Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
10
,
No.
1
,
Febr
uar
y
2020
, pp. 90
0~90
7
IS
S
N:
20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
1
.
pp900
-
907
900
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
Mini
mize MIMO
OFDM
i
nterfe
rence
a
nd
n
oise
r
atio usin
g
polynom
ial
-
time
alg
orithm
Muhame
d K
Husein
D
epa
rtment
o
f
E
l
ectrical
Engi
n
eering,
Ti
kri
t
Univ
ersity
,
Ira
q
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
pr
29
, 201
9
Re
vised
Jun
2
7
,
20
19
Accepte
d
Se
p 27,
2019
In
th
e
d
istri
but
e
d
tr
ansm
it
an
te
n
na
MIM
O
OF
DM
s
y
stem,
e
ac
h
tra
nsm
it
ti
ng
ant
enn
a
has
d
if
fer
ent
fre
qu
ency
offset
b
et
wee
n
each
tr
ansm
it
ting
ant
enn
a
and
r
ecei
ver
du
e
to
the
use
of
ind
epe
nden
t
cr
y
sta
l
oscillat
or
.
Thi
s
p
ape
r
proposes
Pol
y
no
m
ia
l
-
ti
m
e
al
gor
i
thm
for
cor
recti
ng
the
fr
eque
n
c
y
offset
in
a
re
ce
iv
ed
signa
l
b
y
m
axi
m
iz
ing
t
he
condition
al
av
era
ge
signal.
Th
e
al
gori
thm
foc
us
on
r
educi
ng
to
interfe
r
en
ce
and
no
ise
ra
ti
o
of
e
ac
h
subca
rri
er
on
the
r
ecei
ving
a
nte
nna
b
y
fr
eq
uency
offse
t.
T
he
si
m
ula
t
ion
r
esult
show
s
the
per
f
orm
anc
e
of
the
proposed
a
lgori
thm
is
slight
l
y
improved
com
par
ed
wit
h
the
exi
sting
fr
e
quency
o
ffset
c
orre
ction
al
gori
t
hm
,
and
the
co
m
pl
exi
t
y
is
red
uce
d
b
y
50%
or
m
ore
.
Ke
yw
or
d
s
:
Algorithm
BER
MIM
O
OFDM
SN
R
Copyright
©
202
0
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Muh
am
ed
K
Husein
,
Dep
a
rtm
ent o
f El
ect
rical
En
gi
neer
i
ng,
Tikrit
U
nive
rsi
ty
,
Street
of Ti
kr
it
-
Mosil
, Al
-
Qa
dissiy
ah qu
a
rte
r,
0096
42, Tikr
īt
, S
al
la
haldin,
Ir
a
q
.
Em
a
il
:
m
uh
a
m
edm
akh
oodh
@
tu.edu.i
q
1.
INTROD
U
CTION
In
the
MIM
O
OFDM
(Mul
ti
-
Inpu
t
Mult
i
-
Ou
t
pu
t
O
rth
ogon
al
Fr
e
quen
c
y
Divisio
n
Mult
iplexin
g)
syst
e
m
[1
-
3]
,
t
he
diff
e
re
nce
betwee
n
t
he
lo
cal
cryst
al
os
c
il
la
tors
of
eac
h
dis
tri
bute
d
transm
it
antenna
a
nd
the
existe
nce
of
m
ulti
ple
dopple
r
s
hifts,
ca
usi
ng
dif
fer
e
nt
f
reque
ncy
offse
ts
betwee
n
eac
h
distrib
uted
tr
ansm
i
t
anten
na
to
t
he
r
ecei
ver
[4]
,
int
er
-
ca
rr
ie
r
i
nterference
(I
C
I)
a
nd
syst
em
per
for
m
ance
de
gr
a
da
ti
on
.
In
rece
nt
ye
ars,
this
te
ch
no
l
ogy
has
al
s
o
be
en
gr
a
dual
ly
app
li
ed
to
MI
MO
O
FD
M
s
yst
e
m
s
m
ai
nly
f
o
cus
on
ant
enn
a
sel
ect
ion
[5
-
8]
.
An
te
nn
a
sel
ect
ion
f
or
MIM
O
-
OFDM
syst
e
m
s
is
based
on
s
ub
ca
rr
ie
rs
an
d
su
bsy
ste
m
s.
In
order
to
c
om
bat
m
ulti
ple
fr
e
quenc
y
offsets,
the
li
te
ratur
e
[9
-
11]
ad
opts
t
he
e
qual
iz
at
ion
m
et
ho
d
to
el
im
inate
the
f
reque
ncy
offset,
but
the
perform
ance
o
f
this
m
et
ho
d
will
degra
date
with
inc
reasi
ng
the
off
set
f
re
qu
e
ncy.
In
[
4]
,
a
fr
e
qu
ency
offset
c
or
recti
on
al
go
rithm
fo
r
distrib
uted
tra
ns
m
it
anten
na
MIM
O
OFDM
is
pro
posed
,
to
i
m
pr
ove
the
perform
ance
of
t
he
eq
ualiz
at
ion
-
base
d
f
re
quency
offset
el
i
m
inati
on
m
e
t
hod,
w
hich
c
orrect
s
m
ul
ti
ple
fr
e
qu
e
ncy
offsets
befor
e
eq
ualiz
at
io
n
a
nd
reduces
the
i
nterf
e
re
nce
cause
d
by
f
requen
cy
offset
[12
-
15]
.
The
lo
wer
lim
i
t
of
the
c
onditi
on
al
ave
ra
ge
si
gn
al
to
inter
fere
nce
a
nd
noise
rati
o
(
SINR,
Sign
al
-
I
nterf
e
r
ence
-
No
ise
-
Ra
ti
o)
of
the
s
ub
ca
rr
ie
rs
on
the
a
ntenn
a
is
t
he
c
rite
rio
n
c
orrecti
on
fr
e
quency
of
fset,
but
it
has
tw
o
disad
va
ntages.
the
first,
wh
e
n
the
fr
e
quency
dev
ia
ti
on
bet
w
een
the
tr
a
ns
m
it
ti
ng
anten
nas
increases
,
the
lowe
r
lim
it
of
the
c
onditi
onal
ave
ra
ge
S
IN
R
becom
es
m
or
e
an
d
m
or
e
sla
ck
[4,
16,
17]
.
At
this
tim
e,
the
fr
e
qu
e
ncy
offset
is
co
rr
ec
te
d
by
m
axi
m
i
zi
ng
the
lo
wer l
i
m
it
of
the
co
nd
it
io
nal
a
ver
a
ge
SINR.
Co
ndit
ion
al
a
ve
rage
SIN
R
will
pro
duce
a
certai
n
perf
orm
ance
loss;
th
e
seco
nd,
the
a
lgorit
hm
need
s
to
pe
rfor
m
m
or
e
trig
onom
etr
ic
an
d
inv
e
rse
tri
gono
m
et
ric o
pe
rati
ons, a
nd t
he ov
e
rall
co
m
plexity is h
i
gh
e
r
[18,
19
]
.
In this
pa
per a
low
c
om
plexity
f
re
qu
e
ncy
offset co
rrec
ti
on
al
gorithm
p
rop
os
e
d wh
ic
h dir
ect
ly
co
rr
ect
the
f
reque
ncy
offset
by
m
axim
iz
ing
the
c
onditi
on
al
a
ver
a
ge
SINR
of
s
ub
carriers
on
eac
h
re
cei
ving
a
nt
enn
a
.
In
ad
diti
on,
a
na
ly
ses
the
rece
ive
a
nten
na
sel
ect
ion
a
nd
si
gnal
processi
ng
a
lgorit
hm
s
base
d
on
MIM
O
-
O
FD
M
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Mi
nimize M
IM
O OFD
M inter
fe
rence
and n
oi
se ra
ti
o usi
ng
po
ly
nomial
-
ti
m
e a
lg
or
it
hm
(
M
uhamed
K
Hu
s
ei
n
)
901
syst
e
m
s.
By
us
i
ng
the
f
unct
io
n
in
the
c
onditi
onal
a
ver
a
ge
SINR
e
xpressi
on
Po
ly
nom
ial
ap
pro
xim
a
ti
on
,
de
riving
low
-
com
plexity
fr
e
quency
of
fset
c
orrecti
on
analy
ti
c.
Com
par
e
d
with
the
li
te
ratur
e
[
4]
,
the
pe
rfor
m
ance
of
the pr
opos
e
d
al
gorithm
is sli
gh
tl
y im
pr
oved
, a
nd the c
om
plexity
is reduce
d by 50%
or m
ore.
2.
MI
MO
-
OF
D
M
S
YS
TE
M
S
TRU
CTU
RE
AND CH
A
NNE
L MO
DEL
Wh
e
n
t
he
s
ubc
arr
ie
r
-
ba
sed
se
le
ct
ion
m
et
ho
d
is
ad
opte
d,
in
the
a
nten
na
s
ub
s
et
sel
ect
ion
proces
s
of
each
s
ub
ca
rr
ie
r
,
a
sin
gle
carrie
r
MIM
O
syst
e
m
is
act
ually
fa
ced,
s
o
the
a
nte
nn
a
sel
ect
ion
m
et
ho
d
of
t
he
e
xisti
ng
MIM
O
syst
em
c
an
be
i
ntrod
uc
ed
into
t
he
MIM
O
-
A
nten
na
sel
ect
ion
in
O
FD
M
syst
em
s.
Since
the
CS
I
is
fed
back
to
th
e
tra
ns
m
itter
wh
e
n
the
tran
sm
i
t
antenn
a
is
sel
ect
ed,
t
his
is
not
feasible
wh
e
n
the
cha
nnel
ch
ang
e
s
rand
om
l
y,
so
t
he
receive
a
nte
nn
a
sel
ect
ion
is
m
or
e
at
tract
ive.
I
n
the
or
y,
the
c
ho
ic
e
of
th
e
receivi
ng
ant
enna
will
reduce
t
he
ra
nk
of
t
he
c
ha
nn
el
m
at
rix,
wh
ic
h
will
ine
vitably
le
ad
t
o
a
de
crease
in
the
c
hannel
ca
pacit
y.
If
the
c
orres
pondin
g
sig
nal
com
bin
ing
proces
sin
g
al
gorithm
is
com
bin
e
d
af
te
r
t
he
anten
na
sel
ect
ion
,
this pe
rfor
m
ance loss
ca
n be
com
pen
sat
ed.
Figure
1
is
a
blo
ck
dia
gr
am
of
the
ante
nna
sel
ect
ion
at
the
re
cei
vin
g
en
d
of
the
MIM
O
-
O
F
DM
syst
em
.
The
sig
nal
rec
ei
ved
by
the
a
nten
na
sel
ect
s
an
a
nte
nn
a
to
be
dem
odulate
d
by
a
s
pecifi
c
ante
nna
sel
e
ct
ion
al
gorithm
,
and
the
n
rec
ove
rs
the
in
f
or
m
at
ion
of
the
so
urce
th
r
ough
FFT
,
pa
rall
el
string
c
onve
rsion
,
dem
od
ulati
on, a
nd the lik
e.
Figure
1
.
Bl
oc
k diag
ram
o
f
a
nten
na
sel
ect
io
n
at
the
r
ecei
vin
g en
d o
f
MIM
O
-
OFDM sy
st
e
m
2.1.
Tr
an
smit
tin
g si
gn
al
Investi
gate
a
MIM
O
OFDM
syst
e
m
us
ing
t
he
MT
root
-
dis
tribu
ti
on
tra
nsm
it
antenn
a
a
nd
the
MR
roo
t
-
con
ce
ntrate
d
receive
a
nte
nna.
Assum
e
that
the
num
ber
of
OFD
M
subcar
riers
is
K
,
so
(
=
1
,
2
,
…
.
,
;
=
1
,
2
,
…
.
,
)
that t
he
i
nform
at
ion
sym
bo
l carrie
d by the
kt
h
subca
rr
ie
r
on the tra
ns
m
it an
te
nna
m
.
K
-
po
int
fa
st
Fo
uri
er
in
ve
rse
a
fter
t
ransform
ing
an
d
insertin
g
the
cy
c
li
c
pr
efix
(
CP,
Cy
cl
ic
Pr
efix),
at
the d
isc
rete t
i
m
e l, the tim
e d
om
ai
n
sig
nal
on the t
ran
sm
itti
ng
a
nten
na
m
is.
(
)
=
1
√
∑
2
/
−
1
=
0
−
≤
≤
−
1
(1)
Wh
e
re
N
g
is
t
he
l
e
ng
t
h
of
t
he
CP
an
d
the
Ng
is
eq
ual
t
o
or
great
er
tha
n
t
he
su
m
of
t
he
m
axim
u
m
relat
ive
pro
pag
at
io
n de
la
y between
ea
ch
tra
ns
m
it
an
te
nn
a
and t
he
m
axim
u
m
m
ult
ipath d
el
ay
of th
e cha
nn
el
[
2]
.
2.2.
Frequenc
y de
viation
model
Du
e
t
o
t
he
dist
rib
ution
of
th
e
transm
itti
ng
a
nt
enn
as
a
nd
t
he
con
ce
ntrati
on
of
the
receivi
ng
ante
nn
as
,
the
distrib
uted
tra
ns
m
i
tt
ing
anten
nas
us
e
t
heir
res
pect
ive
local
c
rysta
l
os
ci
ll
at
ors,
a
nd
the
c
on
ce
nt
rated
receivin
g
ante
nn
a
s
a
re
us
e
d,
a
l
ocal
os
ci
ll
at
or
[
4]
.
T
her
e
f
or
e
,
this
pap
e
r
ass
um
es
that
the
f
re
qu
e
ncy
offset
betwee
n
the
sa
m
e transm
i
t antenn
a a
nd the
r
ecei
ver
a
nten
na
s of the recei
ve
r
is eq
ual, a
nd
the freque
ncy
offset
betwee
n
diff
e
r
ent
tra
ns
m
i
t
antenn
as
a
nd
the
sam
e
receive
anten
n
a
m
ay
be
dif
fer
e
nt.
Let
ε
m
be
the
norm
al
iz
ed
fr
e
qu
e
ncy
offs
et
betwee
n
the
transm
it
antenn
a
m
and
t
he
re
cei
ver
a
nten
na
s
of
the
receive
r
(t
he
rati
o
of
t
he
tr
ue
fr
e
qu
e
ncy
off
s
et
to
the
subca
rr
ie
r
s
pacin
g).
Si
m
il
ar
to
the
li
te
ratur
e
[
4]
,
this
pa
per
a
ssum
es
that
in
the
sam
e
OFDM
sym
bo
l
per
i
od,
in
t
he
s
yst
e
m
.
The
dif
f
eren
ce
betwe
e
n
the
m
ax
i
m
um
fr
eq
ue
ncy
of
fset
an
d
the
m
i
nim
u
m
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
1
,
Febr
uar
y
2020
:
90
0
-
907
902
fr
e
qu
e
ncy
offs
et
is
no
t
m
or
e
tha
n
half
th
e
su
bc
ar
rier
sp
a
ci
ng
,
ie
.
F
ur
t
her
m
or
e
,
assum
ing
that
ε
m
re
m
a
ins
co
nst
ant
within
on
e
OFDM
sym
b
ol,
it
can
ra
ndom
ly
var
y
betw
een
O
FD
M
sy
m
bo
ls
and can
b
e
acc
ur
at
el
y est
im
a
t
ed by t
he recei
ver
[
20
]
.
2.3.
Receivin
g
si
gnals
It
is
assum
ed
that
the
MIM
O
c
ha
nn
el
is
sp
at
ia
ll
y
un
c
or
relat
ed
a
nd
ha
s
th
ree
cha
rac
te
risti
cs
of
m
ul
ti
path
Ra
yl
ei
gh
fa
ding,
pa
th
los
s,
an
d
s
hado
w
fad
i
ng.
Me
an
wh
il
e,
a
ssu
m
ing
that
the
c
ha
nn
el
re
m
ai
ns
un
c
ha
ng
e
d
within
on
e
OFD
M
sy
m
bo
l,
th
e
fr
e
qu
e
ncy
of
the
kt
h
s
ub
ca
rr
ie
r
on
the
re
cei
vin
g
ante
nna
n
is
assum
ed.
T
he do
m
ai
n
sig
nal
is:
=
∑
√
Λ
,
0
=
1
+
∑
∑
√
Λ
,
+
−
1
=
0
,
≠
=
1
(2)
Wh
e
re
,
is
th
e
f
reque
ncy
dom
a
in
res
ponse
of
the
sm
al
l
-
scal
e
m
ul
ti
path
Ra
yl
ei
gh
fa
ding
on
the
s
ub
ca
rr
i
e
r
k
betwee
n
the
transm
itti
ng
a
nt
enn
a
m
and
t
he
receivi
ng
a
nten
na
n,
a
nd
n
is
a
zer
o
-
m
ean
c
om
plex
G
aussian
rand
om
var
ia
bl
e
with
,
a
va
ria
nce o
f
1
;
P
m
is t
he
tra
ns
m
itti
ng
a
nten
na
m
to
the r
ecei
vi
ng Th
e
la
rg
e
-
scal
e
fad
i
ng co
e
ff
ic
i
ent b
et
ween m
achines
, whic
h cha
racteri
zes
path
l
os
s a
nd s
hado
w fadi
ng
;
Λ
,
=
sin
(
(
+
)
)
sin
(
(
+
)
/
)
(
−
1
)
(
+
)
)
is t
he
IC
I
c
oeff
ic
ie
nt
[21]
;
=
∑
√
Λ
,
0
=
1
is t
he
ef
fecti
ve data
it
em
;
=
∑
∑
√
Λ
,
−
1
=
0
,
≠
=
1
is t
he
IC
I
inte
rference;
is t
he
z
ero
-
m
ean com
plex
Gau
ssia
n wh
it
e
no
ise
with
2
va
riance.
3.
PROP
OSE
D MET
HO
D
The
receive
r
m
ult
ipli
es
the
receive
d
sign
al
of
eac
h
receivin
g
anten
na
by
exp
(
−
2
̃
)
the
fr
e
quen
cy
offset
co
rr
ect
ion
si
gn
al
in
the
tim
e
do
m
ai
n
to
co
rrec
t
the
fr
e
quen
c
y
of
f
set
,
w
here
̃
is
the
norm
al
iz
ed
f
reque
ncy
offs
et
co
rr
ect
io
n
va
lue
on
t
he
rec
ei
vin
g
a
nten
na
n
[
4]
.
T
he
pr
opos
e
d
fr
e
quency
off
set
correct
ion
al
go
rithm
fo
r
m
axi
m
iz
ing
the
lo
w
er
bo
und
of
t
he
conditi
onal
av
erag
e
SINR
(
P
m
,
an
d
̃
based
on
the
co
n
diti
on
al
ave
rag
e
SINR
)
of
t
he
s
ubcar
riers
on
eac
h
r
ecei
vin
g
a
nten
na
,
t
his
pap
e
r
directl
y
m
axi
m
iz
es
the
co
ndit
ion
al
ave
rag
e
SINR
of
the
subca
rri
ers
on
ea
ch
r
ecei
vin
g
a
nten
na.
F
or
t
he
ta
r
get,
a
fter
c
orre
ct
ing
the
f
re
qu
e
ncy
offset
a
nd
f
requen
cy
offset
c
orrect
ion,
t
he
fr
e
qu
e
ncy
do
m
ai
n
sign
al
of
the
kt
h
s
ubca
rr
ie
r
on
the r
ecei
ving a
nten
na n is:
(
̃
)
=
(
̃
)
+
(
̃
)
+
(
̃
)
(3)
w
he
re
(
̃
)
=
∑
√
Λ
,
0
(
̃
)
=
1
a
nd
(
̃
)
=
∑
∑
√
Λ
,
(
̃
)
−
1
=
0
,
≠
=
1
is t
he
ef
fecti
ve si
gn
al
a
nd I
C
I i
nterf
e
ren
ce
aft
er th
e
freq
ue
nc
y offset
c
orrecti
on
;
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Mi
nimize M
IM
O OFD
M inter
fe
rence
and n
oi
se ra
ti
o usi
ng
po
ly
nomial
-
ti
m
e a
lg
or
it
hm
(
M
uhamed
K
Hu
s
ei
n
)
903
Λ
,
(
̃
)
=
sin
(
(
+
−
̃
)
)
sin
(
(
+
−
̃
)
/
)
is t
he
zer
o
m
ean c
om
plex
Ga
ussi
an wh
it
e
no
i
se
with
2
var
ia
nc
e.
Assum
ing
tha
t
the
zer
o
m
ea
n
is
in
de
pende
nt
an
d
i
den
ti
ca
ll
y
distribu
te
d
rand
om
var
ia
bl
e
with
var
ia
nce
of
1,
(
̃
)
the con
diti
on
al
m
ean an
d
c
on
diti
on
al
var
ia
nc
e are
[4]
:
[
(
̃
)
|
,
,
̃
]
=
0
(4)
[
(
̃
)
|
,
,
̃
]
=
∑
(
1
−
(
sin
(
(
+
−
̃
)
)
sin
(
(
+
−
̃
)
/
)
)
2
)
=
1
(5)
Th
us
, t
he
k
th
s
ubcar
rier
on t
he
receivin
g
a
nten
na n is t
he
c
onditi
on
al
a
ver
a
ge
SINR
of
:
(
̃
)
=
[
(
̃
)
|
2
|
,
,
̃
0
]
[
(
̃
)
|
2
|
,
,
̃
0
]
=
∑
2
(
−
̃
)
2
=
1
∑
(
1
−
2
(
−
̃
)
)
+
2
2
=
1
(6)
Since
t
he
num
ber
of
s
ub
ca
rr
i
ers
K
is
m
uch
la
rg
e
r
t
han
(
−
̃
)
,
s
o
(
(
−
̃
)
)
≈
(
−
̃
)
[
22]
,
the
n
(6)
ca
n be
redu
ced to
:
(
̃
)
=
∑
2
(
−
̃
)
=
1
∑
(
1
−
2
(
−
̃
)
)
+
2
=
1
(7)
Am
on
g
t
hem
(
)
=
{
sin
(
)
≠
0
1
=
0
(8)
Fr
om
(
7)
th
at
,
(
̃
)
re
ga
rd
le
ss
of
the
s
ubcar
rier
la
bel
k
,
the
c
onditi
on
al
a
ve
rag
e
S
IN
R
of
each
su
bc
ar
rier
on the sam
e receiv
ing
a
nten
na
is
equ
al
.
F
or co
nvenie
nce,
le
t:
(
̃
)
=
(
̃
)
(9)
Th
us
,
n
the
op
tim
a
l
fr
e
qu
e
nc
y
offset
c
orrec
ti
on
value
that
m
axi
m
iz
es
the
co
ndit
ion
al
a
ver
a
ge
SINR
of
eac
h
su
bc
ar
rier
on the
receivin
g
a
nt
enn
a
̃
,
can
be e
xpresse
d
as:
̃
,
=
̃
(
̃
)
=
1
,
…
.
,
(10)
Sinc
e
(
̃
)
ab
ou
t
i
nc
rem
enting
∑
2
(
−
̃
)
=
1
, (1
0) can
b
e
sim
pl
ifie
d
to:
̃
,
=
∑
2
(
−
̃
)
=
1
(11)
Fr
om
(11)
,
̃
,
it
i
s
diff
ic
ult
to
so
lve
the
pro
ble
m
directl
y.
The
ap
pro
xim
ate
po
ly
nom
ia
l
of
the
functi
on
2
(
)
is use
d
t
o
s
olve
the
pro
blem
.
Sele
ct
the
appr
ox
im
at
e
po
ly
no
m
ial
as
a
qu
a
dr
at
ic
poly
no
m
ia
l,
us
ing
La
gr
a
ngia
n
inter
po
la
ti
on
[
23
]
,
then si
nc
2
(
x) in
the inte
rv
al
[
-
0.5, 0.5]
ca
n b
e ap
pro
xim
a
ted
as:
2
(
)
≈
∑
2
(
)
2
=
0
(
)
−
0
.
5
≤
≤
0
.
5
(12)
Am
on
g
t
hem
,
(
)
∏
(
−
)
(
−
)
2
=
0
,
≠
=
1
2
cos
(
(
2
+
1
)
6
)
=
0
,
1
,
2
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
1
,
Febr
uar
y
2020
:
90
0
-
907
904
Af
te
r
the
opera
ti
on
, t
he
(12) c
an be
wr
it
te
n
a
s:
2
(
)
≈
1
−
2
−
0
.
5
≤
≤
0
.
5
(13)
Wh
e
re a=
2.5
771.
Substi
tuti
ng (1
3) into
(
11), t
he
re ar
e:
̃
,
=
̃
∑
2
(
−
̃
)
=
=
1
̃
(
̃
)
(14)
Wh
e
re
(
̃
)
=
∑
(
1
−
(
−
̃
)
2
)
=
1
.
The
f
irst
de
rivati
ve
of
(
̃
)
the
ord
er
is
e
qu
al
t
o
ze
ro,
the stag
natio
n po
i
nt of
(
)
can
be f
ound as:
̃
,
0
=
∑
/
∑
=
1
=
1
(15)
Substi
tuti
ng
̃
,
0
into of
(
̃
)
, th
e sec
on
d
der
i
vative
with:
2
(
̃
)
̃
2
=
−
2
∑
<
0
=
1
(16)
Ther
e
f
or
e,
(
̃
)
rea
ch
the
m
axi
m
um
v
al
ue
at
̃
=
̃
,
0
.
̃
,
≈
̃
,
0
=
∑
/
∑
=
1
=
1
(17)
Fr
om
(17),
it
is
known
that
̃
,
is
ind
e
pe
nd
e
nt
of
the
receivin
g
anten
na
nu
m
ber
n
,
s
o
the
opti
m
al
fr
eq
uen
c
y
offset c
orrecti
on
value o
n
eac
h recei
ving a
nt
enn
a
is the
sam
e, that is,
̃
1
,
=
̃
2
,
…
.
=
̃
,
=
̃
,
≈
∑
/
∑
=
1
=
1
(18)
4.
RESU
LT
A
N
D DIS
CUSSI
O
N
The
sim
ulati
on
par
am
et
ers
as
fo
ll
ows:
2
tran
sm
it
2
receive,
QP
S
K
m
od
ula
ti
on
,
in
f
or
m
at
i
on
sym
bo
ls
are
in
depen
de
nt
of
each
ot
her;
OF
DM
s
ubca
rr
ie
r
num
ber
is
K
=128,
CP
le
ngth
is
Ng
=
32,
su
bc
ar
rier
s
pac
ing
is
20
kH
z;
sm
al
l
scal
e
fad
i
ng
betwee
n
pairs
of
tran
scei
ve
r
a
nten
nas
T
he
ch
ann
el
s
are
in
de
pende
nt
of
eac
h
oth
e
r
and
are
m
od
el
le
d
as
a
t
wo
-
pat
h
gain
-
Ra
yl
ei
gh
ch
an
ne
l
[24]
with
a
two
-
path
s
pa
ci
ng
of
392.6
23ns;
the
no
rm
alized
fr
e
quency
off
set
ε
m
ob
ey
s
a
unif
or
m
distribu
ti
on
(ie,
~
[
,
]
,
,
th
e
value
s
are
gi
ven
belo
w),
an
d
and
̃
are
ind
ep
en
de
nt
of
eac
h
othe
r
at
≠
̃
;
the
la
rg
e
-
scal
e
fad
i
ng
c
oeffici
ent
Pm
r
e
m
ai
ns
const
ant
durin
g
the
sim
ulati
o
n,
a
nd the
r
ecei
ver ad
opts zer
o
-
f
or
ci
ng d
et
ect
ion.
Figure
2
s
how
s
the
case
wh
e
re
the
sy
m
bo
l
error
rate
var
ie
s
with
the
f
reque
ncy
offset
co
rrec
ti
on
val
ue
ε
2
o
f
t
he
recei
ving
a
nten
na
2
in
t
wo
c
ases
,
w
her
e
SI
R
=
30dB,
[
-
0.2,
0.2].
I
n
case
1,
the
f
reque
ncy
offse
t
correct
ion
valu
e
of
the
receiv
ing
a
nten
na
1
i
s
fixe
d
as
the
su
m
po
int,
w
he
re,
case
2is
th
e
op
ti
m
al
fr
equ
enc
y
offset
c
orrecti
on
value,
p
=1
,...,
5.
Fig
.
1
s
h
ows
t
hat
the
er
r
or
sym
bo
l
rate
in
both
sce
nari
os
is
t
he
sm
al
le
st
at
the tim
e, which indic
at
es that
the
nu
m
erical
r
esults a
re c
onsist
ent w
it
h
t
he
sim
ulatio
n res
ults.
Figure
2
.
The
re
la
ti
on
sh
i
p between
sym
bo
l err
or r
at
e a
nd fr
equ
e
ncy c
orrec
ti
on
s
receive
d by ante
nn
a
2
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Mi
nimize M
IM
O OFD
M inter
fe
rence
and n
oi
se ra
ti
o usi
ng
po
ly
nomial
-
ti
m
e a
lg
or
it
hm
(
M
uhamed
K
Hu
s
ei
n
)
905
Figure
3
sho
ws
the
c
om
par
iso
n
of
the
sub
-
ca
rr
ie
r
t
heoreti
cal
conditi
on
al
a
ver
a
ge
sig
nal
-
to
-
interfe
ren
ce
r
at
io
(S
IR
)
af
te
r
us
in
g
t
he
pro
po
se
d
al
gorithm
,
to
correct
the
f
re
qu
e
ncy
off
set
(11).
Her
e
, th
e
the
oret
ic
al
co
ndit
ion
al
av
e
ra
ge
S
I
R i
s
de
fine
d
as
:
=
[
(
̃
)
|
2
|
,
,
̃
0
]
[
(
̃
)
|
2
|
,
,
̃
0
]
=
∑
2
(
−
̃
)
2
=
1
∑
(
1
−
2
(
−
̃
)
)
2
=
1
(19)
Wh
e
re,
̃
is valu
e of the
freq
ue
ncy off
set
corr
ect
ion
.
Figure
3
s
how
s
that
the
S
IR
curve
of
t
he
propose
d
al
go
rithm
agr
ees
wel
l
with
the
i
dea
l
al
go
rithm
,
wh
il
e
the
SI
R
curve
of
t
he
existi
ng
al
gorithm
is
gr
ad
ually
lower
t
ha
n
2
−
1
the
ideal
al
gorithm
as
the
fr
e
qu
e
ncy
dev
ia
ti
on
bet
w
een
t
he
t
wo
tr
ansm
itti
ng
a
nte
nn
a
s
i
ncr
ease
s.
T
hi
s
is
beca
us
e
with
the
in
crease
,
the
co
ndit
ion
2
−
1
l
ow
e
r
li
m
i
t
of
t
he
a
ver
a
ge
SI
R
will
be
m
or
e
a
nd
m
or
e
relaxe
d
[
4]
.
At
this
ti
m
e,
correct
in
g
the
f
reque
ncy
offset
(t
he
e
xis
ti
ng
al
gorithm
)
with
the
c
rite
rion
of
m
axi
m
iz
i
ng
the
co
ndit
ion
al
av
era
ge
S
I
R
will
br
i
ng m
or
e ob
vious e
rrors.
Figure
3
.
T
he s
ub
ca
rr
ie
rs
a
f
te
r
correcti
ng
fr
e
qu
e
ncy
offset
Figure
4
c
om
par
es
the
error
rate
of
the
pro
posed
al
go
rit
hm
w
it
h
the
existi
ng
al
gorithm
,
wh
e
re
P
1
/P
2
=3
dB,
~
[
−
0
.
24
,
0
.
24
]
.
Fig
ure
4
sho
ws
t
he
error
sig
n
afte
r
fr
e
quency
offset
co
rr
ect
io
n.
The
pe
rfor
m
ance
of
t
he
rate
i
s
bette
r
tha
n
t
hat
of
the
unc
orrecte
d
f
requ
ency
offset,
a
nd
t
he
perform
a
nce
of
the pr
opos
e
d
al
gorithm
is sli
gh
tl
y bett
er th
a
n t
he
e
xisti
ng algorit
hm
.
Figure
4
.
C
omparis
on of sym
bo
l e
rro
r
rates
The
c
om
plexity
analy
sis
is
carried
out,
a
nd
t
he
c
om
ple
x
it
y
of
this
al
gorithm
is
com
par
ed
with
the
existi
ng
a
lgorit
hm
(i.e,
[
4]
).
T
he
frequ
e
ncy
offse
t
correct
io
n
va
lue
of
eac
h
existi
ng
ante
nna
on
the r
ecei
ving a
nten
na
is
[4]
:
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
10
, No
.
1
,
Febr
uar
y
2020
:
90
0
-
907
906
{
̃
,
=
1
2
∑
2
=
1
∑
2
=
1
=
1
,
…
.
,
(
)
≤
̃
,
≤
(
)
(20)
This
pa
pe
r
va
li
date
the
com
plexity
of
the
pro
po
se
d
al
gor
it
h
m
based
on
[25]
.
Accord
i
ng
t
o
(
17)
a
nd
(20),
Table
1
li
sts t
he
co
m
pu
ta
ti
on
a
l com
plexity
o
f
the
pro
po
se
d
a
lgorit
hm
an
d
t
he
e
xisti
ng
al
gorithm
.
Table
1
Com
pa
rison
of the c
om
plexity
w
it
h
the e
xisti
ng algorit
hm
[
3]
Ad
d
itio
n
Multip
licatio
n
Div
isio
n
Sin
e
Co
sin
e
Co
n
stan
t
Total n
u
m
b
er
o
f
op
eration
s
Exis
tin
g
algo
rith
m
[
3
]
2
(
−
1
)
2
M
T
2
M
T
M
T
1
6
M
T
+1
Prop
o
sed
algo
rith
m
2
(
−
1
)
M
T
1
0
0
0
3
M
T
-
1
It
can
be
see
n
from
Table
1
that
the
total
nu
m
ber
of
operat
ion
s
of
the
al
gorithm
is
3M
T
-
1
tim
es,
and
the
existi
ng
al
gorithm
is
6M
T
+1
ti
m
es.
The
al
go
rithm
of
this
pa
per
is
lowe
r
t
han
th
e
existi
ng
al
go
rithm
.
At
the
sam
e
t
i
m
e,
fo
r
a
ny
op
erati
on
require
d
to
c
om
plete
the
al
gorithm
,
the
al
gorithm
needs
the
num
ber
of
tim
es
is
le
ss
t
han
or
eq
ual
t
o
the
e
xisti
ng
al
gorithm
.
In
add
it
io
n,
t
he
e
xisti
ng
al
gorithm
req
uires
M
T
sine
op
e
rati
on,
M
T
cosine
op
e
rat
ion
a
nd
1
arct
ang
e
nt
oper
at
ion.
De
fine
d
(
)
=
(
3
−
1
)
/
(
6
+
1
)
as
the
com
plexity
rati
o
functi
on
of
the
pro
pose
d
al
gorith
m
and
the
e
xi
sti
ng
al
gorit
hm
.
Be
cause
of
the
first
der
i
vative
of
(
)
=
9
/
(
6
+
1
)
2
>
0
,
C(M
T
)
is
i
ncrea
sing
with
re
sp
ect
to
the
nu
m
ber
of
t
ran
s
m
itti
ng
anten
nas
M
T
te
nd
s
to
i
nfi
nity
.
In
te
rm
s
of
t
he
nu
m
ber
of
op
e
r
at
ion
s
li
m
→
∞
(
)
=
0
.
5
,
so
the
pro
posed
m
et
ho
d
com
plexity
is l
ess tha
n or eq
ua
l t
o
50%
of th
e existi
ng alg
ori
thm
.
5.
CONCL
US
I
O
N
In
this
pa
per
,
a
m
ulti
-
fr
eq
ue
ncy
offset
co
rrec
ti
on
al
gorith
m
fo
r
dist
rib
uted
tran
sm
i
t
antenn
a
M
IM
O
OFDM
is
pro
pose
d.
T
he
c
onditi
on
al
a
ver
a
ge
SI
R
is
m
axim
iz
ed
for
eac
h
subcar
rier
on
the
recei
ve
a
ntenn
a
as
the
crit
erio
n
to
co
rr
ect
the
fr
e
qu
e
ncy
off
s
et
,
by
the
f
unct
ion
2
(
∗
)
in
the
conditi
on
al
a
ve
rag
e
S
I
NR
expressi
on.
T
he
poly
nom
ial
ap
pr
ox
im
at
e
m
e
thod
is
obta
ine
d
with
l
ow
c
om
plexit
y
fr
e
quency
offset
c
orrecti
on.
Com
par
ed wit
h
the
existi
ng freq
ue
ncy offse
t correct
ion al
gorithm
, th
e p
e
r
form
ance o
f
t
he
p
r
opose
d
al
gorithm
is sl
igh
tl
y i
m
pr
ov
e
d, an
d
t
he
c
om
plexity
is red
uce
d by
50%
or m
or
e.
ACKN
OWLE
DGE
MENTS
The
a
uthor
s are
grate
fu
l t
o
Ti
kr
it
Un
i
ver
sit
y
for pr
ovidin
g f
inancial
s
uppor
t t
o
com
plete
this
proj
ect
.
REFERE
NCE
S
[1]
N.
H.
Dawod
,
I
.
D.
Marsl
and,
a
nd
R.
H
.
Haf
ez,
"Im
prove
d
tra
n
sm
it
null
st
ee
rin
g
for
MIM
O
-
OF
DM
downlinks
with
distr
ibut
ed
base
station
a
nte
nna
arr
a
y
s
,
"
IEE
E
Journal
o
n
Selec
te
d
Area
s
in
Comm
unications
,
vo
l.
24
,
pp.
419
-
426
,
20
06.
[2]
Y.
Shen
,
Y
.
Ta
n
g,
T.
Kong,
and
S.
Shao
.
,
"O
ptim
al
an
te
nn
a
locati
on
for
STBC
-
OF
DM
downlink
with
distr
ibut
e
d
tra
nsm
it
an
te
nn
a
s in
l
i
near cel
ls,"
IEEE
Comm
uni
cat
ions Le
tt
ers
,
vol.
11
,
pp
.
387
-
389,
2007
.
[3]
P.
A.
C.
Lop
es
and
J.
A.
B.
Ger
al
d,
"L
ea
kag
e
-
b
ase
d
pr
ec
oding
al
gorit
hm
s
for
m
ult
ipl
e
stre
ams
per
t
erminal
MU
-
MIM
O sy
st
ems
,
"
Digit
al
Signal
Proce
ss
ing
,
vo
l.
75,
pp
.
38
-
44
,
2
018.
[4]
K.
Deng,
Y
.
T
an
g,
S.
Shao
,
and
K
.
Sun
.
,
"Correcti
on
of
c
arr
ier
fre
q
uency
offse
ts
in
OF
DM
-
base
d
sp
at
i
al
m
ult
ip
le
x
in
g
MIM
O
with
dist
ribut
ed
tr
ansm
it
ant
enn
as,
"
I
EE
E
Tr
ansacti
ons
o
n
Ve
h
ic
u
lar
Tec
hnology
,
vol
.
58
,
pp
.
2072
-
207
7
,
2009.
[5]
B.
-
b.
Hu
,
Y.
-
a.
L
iu,
G.
Xie
,
J.
-
c.
Gao,
and
Y.
-
l
.
Y
ang
.
,
"Ene
rg
y
eff
ic
i
ency
of
m
assive
MIM
O
wire
les
s
comm
unic
at
io
n
s
y
stems
with
an
te
nna
se
lecti
on
,
"
The
Journal
of
China
Unive
rs
i
ti
es
of
Posts
and
Tele
communic
ati
ons
,
vol
.
2
1
,
pp.
1
-
8,
2014
.
[6]
E.
M.
Okum
u
a
nd
M.
E.
Dlodlo
,
"Tr
ansm
it
an
tenna
sel
ection
fo
r
m
ult
iple
anten
na
s
y
s
te
m
s
with
stall
avoi
d
anc
e
,
"
Computers
&
El
ec
tri
cal
Engi
n
eer
ing
,
vol
.
58
,
pp
.
144
-
153,
2017.
[7]
A.
Dat
ta
and
V.
Bhatia
,
"A
n
ear
m
axi
m
um
li
ke
li
hood
p
erf
orm
a
nce
m
odifi
ed
fir
efly
al
gor
it
hm
f
or
large
MIM
O
det
e
ct
ion
,
"
Swar
m and
Ev
o
lut
ion
ary
Computati
on
,
vol
.
44
,
pp
.
828
-
839,
2019
.
[8]
O.
Font
-
Bac
h
,
N.
Bartzoudis,
A.
Pascua
l
-
Iser
t
e,
and
D.
L.
Bu
eno
.
,
"A
real
-
tim
e
MIM
O
-
O
FDM
m
obil
e
W
iMAX
rec
e
ive
r:
Archi
tectur
e
,
d
esign and
FP
GA
implementa
ti
on
,
"
Compu
te
r Ne
tworks
,
vo
l.
55
,
pp
.
3634
-
3
647,
2011
.
[9]
A.
Tri
m
ec
h
e,
A
.
Sakl
y
,
and
A.
Mtiba
a
.
,
"F
PG
A
Im
ple
m
ent
at
ion
of
ML,
ZF
and
MM
SE
Equa
li
z
ers
for
MIM
O
S
y
stems
,
"
Proc
e
dia
Computer
S
c
ie
nc
e
,
vol. 73, p
p.
226
-
233
,
201
5.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
Mi
nimize M
IM
O OFD
M inter
fe
rence
and n
oi
se ra
ti
o usi
ng
po
ly
nomial
-
ti
m
e a
lg
or
it
hm
(
M
uhamed
K
Hu
s
ei
n
)
907
[10]
A.
Kum
ar
and
P.
R
.
Sahu
,
"
Perform
anc
e
an
aly
s
is
of
DCS
K
-
SR
sy
st
ems
base
d
on
best
rel
a
y
s
elec
t
ion
in
m
ult
ipl
e
MIM
O
rel
a
y
envi
ronm
ent
,
"
AE
U
-
Int
ernati
onal
Journal
of
E
le
c
troni
cs
and
Comm
unic
ati
ons
,
vol
.
70
,
pp.
18
-
24
,
2016
.
[11]
W
.
Zha
ng
,
D.
Q
u,
and
G
.
Zhu
.
,
"P
erf
orm
anc
e
inv
esti
gation
of
d
ist
ribut
ed
STBC
-
O
FD
M
sy
stem
wit
h
m
ult
ipl
e
carrier
fre
quency
off
se
ts,"
in
2006
I
EE
E
17th
Inter
nati
onal
S
ymposium
on
Pe
rs
onal,
Indoor
a
nd
Mobile
Rad
io
Comm
unic
ati
ons
,
pp
.
1
-
5
,
2006
.
[12]
Li
ndner
,
J.,
et
al.,
Com
par
ison
o
f
vec
tor
d
et
e
ctio
n
al
gorit
hm
s
for
MIM
O
-
O
FD
M.
AEU
-
Inte
rna
t
iona
l
Journa
l
of
El
e
ct
roni
cs
and
Com
m
unic
at
ions,
2005.
59(3):
p
.
137
-
146
.
[13]
Upadh
y
a, K.
,
C.
S.
Seelam
ant
ul
a, a
nd
K.V.S.
Hari
,
A r
isk minimiz
at
ion
fra
m
ework
for chann
el
estim
at
ion
in
OF
DM
s
y
stems
.
Signa
l P
roc
essing,
201
6.
128:
p.
78
-
87.
[14]
Golovins,
E
.
and
N.
Ven
tura,
Opt
imis
at
ion
of
th
e
pil
ot
-
t
o
-
d
ata
po
wer
ra
ti
o
in
the
wire
le
ss
MIM
O
-
OF
DM
sy
stem
w
it
h
low
-
complexi
t
y
MM
SE
cha
nnel
esti
m
at
ion
.
Com
pute
r
Com
m
unicati
ons,
2009.
32(
3):
p.
465
-
476.
[15]
Chinna
dura
i,
S.
,
et
a
l.,
W
orst
-
cas
e
weight
ed
sum
-
rat
e
m
axi
m
izat
i
on
in
m
ult
icell
m
assive
MIM
O
downlink
s
y
ste
m
for
5G c
om
m
unic
ations.
Ph
y
s
ical
Com
m
unic
at
ion
,
2018
.
27:
p.
11
6
-
124.
[16]
C.
Pap
at
han
asio
u,
N.
Dim
it
r
iou,
and
L.
T
assiula
s,
"D
y
n
amic
rad
io
resourc
e
and
in
terfere
nc
e
m
ana
ge
m
ent
for
MIM
O
–
OF
DMA
m
obil
e
broa
dband
wire
l
ess a
ccess
s
y
st
e
m
s,"
Computer
Net
works
,
vo
l. 5
7,
pp
.
3
-
16
,
201
3.
[17]
H.
T
ao,
M.
Z.
A.
Bhui
y
a
n,
A.
N.
Abdall
a
,
M.
M.
Hass
an,
J.
M.
Zain,
and
T
.
Ha
y
a
j
neh
.
,
"S
e
cur
ed
d
at
a
co
llection
w
i
th
har
dware
-
b
ase
d
ci
pher
s for
io
t
-
b
ase
d
he
al
th
ca
r
e,
"
IEEE
In
te
rnet
o
f
Things
Journal
,
vol
.
6
,
pp
.
410
-
420,
2019
.
[18]
Y.
W
ang
and
X.
-
f.
Ta
o
,
"Int
er
-
an
t
enna
and
subblo
ck
shift
ing
and
i
nver
sion
for
pe
a
k
-
to
-
ave
r
age
po
wer
ra
ti
o
red
u
ct
i
on
in
MIM
O
-
OF
D
M
s
y
stems
,
"
The
Journal
o
f
Ch
in
a
Unive
rs
ities
o
f
Posts
and
Tel
ec
o
mm
unic
ati
ons
,
v
ol.
14
,
pp
.
41
-
45
,
2007.
[19]
S.
Singh
and
A.
Kum
ar,
"
Perform
anc
e
Ana
l
y
sis
of
Adapti
v
e
Cl
ip
ping
T
ec
hniqu
e
f
or
Reduction
o
f
PA
PR
in
Alamouti
Coded
MIM
O
-
OF
DM
Sy
stems
,
"
Proce
dia
Computer
Sc
ie
nc
e
,
vol
.
93,
pp
.
609
-
616
,
2016.
[20]
A.
S.
Nam
it
ha
and
S.
M.
Sam
ee
r,
"A
b
andwidth
eff
i
ci
en
t
sel
ective
m
app
ing
t
e
chni
que
for
th
e
PA
PR
red
uct
ion
in
spatial
m
ult
i
ple
xing
MIM
O
-
OF
DM
wire
le
ss
comm
unic
at
io
n
s
y
stem,"
Phys
ic
al
Comm
uni
cat
ion
,
vol
.
25
,
pp.
128
-
138
,
20
17
.
[21]
P. R
emlei
n
,
T
.
Ç
ogal
an
,
and
T.
Gucluogl
u
.
,
"O
FD
M
with tra
nsm
it
and re
ce
iv
e an
tenna s
el
e
ction
bas
ed on
subca
r
rier
groups
,
"
in
2012
8th
In
te
rnation
al
S
ymposium
o
n
Comm
unic
at
io
n
Syst
ems,
N
et
w
orks
&
Digit
al
S
ignal
Proce
ss
ing
(
CSNDSP
)
,
pp.
1
-
4,
2012
.
[22]
D.
N.
Dao
and
C
.
Tella
m
bura
,
"In
te
rc
arr
i
er
interfe
ren
ce
se
lf
-
c
ancel
la
ti
on
spac
e
-
fre
q
uency
cod
es
for
MIM
O
-
O
FD
M,"
IEE
E
Tr
ansacti
o
ns
on
Ve
h
ic
u
lar
Technol
ogy
,
vo
l. 54, pp. 1729
-
17
38,
2005
.
[23]
S.
S.
R
a
y
.
,
Nume
rical
ana
ly
sis wi
th
algorit
hms
an
d
progr
amm
ing
:
CRC Press,
201
8.
[24]
D.
Huang
and
K.
B.
Le
t
aief,
"A
n
i
nte
rfe
r
ence
-
c
ancel
l
at
ion
sche
m
e
f
or
ca
rr
ie
r
fre
qu
e
nc
y
offset
s
cor
re
ct
ion
in
OF
DM
A
s
y
st
ems
,
"
IE
EE
Tr
ansacti
ons on Com
municat
ions
,
vol
.
53
,
pp
.
115
5
-
1165,
2005
.
[25]
S.
Marinkovic,
B.
Vuce
tic,
and
A.
Us
hiroka
w
a,
"S
pa
ce
-
t
ime
it
er
at
iv
e
and
m
ult
ista
g
e
r
ecei
ve
r
struct
ur
es
for
CDM
A
m
obil
e
comm
unic
at
io
n
s
y
stems
,
"
IE
EE
Journal
on
select
ed
Area
s
in
Comm
u
nic
ati
ons
,
vo
l.
1
9,
pp.
1594
-
1604
,
2001.
BIOGR
AP
H
Y
O
F
AU
TH
OR
Mohame
d
Kh.
Hus
ei
n
,
He
g
ot
his
B.
Sc
Elec
tron
ic
s
Eng
in
ee
ring
,
Air
-
Mil
i
ta
r
y
Eng
ineeri
n
g
Aca
dem
y
/
Facult
y
of
E
le
c
tronics.
/Sar
aj
evo
Yug
oslavi
a
,
1982
.
Also
he
got
his
M.Sc.
Elec
tron
ic
s
Engi
ne
eri
ng,
Univer
sit
y
of
B
el
gr
ade
/
El
e
ct
r
ic
a
l
Engi
ne
eri
ng
Col
le
ge
,
Bel
g
rad
e
Yugos
la
via
.
198
4
.
Curre
ntly
,
worki
ng
with
col
l
ege
of
engi
n
ee
r
ing
S
hirqa
t
,
d
epa
rtme
nt
of
ele
ct
r
ical
e
ngine
er
ing,
Ti
kr
it
unive
rsit
y
.
His
m
ai
n
rese
arc
h
in
te
rest
including
,
wire
le
ss
comm
unic
a
ti
on,
OF
DM,
and
opt
imiza
t
io
n.
Evaluation Warning : The document was created with Spire.PDF for Python.