw
ork
throu
g
h
the
Sm
art Grid
. 20
14 IEEE Intern
ation
a
l Co
nf. on Commu
n., (ICC). S
y
d
n
e
y
(
A
U). 2014: 4
0
3
6
-40
41.
[1
4
]
AM Al
b
e
r
ti
. A co
n
c
e
p
t
ua
l
-
d
r
i
v
en
su
rvey o
n
fu
tu
re
in
te
rn
e
t
re
qu
i
r
e
m
e
n
t
s, te
ch
no
l
o
g
i
e
s
, and
chall
e
n
ges.
Jo
urna
l of the Bra
z
i
l
i
an C
o
mput
er Society
. 20
1
3
; 19(3): 29
1-3
11.
[15]
M Usman, I Koo. Secur
e
co
oper
ative sp
ec
trum sensin
g for the co
gnitiv
e
radi
o n
e
t
w
or
k using
no
n-
uniform reliabilit
y
.
Scientific W
o
rld Jo
urna
l
. 2014.
[16]
C W
an, X F
eng, IN Center
,
ZT
Colle
ge. Accurate dete
c
tion of dos attacking si
gn
al bas
ed
o
n
spectrum al
iasi
ng pre-
distortio
n
effect.
Bulleti
n of Science &
T
e
chno
logy
. 2
014.
[17]
HP Cor
por
atio
n. A n
e
w
i
n
tr
usio
n d
e
tectio
n s
y
stem b
a
s
ed
on c
l
assific
a
tion
al
gorithm
in
w
i
re
les
s
sensor n
e
t
w
ork
.
Journal of El
e
c
trical & Co
mp
uter Engi
ne
erin
g
. 2014.
[18]
JH Martin,
LS
Dool
e
y
, KCP
W
ong.
A
New
Cross-Lay
er
D
y
na
mic
Spectr
um Access
Ar
chitecture
for
T
V
W
h
ite Sp
a
c
e Co
gn
itive
Radi
o A
ppl
icat
ions
. Inte
lli
ge
n
t
Sign
al Pr
oce
ssing
Co
nf. L
ond
on (GB).
201
3: 1-6.
[19]
Z
hang,
Hu
i
y
in
g
,
Hon
g
zu
o L
i
,
Xi
ao
Do
ng
ya,
Cai
Ch
ao. P
e
rformanc
e A
nal
ysis of
Differe
nt Mod
u
lati
o
n
T
e
chniques fo
r F
r
ee-Space
Optical Comm
unic
a
tion S
y
st
em.
T
E
LKOMNIKA (T
eleco
m
mu
nicati
o
n
Co
mp
uting El
e
c
tronics an
d C
ontrol).
20
15; 1
3
(3).
[20] Indra
w
at
i,
Irme
il
ya
na, F
i
tri
Ma
ya Pus
p
ita,
Ok
y
San
j
a
y
a.
Internet Prici
ng o
n
Ban
d
w
i
dth F
unctio
n
Dimin
ishe
d With
Increasi
ng
Ba
nd
w
i
dth
Utilit
y Func
tion.
T
E
LKOMNIKA (T
ele
c
ommunic
a
tio
n
Co
mp
uting El
e
c
tronics an
d C
ontrol)
. 20
15; 1
3
(1): 299-
30
4.
Evaluation Warning : The document was created with Spire.PDF for Python.