Indonesian J
ournal of Ele
c
trical Engin
eering and
Computer Sci
e
nce
Vol. 1, No. 2,
February 20
1
6
, pp. 354 ~
364
DOI: 10.115
9
1
/ijeecs.v1.i2.pp35
4-3
6
4
354
Re
cei
v
ed Se
ptem
ber 20, 2015; Revi
se
d De
ce
m
ber
28, 2015; Accepted Janu
ary 15, 201
6
The Performance of Dynamic-
static S
p
ectrum Access
Based on Markov Transfer Model
Chan
gbiao Xu*, Lianjun Zhu
Cho
ngq
in
g Uni
v
ersit
y
of Posts
and T
e
leco
m
m
unic
a
tions, C
hon
gqi
ng, Ch
in
a.T
E
L:185238
540
21
*Corres
p
o
ndi
n
g
author, e-ma
i
l
:xuc
b@c
qupt.
c
n, 1852
38
54
0
21@
163.com
A
b
st
r
a
ct
T
h
is study
pr
e
s
ented
a
n
on-
synchro
ni
z
e
d
r
and
o
m
access
mech
anis
m
that su
pp
orts d
y
na
mic-
static spectru
m
.General
ly,
the authori
z
e
d
us
ers of tw
o netw
o
rks (net
w
o
rk A and netw
o
rk B) w
i
th differe
n
t
freque
ncy b
a
n
d
s co
mmu
n
ica
t
e w
i
th each
o
t
her by
usin
g
static spectru
m
. W
h
e
n
co
n
gestio
n
h
a
p
p
e
ned
,
spectru
m
ho
le
s in their ne
tw
orkscan be
detect
ed by
the oppos
ite
netw
o
r
ksand
then utili
z
e
d
to
co
mmu
n
i
ca
te
.
Ba
se
d o
n
qu
eu
i
n
g
th
eo
ry and
Ma
rko
v
tr
an
sfer mod
e
l, a
user b
e
h
a
vior
character
i
z
e
d
by
dyna
mic-static spectru
m
acce
ss w
a
s proposed, and the fe
asibi
lity of th
is theoretic
al
mo
del w
a
s valid
at
e
d
throug
h a
nal
o
g
si
mul
a
tio
n
. T
hereafter, the
theoretic
al p
a
r
ameters of sy
stem
p
e
rfor
ma
nce, lik
e bl
ock
i
n
g
possi
bil
i
ty, forced dro
p
-cal
l p
o
ssibi
lity, and
throug
hp
ut, w
e
re me
asur
ed a
nd co
mp
are
d
betw
een
dyna
mic
-
static mo
de an
d unco
n
j
ugate
d
mod
e
.
Ke
y
w
ords
: Dy
na
mic-static sp
ectrum, Marko
v
transfer
mo
d
e
l, non-sy
nchr
oni
z
e
d ran
d
o
m
access, que
ui
ng
mo
de
l
Copy
right
©
2016 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Strict internat
ional rule
s h
a
ve bee
n e
s
tab
lish
ed for the usage
of wirel
e
ss
spectrum
resou
r
ces. A
c
cordi
ng to t
he cu
rrent p
r
actical
usage
and researches, the
r
e a
r
e three
sp
ect
r
um
pattern
s whe
n
co
mmuni
ca
tion
co
ntrol center com
m
u
n
icate with
te
rminal: (1)
st
atic spe
c
trum
;
(2
)
semi
-dynami
c
spe
c
tru
m
; and
(3
) dy
namic spe
c
tr
um. Gen
e
rall
y, static
spe
c
trum
is fixedly
assign
ed by
the
state a
c
cordi
ng to
national
s
p
e
c
trum
re
so
urce
s, an
d it
belon
gs to
the
una
ssi
gned
a
nd sh
ared sp
ectru
m
s. Be
sides, semi-dy
namic
sp
ectrum is ge
ne
ra
ted via auctio
n
,
and dyna
mic spe
c
tru
m
is
obtaine
d thro
ugh dyna
mic
perceptio
n. In this
study, static resou
r
ce
stand
s
for sta
t
ic
sp
ect
r
um, while dynami
c
re
so
urce
re
pre
s
ent
s both
semi
-dynami
c
an
d dynami
c
s
p
ec
tr
ums
.
The majo
rity of tradition
al cog
n
itive r
adio te
chn
o
l
ogie
s
focu
s on the sh
aring of
spe
c
tru
m
s
wit
h
sa
me frequ
ency b
and,
while few
studi
es h
a
ve be
en
con
d
u
c
ted fo
r the
com
b
ine
d
acce
ss of a
u
thori
z
ed
an
d
unauth
o
ri
zed
spe
c
tru
m
s.
In
order to m
e
et the
se
rvice
dem
and
of l
o
w
delay, la
rge
band
width,
a
nd b
r
o
ad
dat
a servi
c
e
sco
pe, future n
e
twork should
allow the
access
of spe
c
tru
m
s with differe
nt freque
ncy
band. Ho
wever, the u
s
age of dyna
mic spe
c
tru
m
is
different from
that of static spe
c
trum. T
heref
o
r
e, an
effective mod
e
l
that represents dynami
c
-
static spe
c
tru
m
acce
ss i
s
o
f
great signifi
can
c
e.
2. Rese
arch
Statu
s
Dynami
c
spe
c
trum a
c
ce
ss is the b
a
sis fo
r the
co
mbination of
dynamic a
n
d static
resou
r
ces. Howeve
r, the previou
s
researche
s
a
b
o
u
t spe
c
trum
acce
ss g
ene
rally study th
e
acce
ss
strate
gy of dynami
c
spe
c
trum
o
r
the b
ehavi
o
r of
cog
n
itive users in
same fre
que
n
cy
band. Fe
w studie
s
have
been perfo
rmed on th
e combi
ned
acce
ss of authori
z
ed
and
unauth
o
ri
zed spe
c
tru
m
s.
Based o
n
the
access of dynamic
spe
c
trum
, Xie et al. have prop
osed an acce
ss model
that com
b
ine
s
h
e
tero
gen
e
ous networks, in whi
c
h
ma
in and
secon
dary n
e
two
r
ks u
s
e
sp
ect
r
u
m
s
with different
frequ
en
cy b
and [1]. Bian
et al
have con
s
tru
c
ted
a
hete
r
og
ene
ous
coexiste
nce
stru
cture for
cog
n
itive net
work to
solve
the p
o
tential
probl
em
s ab
o
u
t exch
angin
g
an
d
controll
ing
informatio
n a
nd inte
re
st co
nflict [2]. Tzel
atis
a
nd
Be
rb
eridi
s
have
e
s
tablish
ed a system stru
ctu
r
e
for the acce
ss of wirele
ss
spe
c
tru
m
ba
sed on c
ogniti
on, and it fits the hetero
g
e
neou
s wi
rele
ss
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
The Perfo
r
m
ance of Dyn
a
m
ic-static Sp
ectrum
Acce
ss Base
d on M
a
rkov… (Cha
ngbia
o
Xu)
355
netwo
rk
of next gene
rat
i
on [3]. Thro
ugh combi
n
i
ng traditio
nal
emerg
e
n
c
y comm
unication
techn
o
logy a
nd co
gnitive wirel
e
ss sy
stem, Han et a
l
. have con
s
tructe
d Marko
v
transfer mo
del
and an
alyze
d
some
syste
m
para
m
eters [4]. Gene
ra
lly, there are
many wi
rele
ss se
rvicers who
comp
ete with
each
oth
e
r, and se
con
d
a
r
y
users c
an use
th
eir dyn
a
mic biddin
g
model
to ada
pt
their a
c
ce
ss t
o
differe
nt se
rvice
provid
ers [5]. Fu
rth
e
rmore,
cog
n
itive use
r
s can
be divide
d in
to
two cla
s
se
s (high pri
o
rity class and lo
w prio
rity
class) accordi
ng to
their se
rvice
deman
ds
wh
en
dynamic
spe
c
trum a
c
ce
ss is perfo
rme
d
[6]. Li et al.
have pro
p
o
s
e
d
a joint cha
n
nel agg
reg
a
tion
splitting alg
o
rithm, providin
g a que
ueing
model for d
y
namic
spe
c
t
r
um a
c
cess [
7
]. In order t
o
improve
the
capa
city factor of
limited ch
annel,
Sulta
n
a
et
al. have establi
s
h
ed a
n
a
c
cess syst
em
stru
cture with
multi-p
r
iority
and m
u
lti-u
s
er, an
d se
t n
on-p
r
e
e
mptive pri
o
ritie
s
fo
r u
s
er time [
8
].
More
over, Ku
mar
et al. ha
ve mea
s
ured
the pe
rforma
nce
of dyna
m
i
c a
c
ce
ss
wh
en p
r
ima
r
y a
nd
secondary
users are using
IEEE
802.11, a shared frequen
cy band[9]. Jiang et al
. have studied
the distu
r
ba
n
c
e of dyn
a
mi
c spe
c
trum
a
c
cess
to m
a
ster u
s
ers
wh
ose
com
m
uni
cation fe
atures
are un
kno
w
n
to
se
co
nda
ry use
r
s, and
th
us de
signe
d an
a
c
ce
ss protocol
for dyn
a
mic sp
ect
r
um
[10].
3. The Sy
ste
m
Model of D
y
namic-sta
t
ic Spectrum
Acce
ss
3.1. Net
w
o
r
k
Model
It’s assum
ed
that there
are
only two
net
wo
rks (nam
el
y,
network
A and netwo
rk B,
and
their auth
o
ri
zed users (na
m
ely, U
A
and
U
B
) take sp
ectru
m
s
with
different fre
quen
cy ban
d
s
. If
user
requests arrive, th
ey will
preferentially get
access into thei
r own authori
z
ed net
works via
static
spe
c
tru
m
. As the n
e
tworks of U
A
and U
B
have cognitive ability, they will dynami
c
ally
percept
spe
c
t
r
um h
o
le
s if there i
s
n
o
av
ailabl
e
cha
n
n
e
l in their
own networks,
and thu
s
utili
ze
these
spe
c
tru
m
holes to a
c
ce
ss dyn
a
mic spe
c
trum.
In this stu
d
y, the available
band
width
s
o
f
network A a
nd B are re
prese
n
ted by
chann
el
numbe
rs, whi
c
h a
r
e set e
q
ual to facilitate analysi
s
. Authori
z
ed u
s
e
r
s h
a
ve abso
l
ute prio
rity to
use the
cha
n
nels in thei
r o
w
n net
works, in comp
ari
s
o
n
with se
co
nd
ary use
r
s. For example, if the
requ
est of U
A
arrives, stati
c
spect
r
um
s
of network
A will be firstly
checked.
If there is idle
cha
nnel, thi
s
requ
est
will
a
c
cess
ch
ann
e
l
dire
ctly. If there i
s
no i
d
le
cha
nnel
and
some
chan
ne
ls
are taken by U
B
, the com
m
unication of
U
B
will be stoppe
d imme
diately (force
d drop
-call), and
available
cha
nnel
s will b
e
provide
d
to U
A
. If
there is
n
o
idle chan
ne
l and no
ch
an
nel is ta
ken
by
U
B
(nam
ely,
all chan
nel
s in netwo
rk A are taken by U
A
), the requ
est of U
A
will be rejected.
The
acce
ss
situati
on of netwo
rk B is si
milar t
o
that of network A.
It's a
s
sumed
that the
arri
val rate
s
of
requ
est
s
fro
m
UA
and
UB in
net
wo
rk A an
d
netwo
rk B can display the Poisson
distrib
u
tion,
while
servi
c
e
time can show the n
e
g
a
tive
expone
ntial d
i
stributio
n. This stu
d
y aims to
con
s
truct and analyze a model fo
r dynami
c
-static
spe
c
tru
m
access ba
sed o
n
queuin
g
t
heo
ry and Marko
v
transfer m
o
del.
3.2. Queuing
Model
The m
a
jor m
odel utili
zed
in the p
r
e
s
e
n
t st
udy i
s
t
he M/M/m in
stant-refu
s
ing
,
multi-
wind
ow,
and
hybrid
-q
ueui
ng mo
del. T
a
kin
g
the
an
alysis of
U
A
beh
avior ch
ara
c
ters as an
example, it'
s
assume
d that
: (1) the
r
e
are m
ch
annel
s in n
e
two
r
k A
,
while
n
ch
a
nnel
s in
net
work
B; (2) th
e cha
nnel n
u
mb
er
requi
re
d by o
ne requ
est
in
netwo
rk A or
netwo
rk
B is
1; (3
) the
arri
va
l
rate of n
e
w
reque
st can d
i
splay t
he Po
isson di
strib
u
t
ion (inten
sity:
λ
); (4
)
serv
i
c
e time
of on
e
cha
nnel
can
sho
w
th
e n
e
gative exp
o
n
ential di
strib
u
t
ion (paramet
er:
μ
).
As
a
result, a
qu
eu
ing
model is e
s
ta
blish
ed for th
e combi
ned a
c
cess
of dyna
mic-static
spe
c
trum
(Figu
r
e
1).
Figure 1. The
queuin
g
mod
e
l for the com
b
ined a
c
ce
ss
of dynamic-st
a
tic sp
ect
r
um
Numb
er in
cycle: numb
e
r o
f
users in net
work;
u
s
ers from 0 to m use static spe
c
trum in
network A, while users from m+1 to m+n utiliz
e avail
able channel
s in network B via dynamic
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 2, February 201
6 : 354 – 364
356
perceptio
n.
If k users transfer thei
r
st
atus, transi
tion probability
will
be m
a
rked as P
k
. W
h
en
s
t
ab
le
state is a
c
hie
v
ed, the functi
on for statu
s
transitio
n is:
0,
0
0
,
1
0
,
0
(
1
)
The loa
d
parameter of a
singl
e ch
ann
el,
ρ
, is defined a
s
. Ac
c
o
rding to the
feature
s
of Markov tran
sfer, P
k
has polarity. Therefore
,
result of the above fun
c
tio
n
is:
∑
!
(
2
)
By utilizing 0 < k < m
+
n, it
is recursively
obtained that:
ρ
ρ
1
ρ
ρ
ρ
ρ
2
ρ
ρ
ρ
……
(
3
)
Therefore, th
e gene
ral sol
u
tion of state prob
ability P
k
is:
!
(
4
)
The que
uing
model for the
combi
ned a
c
ce
ss of dy
na
mic-static
spe
c
trum in n
e
twork B is
s
i
milar to that in network
A.
3.3. Acce
ss
Strateg
y
Before the ac
cess
of U
A
re
que
st o
r
U
B
requ
est,
the auth
o
ri
ze
d fre
quen
cy
ban
ds
(nam
ely, stati
c
spectrums) of network
A and B
will
be firstly checked.
Network A i
s
taken
as
example. If there i
s
idle
ch
annel in n
e
twork A, U
A
req
uest will
acce
ss
cha
nnel i
n
netwo
rk A in
a
static
man
n
e
r
. If there i
s
no i
d
le
ch
a
nnel i
n
n
e
twork A, the
chann
els in
n
e
twork B will
be
che
c
ked. Fu
rthermo
re, if there i
s
idle
chann
el in net
work B, U
A
reque
st will a
c
cess
chan
n
e
l in
netwo
rk B in
a dynami
c
m
anne
r. Th
e p
r
iority of aut
h
o
r
ize
d
u
s
e
r
s is high
er th
an t
hat of
cog
n
itive
use
r
s in
both
network A
a
nd B, n
a
mely
, prio
rity of
U
A
> p
r
iorit
y
of
U
B
in network A, whil
e priority
of U
B
> p
r
io
rit
y
of
U
A
i
n
n
e
t
work B. T
h
e
r
efore, if
U
B
reque
st a
rrive
s a
nd th
ere i
s
n
o
oth
e
r id
le
cha
nnel in n
e
twork B, the comm
unication of U
A
will bestopp
ed
(forced dro
p
-call), and t
h
e
corre
s
p
ondin
g
ch
ann
el take
n by U
A
will be p
r
o
v
ided to U
B
,
avoiding int
e
rferen
ce to
U
B
comm
uni
cati
on. The a
cce
ss
situation of
network
B is
s
i
milar to that of network
A.
In the
com
b
in
ed a
c
ce
ss of
dynamic-static
spe
c
trum p
r
opo
se
d in
thi
s
study, requ
ests of
U
A
and
U
B
ca
n get acce
ss
into netwo
rks in a non
-ran
dom man
n
e
r
. Again, network A i
s
take
n as
example. The
chan
nel
s in netwo
rk A a
r
e seq
uentia
lly
marked by n
u
mbe
r
s from 1 to m (from l
o
w
freque
ncy
ba
nd to hi
gh freque
ncy b
a
n
d
), while th
e
ch
annel
s i
n
network B
are
se
que
ntially
marked by n
u
mbe
r
s from
1 to n (from l
o
w fr
e
que
ncy
band to hig
h
freque
ncy b
and). If there
is
idle
cha
nnel
i
n
net
wo
rk A,
U
A
requ
est
wi
ll acce
ss th
e f
i
rst i
d
le
ch
an
nel in
chan
ne
l 1-m.
If there
is
no idle chan
n
e
l in netwo
rk A and there i
s
idle
chan
ne
l in netwo
rk
B, U
A
reque
st
will acce
ss t
he
first idle
cha
n
nel in
cha
n
n
e
l n-1
of net
work B
(Figu
r
e 2
)
. The
acce
ss
pro
c
e
ss of netwo
rk
B is
simila
r to that
of netwo
rk A (Figu
r
e 2
)
. In
other wo
rd
s,
autho
rized use
r
s acce
ss
comm
uni
cati
on
system p
r
efe
r
entially via ch
annel
s with lo
w frequ
en
cy band
s othe
r than that with
high freq
uen
cy
band
s, while
cog
n
itive use
r
s a
c
ce
ss
co
mmuni
cati
on
system p
r
efe
r
entially via ch
annel
s with
hi
gh
freque
ncy ba
nds.
(1) If
static
sp
ectru
m
can m
eet user
dem
and,
requ
est of
autho
rized use
r
will
a
c
ce
ss
idle
static spe
c
tru
m
/chan
nel (p
riority: low fre
quen
cy band
> high freq
u
ency ba
nd) i
n
a non-ran
d
o
m
manne
r (Fi
g
u
r
e 2a, 2b, an
d 2c).
(2) If there i
s
no idl
e
chan
nel in
network A (namely,
all the
cha
n
n
e
ls in
net
work A a
r
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
The Perfo
r
m
ance of Dyn
a
m
ic-static Sp
ectrum
Acce
ss Base
d on M
a
rkov… (Cha
ngbia
o
Xu)
357
taken by
U
A
) and the
r
e i
s
idle chan
nel
in network
B, U
A
request will dynami
c
ally access
idle
cha
nnel of ne
twork B (pri
ority: high frequ
ency ba
nd
> low fre
quen
cy
band)
(Fig
ure 2d). Similarly,
the acce
ss p
r
oce
s
s of U
B
in netwo
rk B i
s
sh
own in Figure 2
e
.
(3) After
com
p
leting U
A
re
que
st in net
work A, the
corre
s
p
ondin
g
ch
annel i
s
relea
s
e
d
,
providin
g an
addition
al idl
e
ch
ann
el in
netwo
rk
A. Then, the
co
gnitive user
of netwo
rk B
will
swit
ch the
sp
ectru
m
to its
authori
z
e
d
freque
ncy b
a
n
d
in net
work
A (prio
r
ity: low freq
uen
cy
band
> high fre
que
ncy ban
d) (Fi
gure 2f
).
(4) It's a
s
sum
ed that all the cha
nnel
s in
netwo
rk A a
r
e taken by U
A
, some
ch
annel
s in
netwo
rk B
are taken
by
U
A
, and
there i
s
no oth
e
r idle
cha
nnel
in
ne
twork B. If
U
B
re
que
st a
rriv
e
s,
the com
m
uni
cation
of U
A
will be
stop
p
ed (fo
r
ced d
r
op-call
),
and
the co
rresp
ondin
g
chan
nel
taken by
U
A
will be
relea
s
ed and
provided to U
B
, as priority of U
B
> pri
o
rity of U
A
in network B
(Figu
r
e 2h
).
(5) If all th
e
cha
nnel
s in
netwo
rk A are take
n by
U
A
and the
r
e i
s
no
idle
ch
a
nnel in
netwo
rk B, ca
ll blockin
g
will
happ
en to
U
A
(Figu
r
e 2i
).
Similarly, call
blockin
g
can
also
hap
pen
to
U
B
(Figu
r
e 2j
).
Figure 2. The
access p
r
ocess of netwo
rk A and B (m = n = 5
)
4. The Trans
f
er Mod
e
l of D
y
namic-sta
t
ic Spectrum
Acce
ss
4.1. Markov
Trans
f
er Mo
del
It's assume
d that the arriv
a
l rate of req
uest
s
from U
A
and U
B
can
display the Poisson
distrib
u
tion, while
th
eir service
time can
sh
ow th
e neg
ative e
x
ponential
di
stributio
ns wi
th
para
m
eters
μ
A
and
μ
B
, resp
ectively. In addition,there are m an
d n available cha
nnel
s in network
A and B, resp
ectively. The
use
r
num
be
rs of U
A
and
U
B
are
rep
r
e
s
ent
ed by intege
r
pair (i, j
)
, whil
e
their corre
s
p
ondin
g
po
ssi
bilities a
r
e sh
own a
s
P(i,
j). Beside
s, sta
t
e spa
c
e (i, j) shoul
d meet
the
following formula:
Γ
,
|
0
,
0
,
0
(5)
Then, the
M
a
rkov tra
n
sfe
r
mo
del of
d
y
namic
-stati
c sp
ectrum a
c
cess i
s
con
s
tru
c
ted,
sup
portin
g
m and n availa
b
l
e cha
nnel
s in
network A an
d B, respe
c
tively (Figure 3).
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 2, February 201
6 : 354 – 364
358
λ
blo
c
k
s
t
a
t
e
0,
0
0,
1
0,
n
0,
m
n
1
0,
m
n
1,
m
n
1
m,
n
m1
,
n
1
m
1
,
n
m
1
,
0
m
1
,
1
m,
0
m,
1
m
n
,
0
m
n
1
,
0
m
n
1
,
1
λ
λ
2
λ
n
λ
(n
+1
)
λ
(m
+n
‐
1)
λ
(m
+n
)
λ
λ
λ
(m
+
n
‐
1)
λ
λ
(m
‐
1)
m
λ
(m
+1
)
λ
(m
+n
‐
1)
(m
+
n
)
λ
λ
λ
λ
λ
λ
2
2
n
n
λ
λ
λ
(n
+1
)
λ
λ
λ
(m
‐
1)
(m
‐
1)
m
m
(m
+1
)
λ
λ
(m
+n
‐
1)
λ
λ
λ
λ
λ
λ
λ
λ
λ
λ
(m
‐
1)
1,
0
1,
1
1,
n
λ
λ
λ
λ
λ
λ
2
n
(n
+1
)
2
2
2
c
o
m
p
u
l
s
o
r
y
i
nt
e
rrup
t
t
r
a
n
s
f
e
r
s
t
a
t
e
bl
o
c
k
s
t
a
t
e
c
o
m
p
ul
s
o
ry
i
nt
e
rru
p
t
t
ra
ns
f
e
r
s
t
a
t
e
Figure 3. Markov tran
sfer
model of
dyn
a
mic-stati
c
sp
ectru
m
acce
ss
In Figure 3, the regi
on bet
wee
n
two dot
ted lines
represe
n
ts stati
c
acce
ss. State (0, 0)
mean
s that
neither U
A
nor
U
B
i
s
i
n
communi
catio
n
, whil
e stat
e (1,
0)
me
ans that
U
A
is
in
comm
uni
cati
on a
nd
req
u
e
stsget a
c
ce
ss into
syst
e
m
in a
no
n-random
ma
nn
er
(pri
ority: l
o
w
freque
ncy b
a
nd >
high fre
quen
cy ba
nd). The arro
w
from state
(0,
0) to stat
e (1,
0) an
d
λ
A
(or
λ
B
)
stand
s for th
e
acce
ssi
on of
U
A
(o
r U
B
),
while the a
rro
w from state
(1
, 0) to state
(0, 0) an
d
μ
A
(or
μ
B
) stand
s for the departu
re of U
A
(or U
B
) after answe
ri
ngthe re
que
st
of U
A
(or U
B
).
In Figu
re
3, t
he
regio
n
s on
the
right
of
vertical
dotted
lines re
prese
n
t that U
A
dynamically
take
s un-auth
o
rized freq
ue
ncy band
s in netwo
rk B.
For example, state (m, 0) mean
s that m
U
A
in
system a
r
e i
n
com
m
uni
ca
tion, and all
of
the m stat
ic chann
els
are ta
ken
by U
A
, as
U
A
can
preferentially get access int
o
static fre
quency band in a non-random manner.If
a new U
A
re
q
uest
arrives, it will
get access into
network B
in a n
on-ra
nd
om ma
nne
r (prio
r
ity: high f
r
equ
en
cy ba
n
d
> low f
r
equency band), and state
(m, 0) will
change
into state (m+1, 0).In state (m
+1, 0),
if a
certai
n U
A
o
c
cupying
static ch
ann
el
le
aves sy
stem,
the U
A
which is dynami
c
ally occupyin
g
netwo
rk B
will tran
sfer into t
h
is
ch
annel
in
a n
on-ra
ndo
m man
ner.T
h
i
s p
r
o
c
e
s
s is repre
s
e
n
ted
b
y
the arrow fro
m
state (m
+1,
0) to state (m, 0) and (m
+)
μ
A
.
In the region with sla
s
h on
the right of d
o
tted lines,
. It means that there is
no chan
nel a
v
ailable fo
r U
A
,
and
thu
s
, new re
que
st from U
A
will
be rejected and blocking
will
happ
en (e.g
. state (m+n, 0)).In this circu
m
sta
n
ce, if a new U
B
re
qu
est arrives,
U
A
comm
uni
cati
on will be sto
pped (fo
r
ced
drop
-call) to
avoid interfe
r
ence to U
B
, as the pri
o
rit
y
o
f
U
A
< the pri
o
ri
ty of U
B
in network B. This forced drop
-call
i
s
rep
r
e
s
ente
d
by dotted arro
w fro
m
(m+n, 0) to (m+
n
-1, 1) and
λ
B
.
Similarly, the regio
n
s u
nde
r hori
z
o
n
tal d
o
tt
ed lines m
ean that UB
dynamically take
s the
un-a
u
tho
r
ized
frequ
en
cy b
and i
n
n
e
twork A, an
d
t
he circum
sta
n
ce
s are
co
nsi
s
tent with
U
A
descri
bed a
b
o
ve.
4.2. Theore
t
ical Analy
s
is
Based
on th
e prin
ciple
of con
s
e
r
vation
of energy, the input flow shoul
d equ
a
l
to the
output flow when a ste
ady state is
a
c
hie
v
ed. Therefo
r
e, the followin
g
formula
s
wil
l
be obtaine
d:
Whe
n
0,
0
1
,
,
1
,
1
1
,
1
1
1
,
(6)
Whe
n
0,
,
,
,
1
(7)
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
The Perfo
r
m
ance of Dyn
a
m
ic-static Sp
ectrum
Acce
ss Base
d on M
a
rkov… (Cha
ngbia
o
Xu)
359
Whe
n
0
,
,
,
,
1
1
,
1
,
1
(8)
Whe
n
,
,
,
,
1
1
,
1
,
1
1
,
1
(9)
Whe
n
,
,
,
,
1
1
,
1
,
1
(10)
Whe
n
0
,0
,
,
1
,
1
1
,
1
1
1
,
1
,
(11)
Whe
n
,
0
,
,
1
,
(12)
In the above formul
as, whe
n
0
,
1
; or,
0
.
The sum of static po
ssi
bilities of all states is 1, nam
ely:
∑
,
(
,
)
ϵΓ
1
(13)
In orde
r to a
nalyze the
system perf
o
rmance
of dynamic-static
spe
c
tru
m
acce
ss, it’s
assume
d th
at mutual
interf
eren
ce
do
n
o
t
exist
b
e
twe
en m
a
ste
r
u
s
ers a
nd/or se
con
dary
u
s
ers,
and the
r
e i
s
no del
ay i
n
ch
ann
el switch.
Und
e
r this
circum
stan
ce, thre
e statu
s
e
s
are
determi
ned, i
n
clu
d
ing no
n-blockin
g
state
,
bl
ocking
state, and forced
drop
-call stat
e.
(1)
No
n-bl
ocking
state i
s
the statu
s
o
f
normal
co
mmuni
cation
whe
n
the int
e
rferen
ce
s
from othe
rs a
r
e not taken into accou
n
t.
(2) Blockin
g
state. As
sh
o
w
n in
Figu
re
3,
blockin
gha
ppen
s in
two
circu
m
stan
ces: (a) if
all of the
cha
nnel
sin n
e
two
r
k A
and B a
r
e occu
pied
b
y
their auth
o
ri
zed
user
s, ne
w reque
st
s from
authori
z
e
d
or unautho
rize
d use
r
s will b
e
reje
cted;
(b
) if all of the
cha
nnel
s in network A and
B
are o
c
cupie
d
by their autho
rize
d or
unau
thorized
u
s
e
r
s, new
req
u
e
s
ts fro
m
auth
o
rized u
s
e
r
s
will
not be reje
cte
d
, while ne
w reque
sts fro
m
unauth
o
ri
zed
use
r
s
will be rejecte
d
.
(3) Fo
rced d
r
op-call
state.
As sho
w
n in
Figur
e 3, fo
rced d
r
op
-call
occurs
wh
en
there i
s
no idle
ch
an
nel. In othe
r
words, if a
n
unauth
o
ri
zed
use
r
ta
ke
s the chan
nel o
f
authori
z
ed
use
r
,
the comm
uni
cation of this unautho
rize
d use
r
will b
e
stopp
ed by
force a
nd th
e co
rre
sp
ond
ing
cha
nnel
will be relea
s
e
d
when auth
o
ri
ze
d use
r
nee
ds
this ch
ann
el.
If there is n
o
available
chann
el in ne
twork,
syste
m
will reje
ct
new requ
ests of call
servi
c
e to guarantee the
current
service quality. Therefore, the
probability of call-blocking is
defined a
s
:
Traditio
nally, only stati
c
sp
ectru
m
can
g
e
t
acce
ss int
o
net
work, n
e
twork A
and
B are
indep
ende
nt, and
thei
r a
u
t
horized
u
s
ers
can
not
per
cept
sp
ect
r
u
m
hol
es in t
he oth
e
r net
work
throug
h dyna
mic spe
c
trum
acce
ss. In th
is conditi
o
n
, new
callin
g reque
st will b
e
reje
cted, if there
is n
o
availa
bl
e chann
el. Fo
rce
d
d
r
op
-call
ca
nnot
h
app
en
when
stati
c
spe
c
tru
m
is getting a
c
ce
ss,
as p
r
iority do
es not exi
s
t betwe
en diffe
rent u
s
er
s
(b
usin
ess p
r
iori
ty is not con
s
ide
r
ed i
n
this
study). Therefore, the po
ssibilities of blocki
ng an
d forced drop-call
of U
A
and U
B
are:
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 2, February 201
6 : 354 – 364
360
,
,
,
0
,
0
(14)
In the pre
s
e
n
t study, dynamic-stati
c spectrum
a
c
cess is availa
ble. If all of the stati
c
freque
ncy ba
ndsare
o
c
cu
pied by
auth
o
rized user
s and no ch
an
nel
i
s
avail
a
b
l
e
in perce
ptive
freque
ncy b
a
nd, the req
u
e
s
tsfrom autho
rize
d u
s
ers
will be reje
cted
. The blockin
g
possibilitie
s of
U
A
and U
B
are
:
,
∑∑
,
(15)
,
∑∑
,
(16)
If a user req
u
e
st acce
sse
s
system dyna
mi
cally and a
u
thori
z
ed u
s
e
r
s ne
ed this
chann
el,
the commu
ni
cation of una
uthori
z
ed u
s
er will be
st
oppe
d to avoid influen
cin
g
the authori
z
ed
use
r
s,
as th
e pri
o
rity of autho
rize
d
use
r
>
the
prio
rity of unautho
rized
use
r
. Unde
r this
circum
stan
ce
, the possi
bility of forced drop-call is d
e
fined a
s
:
Accordi
ng to Figure 3, the forc
ed drop-call possibilities of U
A
and U
B
are:
,
∑∑
,
(
,
)
(17)
,
∑∑
,
(
,
)
(
1
8
)
In the above formul
as, whe
n
0
,
1
; or,
0
.
In addition t
o
the po
ssi
b
ilities of blo
c
king
and fo
rced
dro
p
-call
,
the index named
busi
n
e
ss thro
ughp
ut cap
a
city should al
so be mea
s
u
r
ed to evaluat
e a system.
After the arriv
i
n
g
of use
r
req
u
e
sts, only a
part of them
can obt
ai
n servi
c
e
s
thro
ugh sy
stem, while the oth
e
r
requ
est
s
are reje
cted o
r
force
d
to dro
p
. Being
re
pre
s
ented by Th, throug
hput ca
pacity is defin
ed
as th
e n
u
mbe
r
of
req
u
e
s
ts
whi
c
h
obtain
servi
c
e
s
th
ro
ugh
syste
m
p
e
r
unit of time
. If all the
use
r
s
in a system h
a
ve the same
rate of data signaling, the t
h
rou
ghp
ut of U
A
and U
B
will
be:
1
,
1
,
(19)
1
,
1
,
(20)
5. Result o
f
Simulation and Performa
nce An
aly
s
is
5.1. Result o
f
Simulation
In the simul
a
tion pro
c
e
s
s, the cha
nnel
numbe
rs of n
e
twork A an
d
B were
set
as
5
and
3
; the se
rvice rates
of U
A
an
d U
B
wer
e
set
a
s
0
.
3
and
0
.
2
; the re
que
st
arriv
a
l rate
s of
U
A
and
U
B
, namely,
λ
A
and
λ
B
we
re
set a
c
cordi
n
g to the ki
nd
ofperfo
rman
ce
analysi
s
.In a
ddition, “the
o
r
y” stoo
d for the theoreti
c
al value
s
o
b
tained by u
s
ing m
a
tlab2
010
softwa
r
e and
dynami
c
-stat
i
c spe
c
tru
m
acce
ss
ba
se
d on
Markov
tran
sfer mo
del, nam
ely, the
mathemati
c
al
ly computed
values de
rive
d from ma
the
m
atical form
u
l
a (15), (16),
(17
)
, and (1
8
)
.
“sim
ulate” re
pre
s
ente
d
th
e sy
stem
sim
u
lation va
lu
e
s
b
a
sed o
n
C langu
age,
bl
ocking
po
ssi
b
ility,
and forced d
r
op-call po
ssib
ility. The simulation time wa
s set a
s
T=10
0s.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
The Perfo
r
m
ance of Dyn
a
m
ic-static Sp
ectrum
Acce
ss Base
d on M
a
rkov… (Cha
ngbia
o
Xu)
361
(a) T
he rel
a
tionship bet
we
en blo
cki
ng
possibility P
blockA
and arrival rate
λ
A
when
0
.
2
(b) T
he rel
a
tionship bet
we
en blo
cki
ng
possibilityP
blockB
and arrival
rate
λ
B
wh
en
0
.
3
Figure 4. The
relation
ship
betwe
en blo
c
king p
o
ssibilit
y and arrival rate
As shown in
Figure 4, si
m
u
lation results were
pe
rfe
c
tly con
s
i
s
ten
t
withmathem
atically
comp
uted va
lues, indi
cati
ng that it was app
ro
pria
te to use queuei
ng theo
ry and Markov
transfo
rm mo
del
to analy
z
e
the acce
ss
pro
c
e
s
s
of dynamic-static spe
c
tru
m
. Along with
t
h
e
increa
se i
n
arrival
rate
o
f
use
r
requ
e
s
t, bl
o
ckin
g
possibility po
ssesse
d in
creasi
ng trend.
In
addition, blo
c
king p
o
ssibili
ty increa
se
d quite gent
ly
whe
n
the correspon
ding a
rrival rate wa
s
relat
i
v
e
ly
sm
all,
as
sy
st
e
m
re
so
ur
ce
s
co
uld
m
eet
user
dema
n
d
s. Foll
owi
n
g
the in
crea
se of
arrival
rat
e
, system resou
r
ce
s
coul
d n
o
t meet
user d
e
mand
s any
longe
r, resulting in
the
ra
pi
d
increase in bl
ocking possibility.
(a) T
he rel
a
tionship bet
we
en forced d
r
o
p
-call
possibility P
forcedA
and arrival rate
λ
B
when
0
.
3
(b) T
he rel
a
tionship bet
we
en forced d
r
o
p
-call
possibility P
forcedB
and arrival rate
λ
A
when
0
.
2
Figure 5. The
relation
ship
betwe
en forced dro
p
-call p
o
ssibility and
arrival rate
As sh
own i
n
Figu
re 5,
simulatio
n
results
we
re
also
perfe
ctly con
s
i
s
te
nt with
mathemati
c
al
ly computed
values. Along
with the incr
ease in arriva
l rate of mast
er user requ
e
s
ts,
the forced drop-call possi
b
ility of
secondary user al
so elevated,
as forced drop-call happened in
the acce
ss p
r
oce
s
s of dynamic spe
c
tru
m
.
5.2 Compari
s
on of Ma
th
ematically
C
o
mputed Val
u
es
In this secti
on, Matlab20
10 wa
s utilized to
calcul
ate and co
mpar
e the theoretical
para
m
eters o
f
system
pe
rforman
c
e
bet
wee
n
dy
n
a
mi
c-static mod
e
an
d
uncon
jugated
mod
e
,
inclu
d
ing blo
c
king p
o
ssibilit
y, forced drop
-call p
o
ssibilit
y, and throug
hput.
0
0.
2
0.
4
0.
6
0.
8
1
1.
2
1.
4
1.
6
1.
8
2
0
0.
05
0.
1
0.
15
0.
2
0.
25
0.
3
0.
35
0.
4
0.
45
u
s
ers
a
rri
v
e
rat
e
A
b
l
o
ck r
a
t
e
P
bl
oc
k
,
A
t
h
eor
y
si
m
u
l
a
t
e
0
0.
2
0.
4
0.
6
0.
8
1
1.
2
1.
4
1.
6
1.
8
2
0
0.
1
0.
2
0.
3
0.
4
0.
5
u
s
e
r
s
a
rri
v
e
ra
t
e
B
b
l
o
ck r
a
t
e
P
bl
oc
k
,
B
th
e
o
y
si
m
u
l
a
te
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
0.
8
0.
9
1
0
0.
01
0.
02
0.
03
0.
04
0.
05
0.
06
0.
07
0.
08
u
s
er
s
ar
r
i
v
e
r
a
t
e
B
f
o
r
c
ed
in
t
e
r
r
u
p
t
pr
o
b
a
b
ilit
y
P
fo
r
c
e
d
,A
t
h
eor
y
si
m
u
l
a
t
e
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
0.
8
0.
9
1
0
0.
0
1
0.
0
2
0.
0
3
0.
0
4
0.
0
5
0.
0
6
0.
0
7
0.
0
8
0.
0
9
0.
1
us
er
s
ar
r
i
v
e
r
a
t
e
A
fo
r
c
e
d
i
n
te
r
r
u
p
t
p
r
o
b
a
b
l
i
ty
P
fo
r
c
e
d
,B
th
e
o
r
y
si
m
u
l
a
t
e
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 2, February 201
6 : 354 – 364
362
(a) T
he rel
a
tionship bet
we
en blo
cki
ng
possibility P
blockA
and arrival rate
λ
A
when
0
.
2
(b) T
he rel
a
tionship bet
we
en blo
cki
ng
possibility P
blockB
and arrival rate
λ
B
when
0
.
3
Figure 6. The relationship
between theoretical bl
ocking
possibility and arrival rate
Red t
r
ian
g
le:
dynami
c
-sta
tic a
c
cess; g
r
een
s
qua
re:
static
acce
ss o
n
ly. As shown in
Figure 6, along with the
i
n
crease in arrival rate of user
request, blocking possibility possessed
increasing t
r
end in both
modes.
However, the blocking possibility of
U
A
(o
r U
B
) in
dynamic-static
acce
ss
wa
s much lo
we
r than that in
static acce
ss with the sam
e
λ
A
(or
λ
B
)
,
a
s
s
p
ec
tr
um ho
le
s
coul
d be
pe
rceived
and
utilized to
communi
cate
in the a
c
ce
ss p
r
o
c
e
ss
of dynami
c
-sta
tic
s
p
ec
tr
um.
(a) T
he rel
a
tionship bet
we
en theoretical
possibility of forced drop-call P
forced
A
and
arriv
a
l r
a
te
λ
B
(b) T
he rel
a
tionship bet
we
en theoretical
possibility of forced drop-call P
forced
B
and
arriv
a
l r
a
te
λ
A
Figure 7. The
relation
ship
betwe
en theo
retical p
o
ssibi
lity of
force
d
drop
-call and
arrival rate
For the a
c
ce
ss of dynami
c
spe
c
tru
m
, forced
dro
p
-call
happ
en
s in th
e acce
ss pro
c
e
ss
o
f
dynamic-static spectrum, while forc
ed drop-callpossibility is 0 in
the access proc
ess of static
spe
c
tru
m
. As sho
w
n in Fi
gure
7, along
with the in
crease in
λ
A
and
λ
B
, theoretical po
ssibility of
U
A
(o
r U
B
) fo
rced d
r
o
p
-call
elevated. As priority of U
B
> prio
rity of U
A
in the a
c
cess process
of
dynamic spectrum,
theore
tical possibility of U
A
force
d
dro
p
-call
si
gnifica
ntly increa
sed
wh
en
λ
B
increa
sed. When
λ
B
wa
s fi
xed, users u
s
ing dyn
a
mic
spe
c
tr
u
m
in
creased
along
with the
in
cre
a
se
in
λ
A
, and thu
s
, theoretical
possibility of U
A
forced dro
p
-call elevate
d
.
0
0.
2
0.
4
0.
6
0.
8
1
1.
2
1.
4
1.
6
1.
8
2
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
us
er
s
ar
r
i
v
e
r
a
t
e
A
b
l
o
c
k
r
a
te
P
bl
oc
k
,
A
unc
onj
ugat
ed t
heor
y
dy
nam
i
c
-
s
t
a
t
i
c
c
onj
ugat
ed
t
heor
y
0
0.
2
0.
4
0.
6
0.
8
1
1.
2
1.
4
1.
6
1.
8
2
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
0.
8
0.
9
us
er
s
a
rri
v
e
ra
t
e
B
b
l
o
ck r
a
t
e
P
bl
oc
k
,
B
d
y
n
a
m
i
c
-
s
t
a
t
i
c
c
o
nj
ug
at
ed
t
h
eo
ry
un
c
o
nj
ug
at
ed
t
h
eo
r
y
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
0.
8
0.
9
1
0
0.
01
0.
02
0.
03
0.
04
0.
05
0.
06
0.
07
0.
08
0.
09
us
er
s
ar
r
i
v
e
r
a
t
e
B
f
o
r
c
e
d
i
n
t
e
r
r
u
p
t
p
r
o
b
a
b
ilit
y
P
f
o
r
c
ed,
A
A
=
0
.
4
t
heor
y
A
=
0
.
3
t
heor
y
A
=
0
.
2
t
heor
y
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
0.
8
0.
9
1
0
0.
0
2
0.
0
4
0.
0
6
0.
0
8
0.
1
0.
1
2
us
er
s
a
r
r
i
v
e
r
a
t
e
A
f
o
r
c
e
d
in
t
e
r
r
u
p
t
p
r
o
b
a
b
ilit
y
P
fo
r
c
e
d
,B
B
=0
.
3
B
=0
.
2
B
=0
.
1
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
The Perfo
r
m
ance of Dyn
a
m
ic-static Sp
ectrum
Acce
ss Base
d on M
a
rkov… (Cha
ngbia
o
Xu)
363
(a) T
he rel
a
tionship bet
we
en theoretical
throug
hput of
U
A
and arrival rate
λ
A
when
0
.
2
(b) T
he rel
a
tionship bet
we
en theoretical
throug
hput of
U
B
and arrival rate
λ
B
when
0
.
3
Figure 8. The
relation
ship
betwe
en theo
retical th
rou
g
hput and a
rriv
a
l rate
Red t
r
ian
g
le:
dynami
c
-sta
tic a
c
cess; g
r
een
s
qua
re:
static
acce
ss o
n
ly. As shown in
Figure 8, the theoreti
c
al through
put
in dynamic-static
acce
ss
wa
s sim
ilar to that in static a
c
cess
whe
n
λ
wa
s small, a
s
a
u
thori
z
ed
ch
annel
s
coul
d
meet u
s
er
deman
d with
limited u
s
ers.
Ho
wever,
alo
ng
with the
increa
se i
n
a
r
rival ra
te, the
theoretical
throu
ghp
ut in
dynami
c
-stat
i
c
acce
ss
wa
s markedly hig
her than that i
n
static a
c
cess with a same
arrival rate.
6. Conclusio
n
Based
on
no
n-rand
om a
c
ce
ss, thi
s
study si
m
u
late
d the a
c
ce
ss pro
c
e
s
s of
dynamic-
static spe
c
tru
m
by using
queui
ng theo
ry and
Markov transfe
r model, and t
hen obtain
e
d
the
formulas of perform
a
nce param
e
ters, includi
ng bl
ocki
ng possibility, forced
drop-call possibilit
y,
and through
put. The fea
s
ibility and reliability of our mo
del
were validat
ed by com
p
aring
theoreti
c
al value
s
with simulative values.
Th
e pe
rforma
nce p
a
ram
e
ters of
dynamic-sta
tic
spectrum access were
calculated through sim
u
lation,
includ
ing blocking possi
bility,
forced drop-
call p
o
ssibilit
y, and throu
ghput,
which
we
re fu
rthe
r compa
r
e
d
with that of
static
sp
ectrum
acce
ss. However, the interferen
ce
s b
e
twee
n diffe
rent netwo
rks were not co
nsid
ere
d
in this
model, neith
er wa
s buffe
r queu
e leng
th. As these fa
ctors exist in the reality, modification
is
requi
re
d to co
uple with the
actual
situatio
n.
Referen
ces
[1]
Xi
eQin
g
y
an, Z
eng Qi
ng
’an.
P
e
rformanc
e a
n
a
lysis
of op
port
unistic
s
pectru
m
access in he
teroge
ne
ou
s
wi
re
l
e
ss n
e
t
wo
rks
. Wi
r
e
l
e
s
s
Te
l
e
c
o
m
m
u
nic
a
ti
o
n
s
Sy
m
p
o
s
i
u
m (
W
TS
)
.
N
e
w
Y
o
r
k
C
i
ty
, N
Y
: IEEE
Press. 2011: 1-
5.
[2]
Kaig
ui Bi
an,
Jung-Mi
n “Jer
r
y
” Park,
Xi
a
o
jia
ng
Du, et
al.
Ecolo
g
y-
inspir
ed C
oex
istence
of
Hetero
gen
eo
u
s
W
i
reless
N
e
tw
orks
. Global Communic
a
tions
Co
nfer
ence (GLOBECOM), IEEE.
Atlanta, GA: IEEE Press. 201
3: 4921-
49
26.
[3]
Ioann
is T
z
elati
s
, Kostas Ber
beri
d
is.
C
ogn
iti
on-Bas
ed
Dyn
a
mic S
pectru
m
Acc
e
ss
and
Interferenc
e
Coor
din
a
tio
n
for Heterog
ene
o
u
s Netw
orks.
Sign
al Process
i
n
g
Advanc
es in
W
i
reless Com
m
unic
a
tion
s
(SPAWC), IEE
E
14th Worksh
op on. Darms
t
adt: IEEE Press. 2013: 12
5-1
29.
[4]
Han
P, T
i
an H
,
Xi
e W
.
An
aly
s
is of c
o
g
n
itive
rad
i
o s
pectru
m
access
for
emerg
ency c
o
mmu
n
icati
o
n
system
. Wireless Communic
a
ti
ons and Signal Process
i
ng
(WCSP).
Internati
ona
l Co
n
f
erence
on.
Nanj
in
g: IEEE
Press. 2011: 1-
6.
[5]
Di
xit S, Peri
ya
l
w
ar S, Y
anik
o
merogl
u H. S
e
c
ond
ar
y Us
er
Access in
LT
E Architecture
Based
on
a
Base-Statio
n
-C
entric F
r
ame
w
ork W
i
th D
y
n
a
m
ic Pricin
g Ve
hicul
a
r T
e
chno
l
o
g
y
.
Veh
i
cu
lar T
e
chno
logy,
IEEE Transactions on
. 2
013;
62(1): 28
4-2
9
6
.
[6]
T
u
muluru VK,
W
ang P, Ni
yato D, et al. Perform
anc
e ana
l
ysis of cognitive
radio spectru
m
access
w
i
t
h
prioritized traffic.
Vehicul
a
r T
e
chno
logy, IEE
E
T
r
ansactions
on.
201
2; 61(4
)
: 1895-1
9
0
6
.
[7]
Li L, Z
han
g S, W
ang K, et al.
Queuin
g
meth
od in c
o
mbi
n
e
d
chan
ne
l ag
gr
egati
on a
nd fra
g
mentati
o
n
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
0.
8
0.
9
1
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
0.
8
0.
9
1
us
e
r
s
arr
i
v
e
r
a
t
e
A
()
t
h
r
oughput
per
s
e
c
ond
dy
n
a
m
i
c
-
s
t
a
t
i
c
c
o
nj
ug
a
t
ed
u
n
c
onj
uga
t
e
d
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
0.
8
0.
9
1
0
0.
1
0.
2
0.
3
0.
4
0.
5
0.
6
0.
7
us
er
s
ar
ri
v
e
ra
t
e
B
()
t
h
r
o
u
g
h
put
pe
r
s
e
c
o
n
d
dy
na
m
i
c
-
s
t
at
i
c
c
onj
ug
at
ed
unc
o
n
j
u
gat
ed
Evaluation Warning : The document was created with Spire.PDF for Python.