Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
8
,
No.
6
,
D
ece
m
ber
201
8,
pp.
5169
~
51
77
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v8
i
6
.
pp
5169
-
51
77
5169
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
Multi
-
Ch
annel P
reemptiv
e Priorit
y Model
fo
r Spec
trum
Mobility
in Cogn
itive R
ad
io Networks
S.
E.
Saad
1
, I
.
F. Ta
rr
ad
2
, A.
A
.
A
m
mar
3
1
Depa
rte
m
ent of
Elec
tron
ic
s
and Com
m
unic
at
ions
Engi
ne
eri
ng
,
H
igh
Instit
u
te for Engi
n
e
eri
ng
and
Technol
og
y
,
Eg
y
pt
2,
3
El
e
ct
ri
ca
l
Eng
ine
er
ing
Dep
artm
ent
,
Fa
cul
t
y
of
Engi
n
ee
ring
,
Al
-
Azha
r
Univ
ersity
,
Eg
y
p
t
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Ma
r 2
, 2
01
8
Re
vised
Ju
l
2
4
,
201
8
Accepte
d
Aug
11
, 201
8
Cognit
ive
R
adi
o
te
chn
ique
s
hav
e
bee
n
proposed
for
improving
ut
il
izati
on
of
the
spec
trum
b
y
expl
oiting
the
unocc
upi
ed
ban
ds
of
the
li
c
ense
d
spec
trum
.
Thi
s
pape
r
prop
oses
a
pre
emptive
m
ult
i
-
cha
nn
el
ac
c
ess
m
odel
for
priori
tized
cogni
ti
v
e
r
adi
o
n
et
wor
ks using
an
it
er
ative
m
et
hod
of
queui
ng
th
eo
r
y
to
solv
e
the
spec
trum
sca
rcit
y
prob
le
m
.
The
proposed
m
odel
form
ula
te
s
ac
cur
a
te
cl
osed
form
of
a
n
expe
c
te
d
wa
it
i
ng
ti
m
e
in
th
e
q
ueue
,
an
exp
ecte
d
num
ber
of
users
in
the
que
ue,
an
expect
ed
wait
ing
t
ime
in
t
he
s
y
st
em,
and
a
n
expe
c
ted
num
ber
of
users
in
the
s
y
stem.
The
result
s
co
m
par
ed
to
the
basic
m
odel
(without
pre
emptive
prio
rity
)
show
tha
t,
the
wai
ti
ng
ti
m
e
in
qu
e
ue
and
th
e
wait
ing
ti
m
e
in
t
he
s
y
st
em
compare
d
to
the
basic
m
odel
will
b
e
i
m
prove
d
b
y
92.
99%
and
33
.
15%
respe
ct
iv
ely
for
class
one
sec
ondar
y
user.
The
result
s
al
so
show
tha
t,
t
he
wait
ing
ti
m
e
in
queue
and
the
wait
ing
ti
m
e
in
the
s
y
stem
will
be
improv
ed
b
y
43
.
25%
and
15
.
42%
respe
ctively
for
class
two
sec
ondar
y
users
.
The
proposed
m
odel
inve
st
iga
t
es
the
d
esira
b
le
sche
dule
s
of
primar
y
and
sec
ondar
y
users.
Ke
yw
or
d:
Cognit
ive Ra
dio
Mult
i
-
Chan
nel
Pr
eem
ptive Pr
i
or
it
y
Qu
e
uing
The
ory
Sp
ect
r
um
Mobi
li
t
y
Copyright
©
201
8
Instit
ute of
Ad
v
ance
d
Engi
ne
eri
ng
and
Sc
ie
n
ce
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
S.
E
. S
aa
d
,
Dep
a
rtem
ent o
f
Ele
ct
r
on
ic
s
a
nd Com
m
un
ic
at
ion
s E
nginee
ring,
High
In
sti
tute
f
or Enginee
rin
g an
d
Tec
hnol
ogy,
Al
-
O
bo
ur
,
Kilo
21 Cai
ro / Be
lbies R
d
, E
gypt
-
P.
O
. B
ox
27
O
bour Ci
ty
, Eg
y
pt
.
Em
a
il
:
s.elsay
e
d8585@
gm
ail.co
m
1.
INTROD
U
CTION
Durin
g
the
la
st
decad
e,
the
wireless
com
m
un
ic
at
ion
s
ha
ve
bee
n
de
ve
lop
e
d.
Th
e
act
ual
sp
ect
ral
occupa
ncy
ov
e
r
so
m
e
fr
equ
e
ncy
bands
stu
di
ed
was
fou
nd
to
be
virtu
al
ly
e
m
pty.
In
O.
W
.
Bel
lo
et
al
a
uthor
s
carried
ou
t
s
pe
ct
ru
m
m
easur
e
m
ent
in
urba
n
an
d
rural
loc
a
t
ion
s,
co
ve
rin
g
ba
nds
of
50
MHz
a
nd
6
G
Hz
[1]
.
T
he
res
ults
show
that,
the
a
ver
a
ge
sp
ect
ra
l
occu
pa
ncy
of
5.08%
an
d
0.1
8%
in
urba
n
and
r
ur
al
loc
at
ion
s
resp
ect
ively
du
rin
g
weekday
s an
d
1.45
% o
n
week
e
nds
f
or
urba
n
locat
io
ns
.
So
,
the
lim
it
e
d
avail
able
sp
e
ct
ru
m
a
nd
the
i
neffi
ci
ency
us
i
ng
of
the
s
pectr
um
hav
e
bec
om
e
on
e
of
the
cu
rr
e
nt
pro
blem
s
in
w
irel
ess
com
m
un
ic
at
ion
s.
C
ogniti
ve r
adio
te
c
hn
i
que
has bee
n propo
sed
as
a s
olu
ti
on to
these
prob
lem
s.
The
c
ogniti
ve
rad
i
o
te
ch
no
l
ogy
is
base
d
on
the
opport
unist
ic
us
age
o
f
the
sp
ect
r
um
by
al
lowing
un
li
cen
sed
us
e
rs
to
ex
plo
it
frequ
e
ncy
ba
nds
of
li
cense
d
use
rs.
O
pport
uni
sti
c
sp
ect
ru
m
acce
ss
involve
s
two
ta
sk
s
of co
gnit
ive
rad
i
o
syst
e
m
: spectru
m
sen
sin
g
a
nd
dyna
m
ic
sp
ect
ru
m
access.
Ma
ny
sp
ect
r
um
sensing
te
ch
niq
ues
ha
ve
be
en
r
ece
ntly
stud
ie
d,
nam
el
y
E
nergy
detect
io
n
[
2],
hybri
d
sp
ect
r
um
sensi
ng
m
et
ho
d
[
3],
disc
rete
m
ark
ov
c
hain
ba
sed
m
et
ho
d
[
4],
ei
gen
value
ba
se
d
sp
ect
r
um
se
ns
in
g
[5
]
, a
nd m
at
ched
filt
er d
et
ect
or an
d
cy
cl
os
ta
ti
on
ary
detect
or [6
]
-
[
7]
.
Ma
ny
dy
nam
i
c
sp
ect
r
um
acce
ss
m
od
el
s
ha
ve
been
stu
di
ed.
I
n
L
.
Chen
et
al
t
he
a
uthors
pr
opos
e
d
M/
M/
1
qu
e
ue
wh
ic
h
is
repre
sented
by
a
t
w
o
dim
ension
al
sta
te
transiti
on
gr
a
ph
[
8
]
.
Th
e
auth
ors
propose
d
a
qu
e
uing
m
od
el
fo
r
heter
og
e
ne
ous
data
tran
sm
issi
on
s
in
unde
rlay
cogniti
ve
rad
i
o
netw
orks.
In
this
m
od
el
,
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
87
08
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
5169
-
5177
5170
safety
or
em
erg
ency
relat
ed
m
essages
poss
ess
pr
eem
ptive
higher
pr
i
or
it
y
ov
e
r
no
nem
erg
ency
m
essages.
This
m
od
el
h
as tw
o cl
asses;
h
ig
h p
rior
it
y f
or prim
ary us
e
r
a
nd lo
w pr
i
or
it
y f
or
s
econda
r
y use
r.
I
n
M.E
.
Ba
y
r
ak
dar
and
A
.
Ça
lhan
non
-
preem
pti
ve
M/
G/1
pri
ori
ty
qu
eui
ng
m
od
el
of
s
pectrum
han
d
-
off
schem
e
was
pr
opos
e
d
in
co
gnit
ive
rad
i
o
ne
tworks
[
9
-
10
]
.
In
T
.
C.
Chu
et
al
,
the
auth
or
s
pro
po
se
d
a
Dynam
ic
Sp
ect
r
um
Access
schem
e
for
co
gn
it
ive
ra
dio
n
et
w
orks,
wh
e
re
pr
io
riti
es
for
the
band
width,
the
s
pe
ct
ru
m
acce
ss,
a
nd
spe
ct
ru
m
han
d
-
off
a
re
c
on
si
dered
f
or
th
ree
ty
pes
of
t
raffic
s
[
1
1
]
.
Pa
pe
r
ha
s
ad
op
te
d
a
m
ul
ti
-
dim
ension
al
M
arko
v
c
hain
wi
th
th
ree
sta
te
va
riables
t
o
a
naly
ze
the
sta
te
tr
ansiti
on
s
of
t
he
d
ynam
ic
sp
e
ct
ru
m
acce
ss
schem
e
wh
ic
h
en
ables
us
to
ob
ta
in
th
e
ste
ady
sta
te
distrib
ution
of
the
nu
m
ber
of
each
kind
of
traff
ic
s
in the sy
stem
.
It
is
ver
y
diffi
cult
to
analy
ze
the
be
ha
vio
r
of
m
ulti
-
channel
of
the
pr
i
or
it
y
queue
m
od
el
whe
n
channel
us
a
ge
by
pr
im
ary
and
seco
ndary
use
rs
[
1
2
].
Alth
ough
the
vast
researc
h
on
th
e
analy
sis
of
pri
or
it
y
qu
e
ues
with
a
sing
le
cha
nne
l
facil
ity,
eff
or
ts
on
the
c
harac
te
rizat
ion
of
m
ulti
chan
nel
qu
e
ues
we
re
no
t
as
extensi
ve
or
fruit
fu
l
due
to
t
heir
c
om
plica
tio
ns.
T
his
is
es
se
ntial
ly
un
f
or
tun
at
e
si
nce
th
e
m
od
el
ing
of
m
any
te
le
com
m
un
ic
a
ti
on
s
prob
le
m
s
can
be
s
uitably
placed
in
t
he
fr
am
ewo
r
k
of
a
m
ulti
chan
nel
syst
em
,
su
ch
as
m
ul
ti
ch
an
nel
cogniti
ve radi
o netw
orks
[
1
3
].
In
this
pa
per
,
m
ul
ti
-
channel
pr
eem
ptive
pri
or
it
y
m
od
el
ba
sed
on
qu
e
uing
the
ory
ha
s
be
en
pro
po
s
ed
as a s
olu
ti
on to
th
is
prob
le
m
.
The
m
ai
n
co
ntr
ibu
ti
ons
of this
p
a
per are a
s fo
ll
ow
s:
1.
Q
ue
uing
base
d spectr
um
access schem
e for pr
i
or
it
iz
ed
c
og
niti
ve
ra
dio net
works
is
pro
posed
.
2.
An
acc
ur
at
e
a
naly
ti
cal
m
od
el
fo
r
s
pectr
um
acce
ss
queui
ng
base
d
sc
he
m
e
pr
iority
serv
ic
e
d
isc
i
plin
e
is
der
i
ved
.
3.
The p
r
opose
d
m
od
el
can
be
a
pp
li
ed
for m
ult
i
-
pr
i
or
it
y sec
ondar
y
us
e
r.
4.
The p
r
opose
d
m
od
el
can be
ge
ner
al
iz
ed
for
m
ul
ti
-
channels
.
The
rest o
f
this
pap
e
r
is
orga
ni
zed
as
fo
ll
ows
:
In
Sect
io
n
2,
the
pro
posed
m
od
el
has
bee
n
intr
oduce
d.
The
sim
ulati
on
res
ults
of
t
he
pro
po
se
d
m
odel
ha
ve
been
di
scusse
d
in
Se
ct
ion
3.
Finall
y,
Sect
io
n
4
presents
the concl
usi
on
of the
pro
pose
d
m
od
el
.
2.
P
ROP
OSE
D MO
DEL
In
this
sect
io
n,
a
pro
po
se
d
m
od
el
based
on
queui
ng
th
eor
y
ha
s
bee
n
pr
ese
nted
.
S
ym
bo
ls
and
no
ta
ti
ons
us
e
d i
n
this
pa
per
ar
e li
ste
d
in
T
abl
e 1
.
Table
1.
T
he
Desc
riptio
n of S
ym
bo
ls
Sy
m
b
o
l
Descripti
o
n
p
Mean ar
rival
rate
o
f
the
p
ri
m
ar
y
us
er
si
Mean ar
rival
rate
o
f
class i secon
d
ary
u
ser
Mean ar
rival
rate
o
f
all
u
sers
µ
m
e
an
se
rvice r
a
te
Utilizatio
n
f
acto
r
it
’s =
λ
/Cμ
C
n
u
m
b
er
of
chan
n
els in
the syste
m
p
rob
ab
ility
o
f
exactly
n
us
ers in
th
e s
y
ste
m
Av
erage
n
u
m
b
er
o
f
us
ers in th
e sy
ste
m
Av
erage nu
m
b
er
o
f
us
ers in th
e waiti
n
g
buf
f
er
Av
erage waiting
ti
m
e
of
the p
ri
m
ar
y
u
sers in
the syste
m
si
Av
erage waiting
ti
m
e
of
us
ers in th
e
syste
m
Av
erage waiting
ti
m
e
of
us
ers in b
u
f
fer
2
.1
.
Pri
mar
y
User
M
od
e
l
In
this
m
od
el
,
there
a
re
se
ve
r
al
channels
use
d
by
pri
m
ary
us
ers
a
nd
can
on
ly
be
us
e
d
by
seco
nd
a
ry
us
ers
whe
n
t
he
ch
a
nn
el
s
are fr
ee. T
her
e
are t
hr
ee
possi
ble c
ases;
the a
rr
iv
e
d pr
im
ary us
er
m
ay
b
e enc
ount
er:
1.
Ther
e
is a
n
em
pty cha
nn
el
so
;
the
p
rim
ary use
r wil
l be acce
ssed.
2.
Ther
e
isn'
t
an
e
m
pty
channel
an
d
sec
ondary
us
er
i
n
se
rv
i
ce
so
;
the
sec
onda
ry
us
e
r
w
il
l
be
ejected
t
o
wait
ing
buf
fer
and prim
ary us
er accesse
d.
3.
Ther
e
is
neithe
r
a
n
em
pty chan
nel
nor
seco
ndary
us
e
r
in
servic
e s
o; the p
rim
a
ry u
ser
en
te
rs
the
que
ue.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N: 20
88
-
8708
Multi
-
Ch
annel
Preem
ptive
Pri
or
it
y Mo
del f
or S
pectru
m
M
ob
il
it
y in Co
gnit
iv
e Radio
Net
works
(
S. E
. Saa
d
)
5171
Figure
1.
Pr
im
ary Use
r
M
od
e
l
Figure
1 rep
res
ents the
possi
bl
e thr
ee ca
ses
wh
e
n
t
he prim
ary us
e
r has a
rri
ved
.
2.2
.
Seco
nd
ar
y User
M
od
el
The
sec
onda
ry
us
ers
kee
p
se
arch
i
ng
for
c
ha
nn
el
s
t
hat
are
fr
ee
(a
vaila
bl
e
for
us
e
)
at
s
om
e
po
ints
in
tim
e. Th
ere a
re
thr
ee
possi
ble
cases;
the sec
onda
ry
us
er
m
a
y be e
ncou
nter
:
1.
Ther
e
is a
n
em
pty cha
nn
el
so
;
the sec
onda
ry
us
er
w
il
l be
ac
cessed.
2.
Ther
e
is
n'
t
an
e
m
pty
chan
nel
and
the
re
is
lowe
r
pri
ori
ty
seconda
ry
us
er
in
ser
vice
so
;
the
lo
wer
pr
io
ri
ty
seco
nd
a
ry user
w
il
l be e
j
ect
e
d t
o wait
ing
buf
fer
a
nd the
h
i
gher
prio
rity
u
se
r
will
be
acce
ss
ed.
3.
Ther
e
is
neithe
r
an
em
pty
cha
nn
el
nor
l
ow
e
r
pr
i
or
it
y
seco
ndary
us
er
i
n
se
rv
ic
e
s
o;
the
a
r
rive
d
sec
ondary
us
er
en
te
rs
the
qu
e
ue.
Figure
2
.
S
ec
onda
ry
User M
odel
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
87
08
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
5169
-
5177
5172
Figure
2
s
how
s
the
po
ssi
ble
three
cases;
the
secondary
use
r
m
ay
be
encoun
te
r.
Use
rs
ar
e
sel
ect
ed
to
beg
i
n
se
r
vice
in
t
he
order
of
their
pr
i
or
it
y
c
la
sses,
bu
t
on
a
first
com
e
first
ser
ve
d
basi
s
withi
n
eac
h
cl
ass.
Ther
e
a
re
tw
o
basic
pr
i
or
i
ty
discipli
ne
m
od
el
s,
pr
ee
m
pt
ive
pr
i
or
it
ie
s
and
non
-
pr
eem
ptive
pr
ioriti
es.
Pr
eem
ptive
pr
i
or
it
ie
s m
ean th
at
, th
e lo
west
pri
or
it
y us
er
bei
ng serve
d
is
pree
m
pted
w
he
ne
ver
a
h
i
gh
e
r p
rior
it
y
us
er
e
nters
t
he
syst
e
m
.
Non
-
preem
ptive
pri
or
it
ie
s,
a
us
e
r
bein
g
ser
ve
d
cannot
be
pr
ee
m
pted
if
a
higher
pr
i
or
it
y use
r
e
nt
ers
the
queuei
ng syst
em
. I
n
t
his
pap
e
r,
t
he p
reem
ptive p
ri
ori
ty
m
od
el
h
a
s
been sel
ect
ed.
2.3
.
Analy
tical
Model
In
this
m
od
el
,
assum
e
that;
the
a
rr
i
ving
us
ers
a
nd
le
avin
g
us
er
s
of
the
queui
ng
syst
e
m
o
ccu
r
accor
ding
to
t
he
birth
a
nd
dea
th
proces
s.
H
oweve
r,
t
he
te
r
m
birth
re
fer
s
t
o
the
a
rr
ival
of
a
new
us
e
r
int
o
the
syst
e
m
, an
d de
at
h
re
fer
s
to
t
he
d
e
par
t
ur
e
of
a ser
ved use
r.
The
s
olu
ti
on
ha
s the
fo
ll
owin
g
ste
ps
acc
ordi
ng to
t
he gen
er
al
m
et
ho
d desc
ribe
d
in
[
1
4
]:
1.
Ob
ta
in
the
ste
ady stat
e eq
uations g
over
ning
the que
ue.
2.
So
lve
the
e
qua
ti
on
s
f
or
fi
nd
i
ng
out
the
pro
ba
bili
ty
distribu
ti
on
of
qu
e
ue
le
ng
t
h
by:
a)
Iter
at
ive
m
et
ho
d.
b)
Using
ge
ner
at
ing f
unct
ion
s
. c
)
U
sin
g
li
nea
r op
e
rato
rs.
3.
Ob
ta
in
for
m
ula for
Ls, L
q,
Ws
an
d,
Wq
as
shown i
n
T
a
ble
1.
Af
te
r
c
onstr
uc
ti
ng
the
balan
ce
equ
at
io
ns
f
or
al
l
the
sta
te
s
in
te
rm
s
of
the
pro
bab
il
it
ie
s,
this
syst
e
m
o
f
e
qu
a
ti
on
s ca
n be s
ol
ved
.
By
A
pply
ing t
his
procedu
re
yi
el
ds
=
{
1
!
.
(
)
0
0
≤
<
1
!
.
−
.
(
)
0
≥
(1)
A
qu
e
uein
g
m
od
el
is
base
d
on
the
birt
h
a
nd
death
proces
s,
so
the
sta
te
of
t
he
syst
em
n
re
pr
ese
nts
th
e
nu
m
ber
of
us
e
rs
in
the
queue
ing
syst
em
,
the
key
m
easur
es
of
pe
rfor
m
an
ce
fo
r
the
qu
e
uein
g
syst
e
m
(L,
Lq
,
W,
a
nd
Wq) ca
n be
ob
ta
ine
d.
Fr
om
d
efi
niti
on
of
=
∑
(
−
)
∞
=
(2)
By
so
lvin
g
t
his
equati
on w
e
get
=
1
.
!
.
(
)
+
1
(
1
−
)
2
.
0
(3)
It h
as
b
ee
n p
roved that
in
a st
eady sta
te
que
uing syst
em
,
L
=
λ
W
(4)
This e
qu
at
i
on is cal
le
d
Lit
tl
e’s
f
or
m
ula.
Assum
e that t
he
m
ean ser
vice
tim
e is a consta
nt,
1
μ
. I
t t
hen f
ol
lows
t
hat
W
=
W
q
+
1
μ
(5)
These
relat
ionships
are
e
xtre
m
el
y
i
m
po
rtant
because
they
ena
ble
the
fund
am
ental
para
m
et
ers
L
q,
Wq,
L,
an
d
W
to
be
i
m
m
edi
at
el
y
deter
m
in
ed
as
soon
as
on
e
is
f
ound
analy
ti
cal
ly
.
Th
is
relat
ion
is
fo
rt
un
a
t
e
because
s
om
e
of
these
qua
ntit
ie
s
of
te
n
are
m
uch
easi
er
to
fin
d
t
han
ot
he
rs
wh
e
n
a
qu
e
uing
m
od
el
is
so
lve
d
from
b
asi
c p
ri
nc
iples.
2.4
.
Th
e
Av
er
ag
e
W
aitin
g
T
im
e in
t
he
S
ystem
Fo
r
P
reem
pti
ve
pr
i
or
it
y,
wait
ing
ti
m
e
for
pr
im
ary
us
ers
m
us
t
equ
al
wait
in
g
tim
e
fo
r
t
he
corres
pondin
g
on
e
cl
ass
m
od
e
l.
Be
cause
the w
ai
ti
ng
ti
m
es
fo
r
pr
im
ary
us
ers
are
c
om
pletely
un
af
fected by
the
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N: 20
88
-
8708
Multi
-
Ch
annel
Preem
ptive
Pri
or
it
y Mo
del f
or S
pectru
m
M
ob
il
it
y in Co
gnit
iv
e Radio
Net
works
(
S. E
. Saa
d
)
5173
pr
ese
nce of
use
rs
in
the seco
nd
a
ry
cl
asses,
w
ai
ti
ng
ti
m
e
will
be
the
sam
e
for
a
ny
ot
her
va
lues o
f
a
rr
i
val
rates ,
includi
ng
λ
si
= 0. T
he wai
ti
ng
ti
m
e fo
r
prim
ary
u
se
r
ca
n be
form
ulate
d
from
Eq
uations (
3
)
, (
4
),
a
nd
(
5
):
W
p
=
1
[
c
!
(
c
μ
−
λ
)
(
λ
μ
)
c
∑
(
λ
μ
)
n
n
!
+
c
μ
]
[
1
−
λ
p
c
μ
]
c
−
1
n
=
0
+
1
μ
(6)
Wait
ing
ti
m
e fo
r
seco
ndary
use
r
ca
n be calc
ulate
d by an
it
erati
ve pr
oce
dur
e.
Cl
ass
on
e
seco
nd
a
ry
us
e
r
is
c
om
plete
ly
un
aff
ect
ed
by
lowe
r
-
pr
i
or
it
y
cl
asses,
w
hich
ca
n
t
her
e
fore
be
ignore
d
i
n
the
analy
sis.
Let
W
X1
be
the
ex
pected
wait
ing
tim
e
in
the
syst
e
m
of
a
rand
om
arr
ival
in
e
it
her
of
these
pr
im
ary us
ers
a
nd class
one sec
onda
ry u
se
rs:
W
x1
=
1
[
c
!
(
c
μ
−
λ
x1
)
(
λ
x1
μ
)
c
∑
(
λ
x1
μ
)
n
n
!
+
c
μ
]
[
1
−
λ
x1
c
μ
]
c
−
1
n
=
0
+
1
μ
(7)
λ
x1
=
λ
p
+
λ
s1
(8)
S
o
the
prob
a
bi
li
ty
is
that
this
arr
ival
is
pri
m
ary
us
er
λ
p
λ
p
+
λ
s1
a
nd
λ
s1
λ
p
+
λ
s1
,
that
it
is
in
cl
a
ss
on
e
sec
onda
ry
us
er
.
Ther
e
f
or
e,
W
x1
=
λ
p
λ
p
+
λ
s1
W
p
+
λ
s1
λ
p
+
λ
s1
W
s1
(9)
W
s1
=
λ
p
+
λ
s1
λ
s1
W
x1
−
λ
p
λ
s1
W
p
(10)
Fo
r
cl
ass
tw
o
seco
nd
a
ry
us
e
r
:
Let
W
X2
be
the
ex
pected
wait
ing
ti
m
e
i
n
the
syst
em
of
a
rand
o
m
arr
ival i
n
ei
the
r of
t
hese
pr
im
ary us
e
rs,
cl
ass
one sec
onda
ry u
se
rs,
an
d
cl
as
s two sec
onda
r
y user
s:
W
x2
=
1
[
c
!
(
c
μ
−
λ
x2
)
(
λ
x2
μ
)
c
∑
(
λ
x2
μ
)
n
n
!
+
c
μ
]
[
1
−
λ
x2
c
μ
]
c
−
1
n
=
0
+
1
μ
(11)
λ
x2
=
λ
p
+
λ
s1
+
λ
s2
(12)
W
x2
=
λ
p
λ
p
+
λ
s1
+
λ
s2
W
p
+
λ
s1
λ
p
+
λ
s1
+
λ
s2
W
s1
+
λ
s2
λ
p
+
λ
s1
+
λ
s2
W
s2
(13)
W
s2
=
λ
p
+
λ
s1
+
λ
s2
λ
s2
W
x2
−
λ
s1
λ
s2
W
s1
−
λ
p
λ
s2
W
p
(14)
The
wait
ing
ti
m
e
fo
r
cl
ass
t
hr
ee
can
be
de
rive
d
with
t
he
sam
e
pr
oce
dures
,
a
nd
the
oth
e
r
th
ree
par
am
et
ers
can
b
e ea
sil
y form
ulate
d by usi
ng
equati
ons (4
),
and (
5).
3.
SIMULATI
O
N RESULTS
This
sect
io
n
e
valuates
t
he
pro
posed
m
od
el
us
in
g
M
AT
LAB.
Re
s
ults
ha
ve
bee
n
ca
rr
ie
d
out
by
var
yi
ng
t
he
ar
r
ival
rate
of
t
he
pr
im
ary
us
ers
.
The
pro
posed
m
od
el
has
bee
n
e
valuate
d
by
four
m
et
rics
nam
el
y,
exp
ect
e
d
wait
ing
ti
m
e
in
qu
e
ue,
e
xp
e
ct
ed
num
ber
of
us
e
rs
in
que
ue,
e
xp
e
ct
ed
wait
in
g
ti
m
e
in
th
e
syst
em
,
and
exp
ect
e
d nu
m
be
r of
us
e
rs
i
n
t
he
syst
em
. Results ha
ve bee
n ob
ta
ine
d
i
n
t
he
case
of
five
cha
nn
el
s
.
3.1. C
ompari
s
on
between
The Pr
oposed
Model
and T
h
e Basic
Model
Queue
in
[
1
5
]
Au
t
hors
i
n
[
1
5
]
prov
i
de
t
he
descr
i
ption
a
nd
c
om
par
iso
n
of
13
st
ru
ct
ur
e
d
a
nd
sim
ulatio
n
m
od
el
ing
syst
e
m
s
(S
SMS)
.
Str
uct
ur
al
and
sim
ulati
on
m
od
el
ing
syst
e
m
s
are
com
p
ared
to
eac
h
oth
er
.
Str
uctur
al
m
od
el
si
m
ulate
d
in
[
1
5
]
pro
po
se
d
qu
e
uing
m
od
e
l
without
pr
io
r
it
y.
In
this
sub
sect
io
n,
resu
l
ts
ob
ta
in
ed
fro
m
the
pro
po
se
d
m
od
el
will
be
co
m
par
e
d
with
the
resu
lt
s
in
[
1
5
]
.
The
p
aram
eter
s
us
e
d
in
this
si
m
ulati
on
are
the
sam
e as the p
ar
a
m
et
ers
us
e
d
i
n [
1
5
].
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
87
08
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
5169
-
5177
5174
Table
2
.
C
om
par
iso
n of
t
he Pr
opos
e
d
M
od
el
with
t
he
Basi
c
Mod
el
i
n
[
1
5
]
Para
m
eter
Prop
o
sed
M
o
d
el
Res
u
lts o
f
[
1
5
]
PU
SU1
SU2
SU3
Total
Netwo
rk
L
q
0
.00
0
1
9
3
9
0
.00
8
4
0
.06
5
7
0
.27
9
9
0
.35
4
1
9
3
9
0
.35
4
L
0
.75
0
2
0
.75
8
4
0
.81
5
7
1
.02
9
9
3
.35
4
2
3
.35
4
W
q
0
.00
7
7
5
7
2
0
.33
7
4
8
6
9
2
.62
6
7
1
1
.19
7
3
.54
2
2
3
6
3
.54
2
W
3
0
.00
8
3
0
.33
7
3
2
.62
7
4
1
.19
7
3
3
.54
2
2
5
3
3
.54
2
As
s
how
n
in
T
able
2,
total
re
su
lt
s
obta
ine
d
from
the
propo
sed
m
od
el
are
the
sam
e
as
the
res
ults
in
[1
5
]
.
All
m
eas
ur
e
d
tim
e
values
in
un
it
tim
e.
This
si
m
ulati
on
has
bee
n
pe
r
form
ed
to
pr
ov
e
that,
the
pr
opose
d
m
od
el
is v
al
ida
te
d
by c
om
par
ing t
he
t
otal res
ults achie
ved
with the
r
es
ults
of the
c
on
ven
t
ion
al
m
od
el
.
In
I.
Y
aki
m
ov
et
al
a
co
nvent
ion
al
que
uing
m
od
el
witho
ut
pr
io
rity
was
pr
ese
nted
,
w
hi
ch
does
not
m
eet
the
req
ui
rem
ents
fo
r
co
gn
it
ive
ra
dio
ne
tworks
[1
5]
.
This
pa
per
pro
po
s
es
a
m
od
el
fo
r
m
ulti
-
cha
nn
el
,
m
ul
ti
-
cl
asses
cogniti
ve
ra
dio
netw
ork
s
us
i
ng
t
he
preem
ptive
pr
io
rity
m
od
el
based
on
queui
ng
th
eor
y.
A
perform
ance
analy
sis
in
the
m
os
t
gen
e
ral
form
can
be
cond
ucted
by
the
pro
po
se
d
m
od
el
.
I
n
th
e
nex
t
su
bse
ct
ion,
the
pr
op
os
e
d
m
odel
will
be
anal
yz
ed
to
ensure
pr
io
rity
fo
r
pr
i
m
ary
us
ers
an
d
ens
u
re
pri
ori
ty
fo
r
higher
class se
conda
ry u
se
r o
ver lo
wer
class
at
a
di
ff
e
ren
t a
rr
ival
rate
.
3.2. Per
f
orm
ance
Ev
alu
at
io
n
of
t
he
Pro
posed
Mo
d
el
at
Diff
ere
nt Re
q
uest R
at
e
In
t
his
sim
ulati
on,
the
a
rr
i
val
rate
of
the
pr
i
m
ary
us
ers
m
ay
var
y
bet
wee
n
0.0
1
-
4
re
qu
e
st/
sec
.
In
t
his
analy
sis,
the
a
rr
ival
rate
of
t
he
sec
onddary
us
er
s
is
4
re
quest
/se
c
f
or
ea
ch
cl
ass,
a
nd
the
ser
vice
rate
is
4
request/
sec f
or
each c
hannel.
Figure
3
.
Expe
ct
ed W
ai
ti
ng T
i
m
e in
the
Qu
e
ue
Figure
4
.
Expe
ct
ed W
ai
ti
ng T
i
m
e in
the
Syst
e
m
The
pe
rfor
m
ance
in
te
r
m
s
of
wait
ing
ti
m
e
in
bu
f
fe
r
is
sh
ow
n
in
Fig
ur
e
3.
I
t
can
be
obser
ve
d
that,
the
wa
it
ing
tim
e
in
buff
e
r
f
or
prim
ary
us
er
is
the
lo
west
i
n
the
syst
em
.
Re
su
lt
s
sho
w
t
hat,
at
the
poi
nt
of
λ
p
= 4
,
λ
s1
=
λ
s2
=
λ
s3
= 4
r
equ
e
st/
sec;
W
ai
ti
ng ti
m
e fo
r
pri
m
ary
an
d
th
ree classes of
sec
onda
r
y user
is 0
.
24
m
s,
9.
7
m
s,
78
.
6
m
s,
and
465.6
m
s
resp
ect
ively
.
We
can
exam
ine
the
eff
ect
of
pr
iority
schedule
at
this
po
int.
The
wait
in
g
ti
m
e
i
n
the
queue
with
ou
t
pr
io
rity
is
138.5
m
s
,
so
,
the
pri
ori
ty
sched
ule
im
pr
ove
s
the
wait
ing
ti
m
e
in
qu
e
ue
f
or
cl
ass
on
e
sec
onda
ry
us
ers
,
an
d
cl
ass
two
sec
onda
ry
us
ers
by
92.
99%,
an
d
43
.25%
res
pecti
ve
ly
co
m
par
ed
to
the
basic m
od
el
w
i
thout p
rio
rity
.
It
is
i
m
po
rtant
to
note
that,
t
he
ave
ra
ge
of
wait
in
g
ti
m
e
f
or
al
l
use
rs
of
the
pro
posed
m
od
el
at
any
po
i
nt
is
e
xactl
y
equ
al
to
t
he
wait
ing
ti
m
e
in
the
que
ue
without
pr
i
or
it
y
First
Com
e
First
Se
rv
e
(FC
FS)
discipli
ne
at
th
e
sam
e
po
int,
s
o,
th
e
ave
ra
ge
of
wait
ing
tim
e
for
al
l
us
e
rs
at
the
point
λ
p
=
4,
λ
s1
=
λ
s2
=
λ
s3
=
4
request/
sec
i
s
138.5
m
s.
I
t
is
obser
ve
d
th
at
the
wait
in
g
tim
e
in
qu
e
ue
increases
in
a
n
ex
pone
ntial
orde
r
as
the ar
rival
rate
of prim
ary us
ers
inc
rease
d.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N: 20
88
-
8708
Multi
-
Ch
annel
Preem
ptive
Pri
or
it
y Mo
del f
or S
pectru
m
M
ob
il
it
y in Co
gnit
iv
e Radio
Net
works
(
S. E
. Saa
d
)
5175
Figure
4
s
how
s
that,
at
the
point
of
λp
=
4
request/
sec;
w
ai
ti
ng
tim
e
fo
r
pr
im
ary
and
t
hree
cl
asses
of
seco
nd
a
ry
us
e
r
is
25
0.24
m
s,259.7
m
s,
32
8.6
m
s,
an
d
715.6
m
s
respec
ti
vely
.
To
s
tud
y
the
im
pact
of
pr
i
or
it
ie
s,
the
wait
ing
tim
e
i
n
the
syst
em
without
pr
i
or
it
y
is
388.5
m
s
so
,
the
pri
or
it
y
sche
dule
im
pr
ov
e
s
the
wait
ing
tim
e
in
cl
ass
one
seco
nd
a
ry
use
rs,
a
nd
cl
ass
two
seco
ndary
us
e
rs
by
33.
15%,
a
nd
15.
42%
resp
ect
ively
.
As
the
pri
m
ar
y
us
er
arr
i
val
rate
increases,
then
due
to
high
pr
i
or
it
y,
the
pr
im
ary
us
er
ta
kes
th
e
c
hannel
from
s
econda
ry
us
e
r
and
the
sec
onda
ry
us
e
r
will
be
ejected
to
wa
it
ing
bu
ff
e
r.
F
urt
her,
as
t
he
pr
i
m
ary
us
ers
us
e
m
or
e
channels
in
the
syst
e
m
,
then
the
wait
in
g
tim
e
to
the
s
econda
ry
us
er
s
sh
ould
inc
re
ase
as
dep
ic
te
d
in
F
igures
3
a
nd
4.
The
n,
if
the
re
isn'
t
an
em
pty
channel
for
seco
nd
a
ry
use
r
an
d
the
re
is
l
ow
e
r
pr
i
or
it
y
second
ary
us
er
in
se
r
vice
so
;
the
lo
wer
pr
io
rity
seconda
ry
us
er
will
be
ejected
to
wait
ing
buf
fer
an
d
the
higher
pri
ori
ty
seco
nd
a
ry
us
er
will
be
ac
cessed.
Co
ns
e
qu
e
ntly
,
the
lo
wer
pri
ori
t
y
seconda
ry
use
r
will
be
m
or
e aff
ect
ed
by inc
reasin
g
t
he prim
ary us
e
r
ar
rival
rate
.
Figure
5
.
Expe
ct
ed
N
um
ber
of Use
rs
in Wai
ti
ng
Figure
6
.
Ex
pe
ct
ed
N
um
ber
of Use
rs
in
Syst
e
m
As
sho
wn
i
n
F
igure
5,
the
nu
m
ber
of
use
rs
in
the
wait
ing
buf
fer
f
or
pr
im
ary
us
ers
is
lower
tha
n
oth
e
rs
us
er
s.
R
esults
s
how
t
ha
t,
at
the
po
i
nt
of
λ
p
=
4
requ
est
/se
c;
nu
m
ber
of
us
e
rs
in
th
e
wait
ing
bu
ffer
f
or
pr
im
ary
and
t
hree
cl
asses
of
seco
nd
a
ry
use
r
is
0.0
0096,
0.038
8,
0.3
144,
and
1.8
6
res
pe
ct
ively
.
To
st
udy
the
i
m
pact
of
pr
i
ori
ti
es,
the
wait
ing
tim
e
in
the
syst
e
m
witho
ut
pr
i
or
it
y
is
2.2
2
s
o;
the
pr
i
or
i
ty
schedule
im
pro
ves
the
wait
in
g
ti
m
e
in
cl
ass
one
sec
onda
ry
us
ers
,
a
nd
cl
ass
tw
o
sec
ondar
y
us
er
s
by
93%,
a
nd
43.
35%
re
sp
ect
ively
.
Figure
6
s
hows
that,
the
nu
m
ber
of
u
sers
in
t
he
syst
e
m
fo
r
pr
im
ary
us
ers
are
lowe
r
than
oth
e
rs
us
e
rs
.
Re
su
lt
s
sho
w
t
hat,
at
the
poin
t
λp
=
4
request
/se
c;
nu
m
ber
of
use
rs
in
t
he
s
yst
e
m
fo
r
pr
im
ary
an
d
t
hr
ee
c
la
sses
of
sec
ondar
y
use
r
is
1.0
01,
1.0
388,
1.314
4,
an
d
2.8
622
re
sp
ect
ively
.
T
o
stud
y
t
he
im
pact
of
pr
io
riti
es,
the
nu
m
ber
of
us
er
s
in
the
syst
em
without
pri
ori
ty
is
6.
22
so
;
th
e
pr
i
or
it
y
sche
du
le
im
pr
oves
the
num
ber
of
us
er
s
in
the
syst
e
m
in
cl
ass
one
seco
ndary
use
rs,
a
nd
cl
ass
two
sec
onda
ry
u
se
rs
by
33.
2%,
an
d
15
.47%
resp
ect
ively
.
Fr
om
the
per
f
or
m
ance
analy
sis
resu
lt
s,
it
i
s
con
cl
ud
e
d
th
at
,
at
the
arr
ival
rate
of
the
pr
im
ary
us
er
increases
,
the
wait
ing
ti
m
e
a
nd
t
he
nu
m
ber
of
us
e
rs
inc
re
ase
,
but
the
pr
i
or
it
y
of
each
c
la
ss
is
reserve
d.
T
he
wait
i
ng
ti
m
e for pr
im
ary us
ers
are
c
om
plete
l
y un
a
ff
ect
e
d b
y t
he
prese
nce
of u
se
rs
i
n
the
seco
nd
a
ry cla
s
ses.
In
m
os
t
previ
ou
s
stu
dies,
t
he
a
uthors
co
ns
ide
r
that
t
he
seco
ndary
use
r
is
a
sin
gle
cl
ass.
T
his
consi
der
at
io
n
do
e
s
not
m
ee
t
the
nee
ds
of
m
ulti
ple
app
li
cat
ion
s
for
the
us
e
of
c
ogniti
ve
rad
i
o.
T
her
e
for
e,
the
m
od
el
p
rese
nte
d
in
this
pa
per
is a sol
ution t
o t
his problem
.
Com
par
ed wit
h
the
pre
vious
works [
8
-
10
, 1
2
]
, th
e
auth
or
s
hav
e
a
ssu
m
ed
the
c
ogniti
ve
ra
dio
ne
tworks
by
ass
um
ing
the
pro
blem
as
a
sin
gl
e
cha
nn
el
.
H
oweve
r,
these
m
od
el
s
do
not
a
dequa
te
ly
ov
erc
om
e
the
need
to
a
naly
ze
of
c
ogniti
ve
ra
dio
ne
tworks
as
they
ha
ve
m
ul
ti
ple
channels.
T
he
refore
,
the
m
od
el
pr
ese
nted
in
this
pap
e
r
ca
n
be
ef
fici
ently
us
e
d
t
o
a
naly
ze
th
e
beh
a
vior
of m
ulti
-
channel o
f
t
he pri
or
it
y
que
ue
m
od
el
for c
ogniti
ve radi
o netw
orks.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
87
08
In
t J
Elec
&
C
om
p
En
g,
V
ol.
8
, N
o.
6
,
Dece
m
ber
2
01
8
:
5169
-
5177
5176
4.
CONCL
US
I
O
N
In
t
his
pap
e
r,
pr
i
or
it
y
base
d
pr
eem
ptive
qu
euin
g
m
od
el
ha
s
bee
n
pro
posed
for
c
ha
nn
e
l
acce
ss
i
n
cogniti
ve
ra
di
o
netw
orks
.
St
eady
sta
te
analy
sis
of
the
pro
po
s
ed
m
od
el
is
pr
ese
nted
.
Pre
e
m
ption
pri
or
it
y
i
s
util
iz
ed
to
m
eet
requirem
ents
of
the
cl
as
s
one
an
d
cl
ass
tw
o
sec
onda
ry
use
rs.
The
pr
opose
d
m
od
el
give
s
the
com
m
end
able
sche
du
le
s
of
pr
i
m
ary
and
seco
nd
a
ry
us
er
s.T
hi
s
wo
r
k
can
be
extend
e
d
to
e
valuate
a
pr
ee
m
pt
ive
pr
i
or
it
y
discipl
ine
in
eve
nt
sim
u
lator
,
a
nd
c
an
be
e
xten
de
d
to
an
al
yz
e
the
pe
rfor
m
anc
e
in
the
case
of
finite
siz
e o
f
w
ai
ti
ng
buff
e
r.
REFERE
NCE
S
[1]
O.
W
.
Bello,
O
.
A.
Sow
ande
,
S.
O.
Onidar
e
,
M
.
Y.
Muham
m
ada
,
and
A.A.
A
y
e
ni,
“
Large
sc
ale
spec
trum
surve
y
in
rura
l and
urb
an e
nvironments
wit
hin
th
e
50
MH
z
–
6
GH
z
bands”
,
E
lsev
i
er,
M
easurement
Journal
91,
2016.
[2]
F.Z
.
El
B
ahi,
H.
Ghennioui
,
and
M.
Zoua
k
,
”
Per
form
anc
e
Ev
al
u
at
ion
o
f
En
erg
y
Dete
c
tor
Based
Spect
rum
Sensing
for
Cognit
ive
Radi
o
using
NI
US
RP
-
2930”
,
Inte
rnational
Jo
urnal
of
El
e
ct
ri
cal
and
Computer
Engi
ne
ering
(
IJE
CE)
,
vol. 7,
no.
4
,
pp
.
1934
–
1940,
2017.
[3]
A.S.
Khobraga
d
e1,
and
R.
D.
Raut
,
”
H
y
brid
Spect
rum
Sensing
Method
for
Cognit
ive
Radio”
,
Int
ernati
ona
l
Journal
of
Elec
t
rical
and
Computer
Eng
ine
ering
(
IJE
CE)
,
vol. 7,
no.
5,
pp.
2683
–
2695,
2017
.
[4]
Moham
m
adr
ez
a
Am
ini
,
As
ra
M
irz
av
andi
,
Mos
r
afa
Re
zaei
,
“
Discre
t
e
Markov
C
hai
n
Based
Spe
c
trum
Sensing
for
Cognit
ive
Radi
o
”
,
In
te
rnationa
l
Journal
of
El
e
ctr
ic
al
and
Comp
ute
r
Engi
n
ee
rin
g
(
IJE
CE)
,
vol.
5,
no.
2,
pp
.
297
–
30,
2015
.
[5]
S.
S.
Ali,
C.
Liu,
and
M.
Jin
,
“
Minim
um
Ei
genva
lu
e
Det
ect
ion
for
Spe
ct
ru
m
Sensing
in
Cognit
ive
Radio”
,
Inte
rnational
Jo
urnal
of El
e
ct
ri
c
al
and
Comput
er
Engi
n
ee
ring
(
IJE
CE)
,
vol
.
4
,
no
.
4,
p.
623,
2014.
[6]
H.
Sun,
A.
Nallana
th
an,
C.
X
.
W
ang,
and
Y.
C
hen,
“
W
ide
band
spec
trum
sensin
g
for
cogni
ti
v
e
rad
io
net
works
:
a
surve
y
”
,
IEEE Wirel. Commun
.
,
vol
.
20
,
no
.
2
,
p
p.
7481
,
2013
.
[7]
J.
Avila,
and
K.
The
nm
ozhi
,
“
Multi
band
OF
DM
for
Cognit
iv
e
R
adi
o
–
A
W
a
y
f
or
C
y
cl
ost
at
ion
a
r
y
D
et
e
ct
io
n
and
Inte
rfe
r
ence
Ca
n
ce
l
la
t
ion
”
,
Int
ernati
onal
Journal
of
E
lectri
cal
an
d
Computer
Eng
ine
ering
(
IJE
CE
)
,
vol.
6
,
no
.
4
,
pp.
1702
–
1709
,
2016.
[8]
L.
Chen
,
L
.
Huang,
H.
Xu,
and
J.
Hu,
“
Queue
i
ng
Anal
y
sis
for
Pree
m
pti
ve
Tr
an
sm
ission
in
Und
erl
a
y
Cogn
it
iv
e
Radi
o
Networks
”,
Int
ernati
onal
Journal
of
Comm
unic
ati
on
S
yst
ems
,
29:
1138
–
1
155.
doi:
10
.
100
2.
,
John
W
il
e
y
&
Sons
,
Lt
d,
2016.
[9]
M.E
.
B
a
y
rak
d
ar,
and,
A.
Ç
al
han
,
“
Im
proving
spec
trum
handof
f
u
ti
lization
for
pri
orit
ized
cogni
t
iv
e
rad
io
users
b
y
expl
oiting
ch
an
nel
bonding
wi
th
starvation
m
it
igation”,
E
lsevie
r,
Int
ernati
on
al
Journa
l
of
El
e
ct
ronics
an
d
Comm
unic
ati
ons (
AE
Ü)
,
2017.
[10]
M.E
.
B
a
y
rak
d
ar
,
and
,
A.
Çal
h
a
n
,
“
Non
-
Pree
m
pti
ve
Queu
ei
ng
Model
of
Spec
t
rum
Handoff
Scheme
base
d
on
Priorit
ized
Da
ta
Tra
ffi
c in
Cogni
t
ive
W
ireless Netw
orks
”
,
ETRI
Jo
urnal
,
39
(4), 55
8
-
569,
2017
.
[11]
T.
C.
Chu,
H.
Ph
an,
and
H.
Z
epernic
k,
“
D
y
namic
Spect
rum
Ac
ces
s
for
Cognit
iv
e
Radi
o
Network
sW
it
h
Priorit
i
zed
Tra
ffi
cs”
,
IE
EE
communic
ati
ons
le
tters
,
Vol
.
18
,
NO
.
7,
2014
.
[12]
V.
Kum
ar,
S.
M
inz
,
and
Vipin
Kum
ar,
“
Perfor
m
anc
e
anal
y
s
is
of
cogni
ti
v
e
rad
i
o
net
works
under
spec
trum
sharin
g
using que
uing
a
pproa
ch”,
El
se
vier,
Computers a
nd
Elec
tri
cal E
n
gine
ering
52,
20
16.
[13]
Navid
T
aday
on
,
“
Modeli
ng,
D
e
sign
and
Ana
l
ysis
of
Multi
-
Ch
anne
l
C
ognitive
Radi
o
Network
s”,
PH
D
The
sis
,
2016.
[14]
P.
Kanda
sam
y
,
K.
Th
il
ag
ava
t
hi,
and
K.
Gu
nava
th
i
,
“
Proba
bil
ity
s
ta
t
isti
cs
and
Queuing
t
heor
y
”,
S.
Ch
a
nd
&Com
pan
y
LTD
.
,
New D
el
hi
,
2006.
[15]
I.
Yakimov,
A.
Kirpic
hnikov
,
V.
Moks
hin,
Z.
Ya
khina
,
and
R.
Ga
inul
li
n
,
“
The
C
o
m
par
ison
of
Struct
ure
d
Mode
li
ng
and
Sim
ula
ti
on
Modeli
ng
of
Qu
eue
ing
S
y
stems
”,
16th
Inte
rnat
i
onal
Confe
ren
ce,
Information
T
ec
hnolog
ie
s
and
Mathe
mati
cal
M
odel
li
ng
(
ITMM),
Queu
ei
ng
Theo
ry
and
App
lications
,
Springer
,
2
017.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
& C
om
p
Eng
IS
S
N: 20
88
-
8708
Multi
-
Ch
annel
Preem
ptive
Pri
or
it
y Mo
del f
or S
pectru
m
M
ob
il
it
y in Co
gnit
iv
e Radio
Net
works
(
S. E
. Saa
d
)
5177
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
Saad
El
sa
y
ed
is
an
As
sistant
Le
c
ture
r
in
the
High
Instit
ute
for
En
gine
er
ing
and
T
ec
hnolog
y
,
Al
-
Obour,
Cai
ro,
Eg
y
p
t.
He
rece
ive
d
his
BS
c
and
MS
c
in
El
ec
tronics
and
Com
m
unic
at
ion
s
Engi
ne
eri
ng
fro
m
the
Facul
t
y
o
f
Engi
nee
r
ing,
Al
-
Azha
r
Unive
rsit
y
,
Cai
ro
,
Eg
y
pt
,
in
2008
an
d
2015
respe
ct
iv
ely
.
He
is
cur
ren
t
l
y
a
PhD
student
at
Facul
t
y
of
En
gine
er
ing,
Al
-
Az
har
unive
rsit
y
,
Cai
ro. His re
se
ar
ch
a
ct
iv
it
i
es
ar
e withi
n
wir
el
ess
c
om
m
unic
at
ions
and
comm
unic
a
t
ion
ne
tworks.
Ibra
him
Fath
y
Ta
rra
d
re
ceive
d
his
BS
c
and
M
Sc
degr
ee
s
in
El
ectroni
cs
and
Com
m
unic
at
ions
Engi
ne
eri
ng
fro
m
the
Facul
t
y
o
f
Engi
nee
r
ing,
Al
-
Azha
r
Unive
rsit
y
,
Cai
ro
,
Eg
y
pt
,
in
1984
an
d
1989,
respe
ct
iv
e
l
y
.
He
r
ecei
ved
his
PhD
from
th
e
Techni
ca
l
Un
i
ver
sit
y
of
Bud
ap
est
in
1996
.
In
19
96,
he
was
appoi
nt
ed
Lectu
rer
at
the
Dep
art
m
ent
of
E
lectr
oni
cs
and
C
om
m
unic
at
ions
Engi
ne
eri
ng,
A
l
-
Azha
r
Univer
sit
y
.
In
2015,
he
bec
ame
an
As
socia
te
Profess
or
i
n
Com
m
unic
at
ions
Engi
nee
ring
at
the
Facult
y
of
Engi
ne
eri
ng,
Al
-
Azha
r
Univer
sit
y
.
His
rese
arch
act
ivi
ties a
re
wi
t
hin
wire
le
ss
com
m
unic
at
ions
and
digi
t
al c
om
m
unic
a
ti
ons.
Abdelha
d
y
Abd
el
a
zi
m
Am
m
ar
is
a
profe
ss
or
in
El
e
ct
ron
ic
s
and
Com
m
unic
at
io
ns
Engi
ne
eri
ng
depa
rtment
,
Fa
c
ulty
of
Engi
n
ee
r
i
ng,
Al
-
Azh
ar
Un
ive
rsit
y
,
Ca
iro,
Eg
y
p
t
sin
ce
198
8.
His
rese
ar
ch
a
ctivit
i
es
ar
e
with
i
n
digi
t
al
comm
unic
a
ti
ons,
m
obi
le
comm
unic
at
i
ons
and
dig
it
a
l
signal
pro
ce
ss
in
g.
Evaluation Warning : The document was created with Spire.PDF for Python.