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.
3
,
June
2020
,
pp.
3275
~
3283
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
3
.
pp3275
-
32
83
3275
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/IJ
ECE
Network
effici
en
cy enh
ancement b
y rea
ctive chann
el state
based
allocatio
n schem
e
Umesh
a G
.
B
.
,
M
.
N
.
S
hanm
ukh
a
Sw
am
y
Depa
rtment
o
f
E
le
c
troni
cs
and
C
om
m
unic
at
ion
E
ngine
er
ing
,
Sri
Ja
y
ac
h
amara
je
ndra
Coll
ege o
f
Engi
n
ee
ring
(S
JCE)
,
Indi
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
pr
10
, 201
9
Re
vised
Dec
11
,
20
19
Accepte
d
J
an
8
, 2020
Now
a
day
th
e
la
rge
MIM
O
h
as
conside
red
a
s
the
eff
ic
i
ent
a
pproa
ch
to
improve
the
spec
tral
and
ene
rg
y
eff
ic
i
ency
a
t
W
MN
.
Ho
weve
r,
the
PC
is
a
big
issue
th
at
ca
used
b
y
re
using
sim
il
ar
pil
o
t
seque
nc
e
a
t
cells,
which
al
s
o
restr
ict
the
p
er
form
anc
e
of
m
assive
M
IMO
net
work.
Her
e,
we
giv
e
the
a
lt
e
rna
t
ive
answer,
wher
e
ea
ch
of
UEs
re
quire
d
a
ll
ot
ti
ng
a
ch
annel
seque
nce
s
b
efo
r
e
passing
the
pa
y
loa
d
data,
so
as
to
avo
id
the
cha
nn
el
col
li
sion
of
inter
-
cell
.
Our
pr
oposed
protoc
ol
will
rea
d
y
to
det
ermine
the
ch
anne
l
col
l
isions
in
distri
b
ute
d
and
sc
alabl
e
proc
ess,
howe
ver
givi
ng
unique
prope
r
ties
of
the
la
rg
e
MIM
O
cha
nnel
s.
Here
we
ha
ve
proposed
a
RCS
A
(Rea
ct
i
ve
cha
nne
l
state
base
d
al
lo
ca
t
io
n)
sche
m
e,
whic
h
will
ver
y
produc
ti
v
ely
w
ork
with
the
R
AP
bloc
ker
s
at
la
rge
n
et
wo
rk
of
MIM
O.
The
positi
on
of
ti
m
e
-
fre
quen
c
y
of
RAP
bloc
ks
is
m
odifi
ed
in
the
m
iddl
e
of
the
adjac
en
t
ce
l
ls,
bec
ause
of
t
his
design
dec
ision
the
RAP
def
end
from
the
har
d
est
t
ypes
of
interfe
r
enc
e
at
inter
-
ce
l
l.
Furth
er,
to
valida
t
e
the
per
form
ance
of
our
propose
d
sche
m
e
it
will
be
compare
d
with
othe
r
exi
sting
techniq
ue.
Ke
yw
or
d
s
:
Ba
se
sta
ti
on
s
(
BSs)
M
ulti
ple
-
in m
ulti
ple
-
ou
t
(MIMO
)
Pil
ot
co
ntam
in
at
ion
(P
C)
Ra
ndom
acce
ss pro
t
oco
l
(RA
P)
User
eq
uip
m
ent’s
(
UEs
)
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
:
Um
esha G
.
B
.
,
Dep
a
rtm
ent o
f
Ele
ct
ro
nics
and C
omm
un
ic
ation
En
gin
ee
rin
g
,
Sr
i Jay
acham
araj
en
dra Coll
e
ge
of E
ng
i
neer
i
ng
(
S
JCE
)
,
My
so
re, Ka
rn
a
ta
ka,
India
.
Em
a
il
:
u
m
esh2719
@r
e
diff
m
ai
l.com
1.
INTROD
U
CTION
Ma
ssive
MIM
O
is
a
te
ch
no
l
og
y
wh
e
re
cel
lular
BSs
are
hav
i
ng
la
rg
e
num
ber
of
a
nte
nn
a
s
(
NoA).
It
sho
ws
si
gn
i
f
ic
antly
a
m
ou
nt
of
im
pr
ovem
e
nt
in
reli
abili
ty
an
d
ca
pacit
y
of
the
wi
reless
s
yst
e
m
,
al
so
it
al
lows
i
m
pr
ove
orders
of
m
agn
it
ude
in
the
spe
ct
ral.
The
e
nergy
ef
fici
ency
is
al
s
o
can
be
pr
ov
i
de
d
t
hroug
h
relat
ively
si
m
ple
pr
oce
ss
ing
in
m
assive
MIM
O
(m
MIM
O)
,
t
her
e
fore
it
has
been
broad
ly
stu
died
f
ro
m
la
st
two
decad
es,
al
so
co
ns
i
der
e
d
at
m
any
wir
el
ess
sta
nd
a
r
ds.
As
pe
r
the
r
andom
m
at
rix
theo
ry
[
1],
th
e
cha
nn
el
vect
or
s
in
betwee
n
the
B
S
an
d
UE
bec
om
e
the
pair
wis
e
ort
hogo
nal. Th
ere
f
or
e,
it
re
su
lt
s
the
m
inim
iz
at
ion
of
sm
al
l
-
scal
e
fad
i
ng
an
d
un
correla
te
d
noise
ef
fect
as
pe
r
the
increm
ent
of
ante
nnas
at
BS
in
t
he
MIM
O
net
wor
k.
Subseque
ntly
,
the
esse
nt
ia
l
tr
ansm
itted
ene
r
gy
as
pe
r
bit
al
so
ca
n
be
as
ym
pto
ti
cal
ly
l
ess
with
out
of
m
uch
perform
ance lo
ss [2].
Ther
e
are
se
ve
ral
co
ntributi
on
s
w
hich
pro
vid
e
i
ns
ig
hts
on
t
he
e
nergy
eff
ic
ie
nt
desi
gn
of
syst
em
;
in
[
3],
pro
vid
e
d
the
optim
iz
a
ti
on
at
do
wn
li
nk
side
of
m
assive
MIM
O
syst
e
m
us
ing
ze
r
o
-
forcin
g
pr
e
-
c
od
i
ng
schem
e.
Iterati
ve
prot
otype
was
co
ns
id
ere
d
unde
r
a
tru
e
power
c
onsum
pt
ion
syst
em
in
order
to
i
ncr
ease
the
ene
rg
y
e
ff
i
ci
ency.
Th
ough
the
a
ppr
oac
h
aim
ed
to
get
t
he
en
ha
nce
d
e
nergy
ef
fici
enc
y
at
un
if
or
m
rates
for
the
ever
y
us
e
r,
w
it
h
co
ns
ide
ring
the
s
uitable
channel
sta
te
i
nfor
m
at
ion
.
More
over,
a
res
ource
al
locat
io
n
based
appr
oach
has
pro
vid
e
d
in
[
4]
to
obta
in
the
optim
al
ener
gy
eff
ic
ie
ncy
with
the
BS
c
oord
i
nation,
the
outc
om
e
ref
le
ct
ed
that
the
energy
eff
i
ci
e
ncy
op
ti
m
izati
on
dif
ficult
y
was
no
t
co
nvex
but
it
was
a
ble
to
transform
into
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
.
3
,
J
une
2020
:
32
7
5
-
32
8
3
3276
the
co
nvex
fr
a
ct
ion
al
pro
gra
m
m
ing
.
H
ow
e
ver,
the
a
uthor
in
[
5],
ai
m
ed
towa
rd
s
t
he
th
eor
et
ic
al
in
ves
ti
gation
of
e
ne
rg
y
e
ff
ic
ie
ncy
an
d
s
pec
tral
eff
ic
ie
ncy
for
the
m
ulti
-
use
r
ty
pe
m
as
si
ve
MIM
O
syst
e
m
.
It
showe
d
so
m
e
visio
ns
on
the
way
for
ante
nnas
al
locat
ion,
al
so
der
i
ved
ne
arly
expressio
ns
f
or
the
nu
m
ber
of
use
rs
th
at
are
necessa
ry
to
be
serv
e
d
in
a
ne
twork
.
Howe
ver,
it
did
no
t
able
to
pro
vid
e
the
com
plete
structu
re
an
d
th
eref
or
e
the
co
ns
i
der
e
d m
od
el
w
as
no
t
m
uch
p
racti
cal
to use
furthe
r.
Thro
ugh
the
c
om
par
ison
with
tra
diti
on
al
s
yst
e
m
of
MI
MO
[
6]
as
s
how
n
i
n
F
ig
ur
e
1
,
the
la
r
ge
MIM
O
has
m
any
ben
e
fits
su
ch
as
the
pro
cessi
ng
of
m
od
est
li
near
at
BS
and
al
so
the
anten
nas
usa
ge
at
m
ob
il
e
t
er
m
inal.
Wh
il
e
co
ns
ide
rin
g
t
he
Tim
e
DD
(
div
isi
on
dupl
ex)
at
la
r
ge
netw
ork
of
MIM
O,
the
sym
m
et
ric
al
pilots
hav
e
so
m
e
resource
s
an
d
nee
d
to
util
iz
e
in
ad
j
acent
cel
l
ular
s
pace,
w
her
e
ve
r
the
cha
nn
el
uti
li
ze
of
fers
gro
w
f
or
the
c
hannel
co
ntam
inati
on
.
T
hat
cau
s
es
the
ne
gative
i
m
pact,
fo
r
e
xam
ple,
at
ta
inable
rate
execu
ti
on
a
nd
exactness
est
i
m
at
ion
at
the
up
li
nk
c
hannel
[1
]
.
The
a
uthor
in
[7
]
ha
ve
thought
about
the
CR
-
MIM
O
(c
ogniti
ve
rad
i
o)
,
wh
e
rev
e
r
the
inter
f
eren
ce
is
occ
urred
becau
se
of
the
op
ti
m
al
us
ers
of
da
ta
tra
ns
m
issio
n
an
d
t
her
e
f
ore
the
esse
ntial
us
er
s
hav
e
a
ff
ect
ed
th
rou
gh
the
c
hannel
transm
issi
on
.
I
n
[8
]
,
the
essenti
al
us
ers
prom
ise
s
the
QoS
in
la
rg
e
syst
e
m
of
MIM
O
thr
ough
le
arn
i
ng
the
i
m
pact
of
vast
-
scal
e
essenti
al
BS
on
fe
w
optim
al
networ
k
of
MIM
O
an
d
th
rou
gh
set
ti
ng
the
best
interf
eren
ce
le
vel
i
n
la
rge
netw
ork
of
MIM
O.
I
n
[
9],
a
pilot
sc
hem
e
for
la
r
ge
M
IMO
of
co
gnit
ive
netw
ork
has
been
pr
opos
e
d
t
o
i
m
pr
ove
the
qual
it
y
of
c
ha
nnel
eval
uatio
n
of
opti
m
a
l
us
e
rs
with
reduci
ng
the
ne
gativ
e
reacti
on
on
e
ssent
ia
l
netw
ork
of
c
ha
nn
el
est
i
m
at
ion
.
I
n
[10],
pro
pose
a
reciproci
ty
wh
ic
h
is
bas
ed
on
la
rg
e
MI
MO
of
beam
-
f
or
ci
ng
m
et
ho
d,
s
o
as
the
UEs
(
us
e
r
eq
uip
m
ent)
durin
g
a
netw
ork.
Wh
e
reas
a
fu
ll
sp
at
ia
l
ran
ge
s
har
i
ng
at
la
rg
e
MIM
O
of
net
work
ha
s
bee
n
pro
posed
[
11]
wit
h
the
m
i
nim
iz
ed
over
he
ad
of
pr
e
pa
ring
to
reduce
t
he
PC
.
Likewise
it
ut
il
iz
es
2
-
D
D
FT
(
Discrete
Four
ie
r
T
ra
ns
f
or
m
)
for
angular
direct
ion
al
an
d
s
pread
of
arr
ival e
stim
at
i
on.
Figure
1.
MIM
O netw
ork dia
gr
am
Usu
al
ly
the
ce
ll
ular
netw
ork
assigns
the
de
di
cat
ed
re
sou
rces
to
i
nd
i
vidual
act
ive
of
UE,
i
n
this
m
ann
er
it
is
vi
ta
l
fo
r
the
B
S
to
tran
sm
i
t
tim
e
-
fr
eq
uen
cy
of
resou
rces
area.
T
he
la
r
ge
netw
ork
of
MIM
O
al
locat
es
al
l
asset
s
of
ti
m
e
-
fr
equ
e
ncy
to
al
l
UEs
a
nd
c
onfi
nes
t
hem
sp
at
ia
ll
y
in
unde
rst
and
i
ng
to
t
hei
r
channel
seq
uen
ces
.
T
he
re
are
restrict
ed
num
ber
s
of
c
hannel
beca
us
e
of
rati
on
al
bloc
k
siz
e,
wh
il
e
the
co
nce
pt
of
la
rge
MIM
O
[
12
]
, the
i
m
po
rtant ch
ann
el
use agai
n
over th
e cel
lular area a
nd
it
causes
the ch
a
nn
el
contam
inati
on
of
inter
cel
l,
wh
i
ch
le
ads
to
w
a
rd
the
e
xtra
i
nterf
e
re
nce
[
13]
.
Thi
nk
i
ng
a
bout
the
cr
ow
ded
urba
n
sit
ua
ti
on
s,
the
num
ber
of
act
ive
and
i
nac
ti
ve
UEs
pr
ese
nt
in
the
CN
(c
el
lular
netw
ork
)
w
hich
is
sig
ni
ficantl
y
m
or
e
than
the
al
l
acce
ssible
cha
nnel
carrier
gro
up
i
ngs
.
I
n
this
way
the
cha
nne
l
can’
t
be
dete
rm
i
ned
f
or
al
l
tim
e
with
the
sp
eci
fic
U
E
in
cel
l,
beca
us
e
of
t
his
pr
es
ent
it
is
nee
ded
to
be
de
vi
ou
sl
y
al
locat
ed
an
d
de
-
al
l
ocated
s
o
as
t
o
purs
ue
it
s
unpredict
able
rec
urre
nt
in
form
at
i
on
tra
ff
ic
desi
gn.
T
he
R
AP
(r
a
ndom
acce
ss
prot
oco
l
)
ca
n
b
e
the
be
st
decisi
on
f
or
this
s
or
t
of
ch
an
nel
re
s
ources
al
locat
i
on
;
howe
ve
r
th
e
desi
gn
is
ad
di
ti
on
al
ly
exp
ec
te
d
to
beat
the
colli
s
ion
in
de
ns
e
c
el
ls
i
m
m
ediat
e
ly
.
In
[
9],
pr
opose
d
a
te
ch
ni
qu
e
w
her
e
the
UEs
ca
n
rea
dy
to
transm
it
the
inform
ation
unde
r
a
c
ho
se
n
ra
ndom
channel
from
the
nor
m
al
po
ol,
t
his
reduces
t
he
e
ntra
nce
delay
s at the c
os
t
of
c
ha
nn
el
colli
sion
t
hat c
auses
t
he
int
ra
-
cel
l PC.
The
colli
sio
n
is
introd
uced
be
cause
the
gr
a
ph
c
ode
in
[
14]
and
al
so
the
belie
f
pro
pa
gation
is
ta
ke
n
into
acco
unt
to
m
od
e
rate
the
PC
thr
ough
ap
plyi
ng
the
pr
i
nc
iple
of
R
AC
(
rand
om
acce
ss
cod
e
).
T
he
m
a
ny
SE
(sp
ect
ral
eff
ic
i
ency)
ex
pressi
on
s
a
re
c
on
si
der
e
d
in
[15]
and
util
iz
ed
to
en
han
ce
t
he
prob
a
bili
ty
of
UE
act
ivati
on
an
d
al
so
the
channel
le
ng
th
.
Al
so
,
the
asy
nc
hro
nous
of
tim
i
ng
U
E
is
sp
eci
fied
in
[9
]
so
as
to
determ
ine
and
detect
the
c
ha
nnel
c
olli
sion
i
n
RAP
.
Durin
g
this
pa
pe
r,
we
giv
e
t
he
al
te
r
na
ti
ve
ans
we
r,
wh
e
re
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
Ne
tw
or
k Eff
ic
ie
ncy e
nhancem
ent b
y
re
activ
e chan
nel state
base
d all
oca
ti
on
sch
e
me
(
U
mes
ha G. B
.
)
3277
ever
y
of
U
Es
re
qu
ire
d
al
lo
tt
ing
a
c
ha
nnel
seq
uen
ce
s
befor
e
pa
ssin
g
the
payl
oa
d
data,
s
o
a
s
to
av
oid
the
channel
col
li
sion
of
int
ra
-
cel
l.
These
represent
the
th
ough
t
ab
out
scen
ario
of
payl
oa
d
transm
issi
on
within
the
la
rg
e
MIM
O
netw
ork
[12
-
13]
.
He
re,
we
hav
e g
ot
ta
r
geted
on
the
urba
n
orga
nizat
ions
with
the
help
of
f
e
w
init
ia
l
var
ia
ti
on
of
the
ti
m
ing
and
pro
po
se
a
RAP
f
or
UEs
t
hat
nee
d
to
get
the
cel
lular
ne
twork
.
O
ur
pro
po
s
ed
protoc
ol
will
r
eady
to
deter
m
ine
the
c
hannel
colli
sio
ns
i
n
distrib
uted
a
nd
scal
able
pr
ocess,
ho
wev
e
r
giv
in
g
un
i
qu
e
pro
pe
rtie
s
of
t
he
la
r
ge
MIM
O
c
ha
nnel
s.
Her
e
we
hav
e
co
ntri
bu
t
ed
a
RC
S
A
(R
eact
ive
cha
nn
e
l
sta
te
base
d
al
locat
io
n)
sc
hem
e,
whic
h
will
ver
y
pro
duct
ively
work
the
R
AP
blo
cke
rs
at
la
rg
e
netw
ork
of
MIM
O
.
The
posit
ion
of
tim
e
-
fr
eq
ue
nc
y
of
R
AP
bl
oc
ks
is
m
od
ifie
d
in
the
m
idd
le
of
t
he
a
djacent
cel
ls,
beca
us
e o
f
t
his
desig
n decisi
on the
RA
P
def
e
nd fro
m
the h
a
rd
est
ty
pe
s
of
i
nterf
e
re
nce at i
nter
-
cel
l.
2.
LIT
ERATUR
E SU
RV
E
Y
A
m
MIM
O
is
on
e
of
t
he
m
os
t
well
-
kn
own
te
c
hnologi
es
for
the
wir
el
ess
com
m
un
ic
at
ion
an
d
at
tract
ed
the
huge
nu
m
ber
of
r
esearc
hers
du
e
to
it
s
va
st
sp
at
ia
l
div
ersit
y
and
hi
gh
data
tra
ns
m
issi
on
rate
[16,
17
]
.
Howe
ver,
in
m
MIM
O
syst
em
,
the
PC
is
a
big
issue
t
ha
t
caused
by
reu
si
ng
sim
il
a
r
pilot
seq
uen
ce
at
cel
ls,
wh
ic
h
al
so
r
est
rict
the
perf
or
m
ance
of
net
work
a
nd
can
not
be
re
duced
t
hro
ugh
m
axi
m
iz
ing
the
NoA
[
1].
P
il
ot
or
cha
nn
el
al
locat
ion
is
t
he
eff
ect
iv
e
pro
cess
to
m
ini
m
iz
e
the
PC
[1
8
-
20
]
.
I
n
pa
pe
r
[21],
pro
po
se
d
a
pilot
seq
uen
ce
ba
sed
al
locat
ion
appr
oach
in
order
to
m
itigate
the
con
se
quen
ce
of
PC
in
m
assive
MIM
O
syst
em
;
firstly
de
riv
ed
t
he
s
um
rate
of
up
li
nk
syst
e
m
wh
e
n
the
BS
ante
nnas
te
nd
to
infi
nity
.
In
acc
orda
nce
to
the
der
i
ve
d
ex
pr
e
ssio
n,
the
co
ntam
inati
on
of
pilot
is
on
ly
i
m
pairm
ent
with
res
pe
ct
to
the
syst
em
per
form
ance.
A
fterw
a
rds,
pro
vid
ed
an
op
ti
m
i
zed
m
od
el
of
pilot
al
locat
io
n
in
order
to
i
ncr
eas
e
the
syst
em
su
m
-
rate,
wh
ere
the
outc
om
e
is
ob
ta
i
ned
t
hro
ugh
al
l
the
c
oncei
va
ble
al
loc
at
ion
s
require
d
to
be
exam
ined
that
le
ads
to
wards
the
higher
co
m
pu
ta
ti
onal
c
omplexit
y.
In
pa
per
[
22]
,
pro
po
se
d
t
he
a
synch
ron
ous
s
cheduli
ng
a
ppr
oach
that
base
d
upon
t
he
f
ra
ct
ion
al
re
use
o
f
pilot
re
us
e,
wh
ic
h
al
lows
the
us
e
rs
to
be
fr
ee
f
ro
m
PC
a
t
the
tim
e
of
up
li
nk
tra
ns
m
is
sion.
In
acc
ord
ing
to
the
interfe
ren
c
e
le
vel,
two
gro
ups
are
crea
te
d
from
the
us
ers
that
ca
n
be
denoted
a
s
center
us
e
rs,
who’
s
hav
i
ng
the
m
ino
r
PC;
al
so
the
edge
us
e
rs,
w
ho’s
hav
i
ng
th
e
si
m
ple
PC
an
d
ba
sed
upon
t
his
di
ff
e
ren
ce
,
a
cel
l
-
centric
pilot
set
can
be
reu
s
ed
for
entire
center
us
e
rs
in
cel
ls;
wh
ere
as,
a
set
of
cel
l
-
edg
e
(CE
)
pilot
is
consi
der
e
d
for
the
ed
ge
us
er
s
unde
r
the
a
djacent
cel
ls.
T
her
e
fore,
t
he
use
d
pilots
thr
ough
the
CE
use
rs
a
r
e
ind
e
pende
nt
to
each
oth
e
r,
so
the
edg
e
us
er
s
able
to
transm
i
t
pilots
at
any
per
i
od
of
tim
e;
nonetheless,
t
he
set
of
pil
ot
fo
r
ce
nter
us
e
rs
can
be
re
us
ed
f
or
entire
cel
ls,
wh
ere
the
cel
l
center
us
e
rs
tra
ns
m
it
their
pil
ots
ove
r
the
non
-
overla
pp
e
d
ti
m
es to ev
ade
PC.
A
fair
pilot
al
locat
ion
m
et
ho
d
ha
s
pro
pose
d
in
[23]
f
or
t
he
m
ulti
-
cel
l
m
MIM
O,
to
r
edu
ce
t
he
P
C
eff
ect
a
nd
incr
ease
the
s
um
rate
(S
R)
of
syst
e
m
with
assu
rin
g
the
se
r
vices
am
on
g
the
us
ers
.
Mi
n
Le
akage
al
gorithm
(ML
A)
that
i
s
a
he
ur
ist
ic
al
gorith
m
is
pr
opos
e
d
to
reso
l
ve
the
pro
blem
of
optim
iz
ation
with
lower
syst
e
m
co
m
plexity
,
wh
il
e
c
on
si
der
i
ng
t
he
MLA
can
on
ly
deliver
a
s
uboptim
al
per
form
ance.
Mo
reover
,
a
gr
ee
dy
pilot
al
locat
ion
base
d
Use
r
-
E
xc
hange
ap
proac
h
is
propose
d
to
f
ur
t
her
im
pr
ov
e
the
syst
e
m
S
R
a
nd
the
serv
ic
es
a
m
on
g
us
ers
[23].
In
[
24]
,
aut
hor
co
ns
ide
re
d
the
pr
oble
m
of
resou
rce
al
locat
ion
f
or
the
up
li
nk
m
MIM
O
syst
em
,
wh
ere
the
anten
nas
at
BS
are
pri
ced
a
nd
the
virtua
li
zat
ion
has
done
acro
ss
the
s
erv
ic
e
pro
vid
er
s;
th
ou
gh
the
m
ob
il
e
netw
ork
pro
vi
der
(M
NP)
hol
ds
the
com
plete
infr
ast
ru
ct
ur
e
of
ante
nn
a
s
de
ci
des
the
pri
ce,
al
so
the
Stac
kelbe
rg
gam
e
is
co
ns
ide
red
f
or
t
he
m
axi
m
iz
ati
on
of
net
pro
f
it
of
M
NP,
w
her
eas
the
m
ini
m
u
m
SR
necessit
ie
s
of
se
rv
ic
e
prov
i
der
s
a
re
f
ul
fill
e
d.
I
n
or
der
to
res
olv
e
the
bi
-
le
vel
prob
l
e
m
of
MNP,
init
ia
ll
y
pro
vid
e
d
the
cl
os
e
d
-
form
op
tim
al
resp
onses
of
se
rv
ic
e
pro
vi
der
s
with
respec
t
of
MNP
pri
ci
ng
strat
egies.
The
refor
e
,
the
M
NP
pro
blem
can
be
dec
reas
e
to
wards
t
he
sin
gle
le
vel
op
ti
m
iz
ation
proces
s.
Af
te
r
wa
rd
s
,
th
e
ap
prox
im
at
i
on
s
an
d
tra
nsf
or
m
at
ion
s
are
us
e
d
to
cast
th
e
MNPs
dif
ficult
y
with
the
i
ntege
r
const
raints
to
Signom
ia
l
geo
m
et
ric
pr
ogra
m
and
,
t
hen
a
pp
li
ed
an
it
era
ti
ve
ap
proach
base
d
up
on
suc
cessi
ve
conve
x
est
im
ation
i
n order
to resol
ve
Si
gnom
ia
l geo
m
et
ric
progr
am
.
Energy
e
ff
ic
ie
nt
based
re
sour
ce
al
locat
ion
a
ppr
oach
is
pro
po
s
ed
in
[
25
]
f
or
the
m
ulti
-
use
rs
m
MIM
O
syst
e
m
s
with
the
am
plify
-
and
-
forwar
d
rela
y,
wh
e
re
the
ded
ic
at
e
d
rela
y
gu
ides
th
e
pair
wise
inf
orm
at
ion
intercha
nge
at
sever
al
pa
rts
of
sin
gle
a
nten
na
UE.
T
he
e
ne
rg
y
e
ff
ic
ie
ncy
syst
e
m
has
the
or
et
ic
al
ly
exa
m
ined
thr
ough
co
ns
id
erin
g
ra
ndom
m
at
rix
theor
y
and
huge
syst
e
m
analy
sis,
this
resu
lt
s
the
optim
al
app
r
ox
i
m
at
ion
for
MIM
O
syst
e
m
with
reas
onable
NoA.
It
al
so
pro
vide
s
the
s
om
e
eff
ic
i
ent
ap
pro
ac
hes
with
diff
e
ren
t
ty
pe
of
channel
sta
te
inform
ation
in
or
de
r
to
increase
the
syst
e
m
ener
g
y
eff
ic
ie
ncy
through
sc
he
du
li
ng
the
ap
pro
pr
ia
te
relay
anten
na
s
an
d,
UE
pa
ir
s
with
thei
r
co
rr
es
pondin
g
tr
ansm
issi
on
po
wer.
W
it
h
res
pe
ct
to
the
co
nventio
nal
res
ource
al
locat
ion
a
ppro
ac
hes,
the
pro
po
se
d
prot
otype
a
vo
i
ds
the
cal
culat
ion
s
of
com
plica
te
d
m
at
rix
a
nd
the
i
nst
ant
cha
nnel
s
ta
te
inform
at
io
n
of
t
he
sm
al
l
-
scal
e
fad
i
ng
; s
o
it
pr
ov
i
des
e
f
fici
ent
com
pu
ta
ti
on
w
it
h
sm
a
ll
ov
er
head
of
cha
nn
el
sta
te
inform
at
ion
.
M
os
tl
y
o
f
t
he
w
orks
a
i
m
ed
on
m
axim
iz
ing
su
m
-
rate
of
U
Es
bu
t
they
ha
ve
not
consi
dered
the
delay
ti
m
e
that
can
decr
ease
the
al
l
syst
e
m
per
fo
rm
ance
f
or
m
assive MIM
O
syst
em
.
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
.
3
,
J
une
2020
:
32
7
5
-
32
8
3
3278
3.
PROP
OSE
D MET
HO
DOL
OGY
Her
e
,
w
e
ha
ve
co
ns
ide
red
th
e
m
ajo
r
f
our
s
te
ps
of
R
CS
A
ap
proac
h
as
s
how
n
i
n
F
ig
ure
2
,
w
here
,
the
base
sta
ti
on
(BS
)
tra
ns
m
i
t
the
co
ntr
ol
sign
al
a
nd
each
of
t
he
U
E
will
ta
ke
us
e
of
thi
s
sign
al
t
o
pr
e
dict
it
s
channel
gain
aver
a
ge
a
nd
s
ynch
ronizes
it
sel
f
thr
ough
t
he
BS.
A
fter
wards,
unde
r
a
syst
e
m
init
i
al
iz
at
ion
process
an
d
queryi
ng
ste
p,
an
i
nacti
ve
s
ubset
of
UE
s
is
co
ns
ide
re
d
in
cel
l
that
want
s
to
act
ive
.
E
ach
of
the
UE
sel
ect
s
sequ
e
nces
of
channel
at
ra
ndom
sta
te
fr
om
the
pr
ede
fi
ned
R
AP
c
hannel
pool,
s
o
t
he
BS
an
est
im
at
es
its
cha
nnel
that
has
t
o
be
tra
nsm
itted
over
.
T
he
UEs
sel
ect
ion
is
sim
il
ar
fo
r
the
R
AP
ch
ann
el
s
,
if
the
c
olli
sion
has
ta
ke
n
place
then
the
BS
acq
uires
a
s
up
e
r
po
sit
io
n
U
E
est
i
m
at
ion
c
hannels.
A
fter
wards
,
the BS ca
nnot
able to i
de
ntify t
he follo
we
d
c
olli
sion
s t
hat si
m
il
ar to
the sit
uation i
n LTE.
Figure
2.
A
pr
opos
e
d
RC
S
A
schem
e fo
r
m
a
ssive
MIM
O
Ther
e
f
or
e,
i
n
r
eact
ion
of
que
uing
the
BS
re
sp
onses
t
hro
ugh
sen
ding
the
dow
nlink
c
ha
nnel
s
that
ar
e
pre
-
co
de
d
th
rough
the
cha
nn
el
est
i
m
a
te
s,
w
hich
resu
lt
s
at
sp
at
ia
ll
y
aimed
sig
nals
to
wa
rd
t
he
UE
s
an
d
offe
r
the su
it
a
ble RAP
res
ource. I
n
ste
p
3,
c
hann
el
stat
e analy
s
i
s an
d
res
ource
con
te
ntion
m
echan
ism
is con
sidere
d,
wh
e
re
t
he
pro
bab
il
it
y
of
non
-
c
olli
ding
c
ha
nn
el
tran
sm
is
sion
is
c
onsid
erab
ly
inc
reas
ed
i
n
RC
S
A
a
ppr
oach.
That
al
lows
th
e
m
assive
MI
MO
syst
e
m
to
acknow
le
dg
e
UEs
unde
r
the
cro
w
de
d
scenari
os
,
al
so
pr
ov
i
de
the
ide
ntit
y
of
UE
a
nd
payl
oad
tran
sm
issio
n
re
qu
e
st,
w
hich
help
t
o
r
esem
ble
the
rad
io
res
ource
con
t
ro
l
connecti
on
fo
l
lowe
d
by
the
dem
and
in
LT
E.
Wh
e
reas
c
onside
rin
g
t
he
s
te
p
4,
it
al
lo
w
s
these
res
ourc
es
by
ass
ign
i
ng
a
pilot
seq
uen
ce
th
at
fu
rth
er
can
be
us
e
d
at
the
payl
oad
blo
c
ks
or
to
sta
rts
c
onflic
t
reso
luti
on
at
the
fe
w
cases
wh
e
ne
ver
R
A
P
colli
sion
s
oc
curred
.
T
her
e
f
or
e
,
the
propos
ed
RC
SA
a
ppr
oach
sta
nds
f
or
both;
on it
s ow
n,
als
o for the
orth
odox c
onte
ntio
n
res
olu
ti
on m
eth
ods.
Each
of
0
UEs
se
le
ct
s
one
of
t
he
unif
or
m
ly
pilots
at
ra
ndom
unde
r
a
RA
blo
ck
,
UE
c
hoose
s
the
pilot
(
)
∈
{
1
,
2
,
.
.
.
.
,
}
.
Mo
reover
,
UE
ge
ner
al
ly
li
ke
to
be
act
ive
at
the
prese
nt
blo
c
k
wit
h
the
pro
ba
bili
ty
of
≤
1
,
wh
ic
h
is
pe
rm
anen
t
scen
ario
base
d
pa
r
a
m
et
er
that
shows
ho
w
the
UE
tra
ns
m
it
or
receive
the
da
ta
pac
kets.
Th
eref
or
e
,
eac
h
of
the
inact
ive
UE
tra
ns
m
it
s
a
s
pecific
pil
ot
se
quence
wi
th
the
⁄
pro
ba
bili
ty
.
The
set
is
gi
ven
be
that
consi
st
the
U
Es
in
di
ces
th
at
transm
it
t
pilot,
a
nd
transm
it
seq
ue
nces
has
a
b
i
nom
ia
l distri
bu
ti
on.
=
{
:
(
)
=
,
ℎ
>
0
}
(1)
|
|
~
(
0
,
⁄
)
(2)
Wh
e
re
,
is
de
note
s
the
ci
rcu
la
r
ly
-
sy
m
m
et
ric
com
po
sit
e
Ga
ussi
an
distrib
utio
n.
W
e
c
on
si
dered
t
hat
the
pilot
is
not
in
us
e
d
(
|
|
=
0
)
with
t
he
pro
ba
bili
ty
(
1
−
⁄
)
0
an
d
c
ho
os
e
only
one
UE
t
hat
is
(
|
|
=
1
)
w
it
h
the
0
⁄
(
1
−
⁄
)
0
−
1
pro
ba
bili
ty
.
Ther
e
fore,
t
he
colli
sio
n
of
RA
(
|
|
≥
2
)
occ
urs
duri
ng
a
t
arb
it
ra
ry p
il
ot
with the
pr
ob
a
bili
ty
o
f;
1
−
(
1
−
⁄
)
0
−
0
⁄
(
1
−
⁄
)
0
−
1
(3)
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
Ne
tw
or
k Eff
ic
ie
ncy e
nhancem
ent b
y
re
activ
e chan
nel state
base
d all
oca
ti
on
sch
e
me
(
U
mes
ha G. B
.
)
3279
In
a
m
assive
MIM
O
netw
ork
[
26]
,
the
c
ol
li
sion
s
are
nee
ded
to
recog
nize
an
d
re
so
l
ve
as
be
fore
of
UE
enteri
ng
at
payl
oad
bl
ocks.
O
ur
pro
po
s
ed
a
ppro
ac
h
is
a
distrib
uted
pr
ocess
in
order
t
o
r
esolve
the
colli
sion
s
at
UEs
by
util
iz
ing
the
m
ass
ive
MIM
O
pri
nciple.
Wh
e
re
the
channel
ve
ct
or
s
at
UE
c
an
be
represe
nted
as
∈
0
and
the
B
S
ca
n
be
giv
e
n
as
∈
C
,
w
her
e
we
co
ns
ide
r
a
ve
ry
gen
e
ral
propag
at
ion
m
od
el
an
d
t
he c
hannels a
re al
so
ca
n be
pr
e
-
a
ssu
m
ed
to sati
s
fy these t
wo cond
it
i
ons:
‖
d
‖
2
→
∞
→
,
∀
,
(4)
d
D
d
→
∞
→
0
,
∀
,
,
≠
,
(5)
Fo
r
so
m
e
stric
tl
y
value
of
that
is
kn
ow
n
to
UE,
th
us
th
e
channels
ca
n
of
fer
e
d
the
asy
m
pto
ti
c
favor
a
ble
pr
opagati
on
a
nd
c
hannel
rei
nforci
ng
[
27]
.
Whi
le
con
side
rin
g
ste
p
1
f
or
t
he
RAP
,
the
sign
al
receive
d
th
r
ough the B
S is
∈
C
×
fr
om
p
il
ot tran
sm
i
ssion can
b
e
g
i
ven as:
=
∑
√
ℎ
∈
0
d
(
)
T
+
+
(6)
w
he
re,
inte
r
-
ce
ll
interfer
e
nce
is
sy
m
bo
li
zed
thr
ough
the
,
∈
C
×
is
the
sel
f
-
deter
m
ining
recei
ve
r
noise
and
in
div
i
du
al
distri
bu
ti
on
of
el
em
ent
can
be
gi
ve
n
as
(
0
,
2
)
t
hat
is
a
c
om
pl
ex
Ga
us
sia
n
di
stribu
ti
on,
wh
e
re
the
m
a
trix
C
×
def
ines
the
inter
fer
e
nces
f
ro
m
the
di
f
fer
e
nt
a
djacen
t
cel
ls.
T
he
ti
m
e
-
fr
e
qu
e
nc
y
pro
per
ti
es
are
div
ide
d
int
o
th
e
coh
e
re
nce
bl
ocks
of
T
channel
and
that
re
pr
e
sents
N
oA.
T
r
ansm
it
po
wer
i
s
represe
nted
by
ℎ
,
C
show
s
t
he
co
m
plex
value
a
nd
tra
ns
m
it
te
d
pilot
is
re
pr
e
se
nted
by
.
T
hro
ugh,
c
orrelat
in
g
F w
it
h a
norm
a
li
zed
pil
ot sequen
ce
, th
e
BS
ob
ta
in
s as;
f
=
∗
‖
‖
=
∑
√
ℎ
‖
‖
d
=
+
∗
‖
‖
+
(7)
f
=
∑
√
ℎ
d
=
+
∗
‖
‖
+
(8)
w
he
re
,
‖
‖
is eq
ual
to
√
,
is t
he
cu
rr
e
nt
receive
r no
ise
and it ca
n b
e
gi
ven
a
s;
=
∗
‖
‖
~
(
0
,
2
)
(9)
w
he
re
,
de
no
te
s
the
×
identit
y
m
at
rix.
Re
cal
l
that
the
is
a
U
Es
set
w
hich
transm
itted
thr
ou
gh
pilot an
d, the
interf
e
re
nce
L
a
t i
nter
-
cel
l ca
n be
giv
e
n
as;
=
∑
L
κ
T
+
∑
∑
√
ℎ
,
Υ
,
,
T
∈
=
1
(10)
The
fi
rst
su
m
m
at
ion
in
ab
ove
(
10)
s
hows
the
interfe
rin
g
tra
ns
m
issi
on
of
data
that
c
arr
ie
d
out
at
neig
hbori
ng
ce
ll
s,
w
her
e
,
κ
sho
ws
ra
ndom
data
seq
ue
nce
a
nd
Υ
is
the
c
ha
nn
el
to
t
he
BS
.
In
re
sp
onse
qu
e
uing
ste
p
,
the
BS
re
plies
towa
rds
the
RAP
pilots
by
sen
ding
the
pr
e
-
co
de
d
do
wn
li
nk
si
gn
al
,
wh
ic
h
co
rr
e
sponds
t
o
each
RA
P
pilo
ts
that
wer
e
be
ing
us
e
d
at
the
up
li
nk.
The
re
act
ion
to
wards
is
the
seq
uence
of
c
ha
nn
el
∈
C
and,
t
he
dow
nl
ink
cha
nnel
se
qu
e
nces
ca
n
be
prese
nted
as
1
,
2
,
.
.
.
,
∈
C
that
a
re
ort
hogonal
to
eac
h
oth
e
r
a
nd
sa
ti
sfy
the
c
onditi
on
‖
‖
2
=
.
If
the
BS
c
on
si
der
the
no
rm
alized
co
njugate
the
n
∗
‖
‖
as
a
pr
e
-
cod
e
d
vect
or
at
pilot
se
quence
in
dow
nlin
k.
The
n
afte
r
the
sign
al
directe
d
to
the
m
ulti
-
ca
st
m
axi
m
u
m
ra
ti
o
of tran
sm
issi
on
at UEs
in
, so t
he
c
om
pl
et
e p
r
e
-
co
de
d dow
nlink
pilot si
gn
al
can
be give
n a
s;
=
√
∑
∗
‖
‖
=
1
T
(11)
w
he
re
,
t
he
down
li
nk
tra
ns
m
i
ts
power
an
d
ha
s
a
pre
-
def
i
ne
d
va
lue,
w
her
e
as
the
pilot
cha
nn
el
le
ng
t
h
is
giv
e
n
as
an
d
it
is in
de
pende
nt fro
m
the
NoA. T
he
r
ecei
ved
sig
nal
∈
C
at
the UE
∈
is;
T
=
d
T
+
T
+
T
(12)
w
he
re
,
h
T
shows
t
he
recip
ro
cal
dow
nlink
c
hannel,
cor
re
spo
nds
to
C
is
a
n
i
nter
-
cel
l
inte
rf
e
re
nce,
is
a
receive
r
noi
se
sign
al
.
Th
r
ough
co
rrel
at
ing
t
he
receiver
sign
al
with
t
he
no
rm
alized
dow
nlink
pilot
seq
uen
ce
, th
e
UE o
btain as;
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
.
3
,
J
une
2020
:
32
7
5
-
32
8
3
3280
=
T
∗
‖
‖
=
√
d
T
+
T
∗
‖
‖
+
(13)
w
he
re
,
the
e
ff
e
ct
ive
receive
r
no
ise
=
T
∗
‖
‖
~
(
0
,
2
)
,
the
refore
the
asy
m
pto
ti
c
favor
a
ble
prop
a
gati
on
is co
ns
ide
red.
√
⁄
=
√
(
d
T
)
∗
1
√
1
‖
‖
2
+
T
∗
√
‖
‖
+
√
(14)
→
∞
→
√
ℎ
√
∑
ℎ
+
+
2
=
The
noise
do
e
s
not
m
axi
m
ize
with
,
s
o
t
he
i
nter
-
cel
l
i
nterference
is
no
t
a
ff
ect
ed
by
.
Lat
er
on
the
ass
um
ption
is
m
ade
if
the
nea
rest
inte
rf
e
rin
g
cel
ls
tra
nsm
it
RA
pilots
thr
ough
assi
gn
dow
nlin
k
pilo
ts
in
diff
e
re
nt
m
e
tho
ds,
w
hich
c
a
us
es
the
m
inim
iz
at
ion
at
pilot
con
ta
m
inate
d
inter
fer
e
nc
e.
Co
ns
ide
rin
g
(14),
we
ca
n defi
ne
t
he
s
um
o
f
inte
r
fe
re
nce
gain
and si
gn
al
receiv
ed
at
BS
at u
pl
ink
t
ran
sm
issi
on
of
pilot
;
=
∑
ℎ
+
=
(15)
So
,
t
he
f
un
ct
io
n
(
.
)
pr
ovide
s
the
real
blo
c
k
of
it
s
inp
ut
a
nd
ba
sed
on
the
(14
),
the
a
ppr
ox
i
m
at
ion
can
be give
n
as
;
(
)
√
≈
√
ℎ
√
+
2
(16)
w
he
re
,
we
ha
ve
rem
ov
e
d
t
he
im
aginar
y
pa
r
t
of
,
w
hich
c
on
ta
in
s
t
he
est
i
m
ation
er
ror,
interfe
ren
ce
a
nd
no
ise
. T
her
e
for
e, the b
UE
can
u
se
the a
ppr
oxi
m
ation
i
n orde
r
to
est
i
m
at
e
;
̂
,
=
max
(
ℎ
2
2
(
(
)
)
2
−
2
,
ℎ
)
(17)
w
he
re
,
m
ax
functi
on
will
ta
ke
m
axi
m
u
m
of
two
val
ues
a
nd
this
est
i
m
at
or
beco
m
e
asy
m
pto
ti
cally
err
or
-
f
ree
wh
e
n
→
∞
(
14).
In
ste
p
3
pr
ovide
the
c
hannel
sta
te
analy
sis
and
res
ource
c
on
te
ntio
n
m
echan
is
m
to
reso
l
ve
the
cha
nnel
co
ntentio
ns
iss
ue
at
the
distri
bute
m
ann
er,
the
refor
e
eac
h
c
ha
nn
el
is
rep
eat
ed
ones
at
one
UE.
UE
∈
is
ta
ken
f
or
the
si
gn
al
gai
n
ℎ
and
,
est
im
a
te
̂
,
sign
al
gain
sum
that
con
te
ndin
g
UEs
a
nd
inter
-
cel
l
interf
eren
ce
.
The
B
S
ta
kes
the
RAP
re
petit
ive
pi
lot
transm
iss
ion
that
f
ollowe
d
by
upli
nk
m
essage
,
BS
ta
kes
the
ℎ
pilot
sig
nal
t
o
pr
e
dict
the
cha
nn
el
at
U
Es,
wh
ic
h
se
nt
t
he
and
decodes
th
e
co
rr
es
ponds
m
essage.
Af
te
rw
a
rd
s
the
suc
cessf
ul
dec
od
ing
,
the
B
S
a
ckno
wled
ged
on
e
U
Es
in
to
pro
vid
e
c
o
herent
blo
c
ks
of
payl
oad
by
al
locat
ing
the
pilot
channel
seq
ue
nce.
This
al
lo
cat
ion
res
ult
i
s
fo
r
wa
rd
e
d
towa
r
ds
the
dow
nlin
k
in
ste
p
4
a
nd
sa
m
e
fo
r
the
st
ep
2.
T
he
RC
SA
ap
proac
h
is
rep
eat
e
d
for
the
fixe
d
num
ber
of
interval,
if t
he UEs
were
no
t
r
ecognise
d
i
n st
ep 4,
t
hen it
’s
i
ns
tr
ucted
t
o
tra
ns
m
it
v
ia
a n
e
w
RA
P c
ha
nn
e
ls.
4.
RESU
LT
S
AND A
N
ALYSIS
Her
e
,
we
pro
vi
de
re
su
lt
a
naly
sis
us
in
g
our
pro
posed
RC
SA
ap
proach
that
com
par
ed
with
the
ve
rtex
gr
a
ph
-
c
olorin
g
-
base
d
pilot
a
ssign
m
ent
(VGCPA
)
a
ppr
oa
ch
[
28]
.
The
pro
po
se
d
sc
hem
e
exh
ibit
s
the
f
our
m
ajo
r
ste
ps
;
syst
e
m
init
ia
li
z
at
ion
an
d
que
ryi
ng,
res
ponse
qu
e
uing,
c
ha
nn
el
sta
te
an
al
ysi
s
and
res
ource
con
te
ntion,
a
nd
re
sourc
e
al
lo
cat
ion
,
w
hich
al
lows
decr
ea
s
ing
t
he
delay
tim
e
and
fail
ed
acce
ss
at
te
m
p
ts,
al
so
increases
the
end
-
to
-
en
d
suc
cess
rate.
I
n
orde
r
to
vali
date
the
res
ult
analy
sis
two
scena
rios
has
be
e
n
co
nsi
der
e
d,
where
al
l
the
oth
er
par
am
et
ers
are
sa
m
e
excep
t
the
nu
m
ber
of
BS.
The
pa
ra
m
et
ers
are
sh
own
i
n
T
able
1,
wh
ic
h
has
been
ap
plied
be
f
or
e
sim
ulati
on
of
c
od
e
.
Her
e
we
will
com
pu
te
the
th
r
ee
m
ajo
r
thin
gs
that
cause
the
PC
in
a
netw
ork
su
ch
as;
e
nd
-
to
-
e
nd
su
cc
es
s
rate,
delay
tim
e
in
m
ille
seconds
a
nd
fail
e
d
acce
ss att
em
pt
s.
In
sce
na
rio
A,
we
com
pu
te
d
end
-
to
-
en
d
s
uc
cess
rate,
delay
tim
e
in
m
ille
seco
nd
s
an
d
f
ai
le
d
acce
ss
at
tem
pts
at
100
nu
m
ber
of
BS.
Fi
gure
3
sh
ows
en
d
-
to
-
end
s
uccess
ra
te
,
w
her
e
our
pro
po
s
e
d
m
odel
go
t
72.83%
of
im
pro
vem
ent
w
ith
res
pect
to
V
GCPA.
Fig
ur
e
4
show
s
dela
y
tim
e
in
m
se
c,
where
our
pro
pose
d
m
od
el
go
t
43.
63%
le
ss
a
verage
delay
tim
e
with
res
pect
to
VG
C
PA
.
Figure
5
s
ho
ws
t
he
fail
e
d
acce
ss
at
tem
pts,
wh
e
r
e
the
m
ean
fai
le
d
acce
ss
at
te
m
pts
fr
om
our
RC
SA
schem
e
is
0.0
844,
w
her
eas
f
r
om
VG
CPA
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
Ne
tw
or
k Eff
ic
ie
ncy e
nhancem
ent b
y
re
activ
e chan
nel state
base
d all
oca
ti
on
sch
e
me
(
U
mes
ha G. B
.
)
3281
we go
t
0.7
5.
Sim
il
arly
co
m
ing
to
sce
nar
i
o
-
B,
wh
e
re t
he
c
on
s
idere
d nu
m
ber
of
B
S is
200. F
igure
6
s
hows end
-
to
-
e
nd
s
uccess
rate,
wh
e
re
our
propose
d
m
od
el
go
t
73.
38
%
of
im
pr
ov
e
m
ent
com
par
ed
to
V
GCP
A
m
od
el
.
F
igure
7
s
how
s
delay
tim
e,
wh
e
re
our
pro
po
s
ed
m
od
el
go
t
43.
68%
le
ss
aver
a
ge
del
ay
tim
e
co
m
par
ed
to
VG
CP
A.
Fig
ure
8
sho
ws
t
he
fail
ed
acce
ss
at
tem
pts,
wh
e
r
e
the
m
ean
fail
ed
acce
ss
at
te
m
pt
fr
om
ou
r
RC
SA
schem
e is 0.
08
55, whic
h
is
88
% less c
om
par
ed
to
ex
ist
in
g
VG
CP
A
a
ppr
oa
ch.
Table
1
.
Syst
em
p
ara
m
et
ers
Serial
No
.
Para
m
eters
Para
m
eters
Values
1
Mon
te
-
Carlo
r
ealiz
atio
n
s
500
2
BS an
ten
n
as
1
0
0
or 20
0
3
Nu
m
b
e
r
o
f
UE
s
1
0
0
0
to 2
0
0
0
0
4
Maxi
m
u
m
nu
m
b
er
of
atte
m
p
ts
10
5
Prob
ab
ility
o
f
r
etra
n
s
m
ittin
g
0
.5
6
RA p
ilo
t
sig
n
als
10
7
Tr
an
s
m
it
po
wer
o
f
UE
s
1
dB
8
Tr
an
s
m
it
po
wer
o
f
the BS
1
dB
9
No
ise v
ariance
1
dB
10
Stan
d
ard d
ev
iatio
n
of
f
ad
in
g
1
0
dB
11
Cell Rad
iu
s
2
5
0
m
12
Nu
m
b
e
r
o
f
active
UEs in n
eig
h
b
o
ring
cells
10
Scena
rio
-
A
Figure
3.
En
d
-
to
-
e
nd s
uccess
r
at
e
Figure
4. Dela
y t
i
m
e (
m
s)
Figure
5.
Fail
ed
acce
ss
att
em
pts
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
.
3
,
J
une
2020
:
32
7
5
-
32
8
3
3282
Scena
rio
-
B
Figure
6.
En
d
-
to
-
e
nd s
uccess
r
at
e
Figure
7.
Dela
y t
i
m
e (
m
s)
Figure
8.
Fail
ed
acce
ss
att
em
pts
5.
CONCL
US
I
O
N
In
m
assive
MIM
O
syst
e
m
,
t
he
PC
is
a
bi
g
issue
that
cau
sed
by
reu
si
ng
si
m
il
ar
pilot
sequ
e
nce
a
t
cel
ls.
The
pro
pose
d
sch
em
e
e
xh
i
bits
the
f
ou
r
m
ajo
r
ste
ps
,
wh
ic
h
al
lo
ws
decr
easi
ng
the
delay
tim
e
and
fail
ed
acce
ss
at
tem
pt
s,
al
so
increas
es
the
end
-
to
-
e
nd
s
uccess
rate
.
In
reacti
on
of
qu
e
uing
the
BS
respo
ns
es
th
rou
gh
sen
ding
the
down
li
nk
c
hanne
ls
that
are
pre
-
cod
e
d
th
r
ough
the
cha
nnel
est
i
m
at
es,
wh
ic
h
res
ults
at
spa
ti
al
ly
aim
ed
sign
al
s
towa
rd
the
UE
s
an
d
offer
the
su
it
able
R
AP
resou
rce.
T
he
channel
sta
te
a
naly
sis
an
d
res
ource
con
te
ntion
m
echan
ism
is
consi
der
e
d
f
or
t
he
non
-
c
olli
ding
channel
tra
nsm
issi
on
is
consi
der
a
bly
incre
ased
in
m
assive
MIM
O
netw
ork
.
So
RC
SA
ap
proa
ch
al
so
al
lo
ws
the
m
assive
MIM
O
syst
em
to
ackno
wled
ge
UEs
unde
r
the
c
rowd
e
d
sce
nar
i
os.
I
n
the
res
ult
analy
sis
sect
ion
our
pro
pos
ed
RC
SA
a
ppr
oach
is
c
om
par
ed
with
the
V
GCP
A
a
ppr
oach,
an
d
we
s
howe
d
th
at
our
propose
d
m
od
el
has
pe
rfor
m
ed
co
nsi
der
a
bly
well
at
bot
h
consi
der
e
d sce
nar
i
o wh
ic
h va
li
date the
perform
ance o
f o
ur
m
od
el
.
REFERE
NCE
S
[1]
T.
L
.
Marz
e
tt
a
,
“
Noncoope
rat
iv
e
ce
ll
u
la
r
wire
l
ess
with
unli
m
i
te
d
num
ber
s
of
base
stat
ion
a
nte
nnas,
”
in
IE
EE
Tr
ansacti
ons on Wirele
ss
Comm
unic
ati
ons
,
vol
.
9
,
no.
11,
pp.
3590
-
3600,
Nov
.
201
0.
[2]
H.Q.
Ngo,
E
.
G.
La
rss
on,
and
T.
L
.
Marz
etta,
“
Ene
rg
y
and
sp
ec
tr
al
effic
ie
n
c
y
of
ver
y
l
arg
e
m
ult
iuser
MIMO
s
y
stems
,
”
in
I
EEE
Tr
ansacti
ons
on
Comm
unic
ations
,
vol.
61,
no.
4,
pp
.
1436
-
144
9,
Apr.
2013.
[3]
Adee
b
Salh,
Lu
km
an
Audah,
N
or
Shahi
da
M.
Shah,
Shipun
A
.
Ham
za
h,
“
Maximizi
ng
Ene
rg
y
Eff
i
cienc
y
fo
r
Consum
pti
on
Circ
uit
Pow
er
in
D
ownlink
Ma
ss
ive
MIM
O
W
ire
le
ss
Netwo
rks,”
in
Inte
rna
ti
onal
Journal
of
El
e
ct
rica
l
and
C
omputer
Engi
n
e
ering
(
IJE
CE)
,
v
ol.
7
,
n
o
.
6
,
pp
.
2
977~2985,
Dec
.
2017.
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
Ne
tw
or
k Eff
ic
ie
ncy e
nhancem
ent b
y
re
activ
e chan
nel state
base
d all
oca
ti
on
sch
e
me
(
U
mes
ha G. B
.
)
3283
[4]
Anuj
Singal
,
De
epa
k
Kedia,
“
Perform
anc
e
Anal
y
sis
of
Antenna
Sele
ction
Tech
nique
s
in
MIMO
-
OF
D
M
Sy
ste
m
with
Hardwar
e
I
m
pai
rm
ent
s:
En
erg
y
Eff
ic
i
ency
per
spec
t
ive,”
in
Inte
rnational
Jo
urnal
of
El
e
ct
ric
al
and
Comput
e
r
Engi
ne
ering
(
IJ
ECE
)
,
vol
.
8
,
n
o
.
4,
pp.
2272~2279,
Aug
.
2018.
[5]
Y.
Xin,
D.
W
a
ng,
and
J.
Li,
“
Area
spec
t
ral
ef
ficie
nc
y
and
ar
e
a
ene
rg
y
effic
i
e
nc
y
of
m
assive
MIM
O
ce
ll
ula
r
s
y
stems
,
”
in
I
EEE
Tr
ansacti
ons
on
Ve
h
ic
ular
Te
chnol
ogy
,
vo
l. 6
5,
no
.
5
,
pp
.
324
3
-
3254,
Ma
y
20
16.
[6]
Hus
sein
A
.
Le
fta
h
,
Huda
N.
Alm
inshid,
“
Channe
l
ca
p
ac
i
t
y
and
per
form
anc
e
ev
a
lua
ti
on
of
pre
co
dedMIMO
-
OFD
M
s
y
stem
with
la
r
ge
-
size
const
el
l
at
ion
,
”
Int
ernational
Journal
of
El
ectric
al
a
nd
Computer
Engi
nee
ring
(
IJE
CE
)
,
v
ol.
9
,
no
.
6
,
pp
.
5024
-
5030
,
De
c
.
2019
.
[7]
L.
Zh
ang,
Y.
Xin,
and
Y.
C
.
Li
ang
,
“
W
ei
ghte
d
sum
rat
e
o
pti
m
iz
ation
for
cogni
ti
v
e
rad
io
MIM
O
broa
dca
st
cha
nne
ls,”
in
IE
EE
Tr
ansacti
ons
on
Wire
le
ss
Co
mm
unic
ati
ons
,
v
o
l.
8
,
no
.
6
,
pp
.
2
950
-
2959,
Jun.
2
009.
[8]
L.
W
ang,
H.
Q.
Ngo,
M.
El
k
ashl
an,
T.
Q.
Duong,
and
K
.
W
ong,
“
Mass
ive
MIMO
in
s
pec
trum
sharing
ne
tworks:
ac
hi
eva
bl
e
r
ate
a
nd
power efficie
nc
y
,
”
in
IEEE
S
y
stems J
ournal
,
v
ol.
11
,
no
.
1
,
pp
.
20
-
31,
Mar
.
201
7.
[9]
M.
Fili
ppou,
D.
Gesbert
,
and
H.
Yin,
“
Dec
on
ta
m
ina
ti
ng
pil
o
t
s
in
cogni
ti
v
e
m
assive
MIM
O
net
works
,
”
201
2
Inte
rnational
Sy
mpos
ium
on
Wirel
ess Comm
unication
S
yste
ms
(
ISWCS)
,
Paris,
p
p.
816
-
820
,
201
2.
[10]
B.
Kouass
i,
I
.
Ghauri
,
and
L
.
Dene
ire,
“
Recip
roc
ity
-
b
ase
d
cog
nit
ive
tra
nsm
issions
using
a
MU
m
assive
MIM
O
appr
oac
h
,
”
2013
IEEE
In
te
rnatio
nal
Conf
ere
nce
on
Comm
unic
ations
(
ICC
)
,
Buda
pest,
pp
.
2738
-
2
742
,
2013
.
[11]
H.
Xie,
B
.
W
an
g,
F.
Gao
and
S
.
Jin,
“
A
full
-
sp
ac
e
spe
ct
rum
-
sharing
stra
te
g
y
fo
r
m
assive
MIMO
cogni
ti
v
e
rad
i
o
s
y
stems
,
”
in
I
EEE
Journal
on
S
elec
t
ed
Ar
eas
in
Comm
unic
ati
ons
,
vol.
34
,
no
.
10
,
p
p.
2537
-
2549
,
O
ct
.
2016.
[12]
J.
Jos
e,
A.
As
hi
khm
in,
T.
L.
Ma
rze
t
ta
,
and
S.
Vi
shw
ana
th,
“
PC
and
pre
codi
ng
i
n
m
ult
i
-
cell
TDD
s
y
stems
,
”
IE
E
E
Tra
ns.
Com
m
un.
,
vol
.
10
,
no
.
8
,
p
p.
2640
–
2651
,
2
011.
[13]
L.
Sanguin
etti,
A.
A.
D’A
m
ic
o,
M.
Morell
i
,
and
M.
Debba
h
,
“
Random
ac
c
ess
in
upli
nk
m
assive
MIM
O
sy
st
ems
:
How
to
expl
oit
as
y
nchr
oni
ci
t
y
and
exc
ess
ant
enna
s,
”
in
2
016
IEE
E
Global
Comm
unic
ations
Confe
renc
e
(
G
LO
BE
COM
)
,
W
ashingt
on,
DC
,
pp
.
1
-
5
,
2016
.
[14]
J.
H.
Sørensen,
E
.
de
Carvalho,
a
nd
P.
Popovs
ki,
“
Mass
ive
MIMO
for
cro
wd
sce
nar
ios:
A
soluti
o
n
base
d
on
ran
do
m
ac
c
ess,”
in
2014
IEE
E
Globe
com Wor
kshops (
GC
Wk
shps
)
,
Aus
ti
n,
TX
,
pp.
352
–
35
7
,
2014
.
[15]
E.
de
Carvalho,
E.
Bj
ornson,
E.
G.
La
rss
on,
and
P.
Popovs
ki,
“Ra
ndom
ac
ce
ss
for
m
assive
MI
MO
sy
stems
with
int
ra
-
ce
l
l
PC
,
”
in
Proc. I
E
EE ICA
SSP
,
2016
.
[16]
F.
Rusek,
D.
Perss
on,
B.
K.
La
u
,
E.
G.
L
arsson,
T.
L
.
Marz
et
t
a,
O.
Edfor
s
and
F.
Tufve
ss
on,
“
Scaling
up
MIM
O:
Opportunit
ie
s
an
d
challe
ng
es
wit
h
ver
y
la
rg
e array
s,
”
i
n
IE
EE Si
g
nal
Proc
essing
Magazine
,
vo
l. 3
0,
no
.
1
,
2012
.
[17]
L.
Fa
n
,
S.
Jin,
C.
W
en
and
H.
Zh
ang,
“
Uplink
A
c
hie
vab
le
R
at
e
fo
r
Mass
ive
MIM
O
S
y
stems
W
it
h
Low
-
Resolution
AD
C,
”
in
IEEE C
omm
unic
ati
ons Let
t
ers
,
vol
.
19
,
no.
12
,
pp
.
2186
-
2189,
De
c. 201
5.
[18]
S.
Jin,
M.
Li,
Y.
Huang,
Y.
D
u,
and
X.
Gao
,
“
Pilot
sche
duli
n
g
sche
m
es
for
m
ult
i
-
cell
m
assive
m
ult
iple
-
inpu
t
-
m
ult
ipl
e
-
outpu
t t
ran
sm
ission,”
in
IET
Comm
unic
a
ti
ons
,
vo
l. 9, no.
5,
pp
.
685
-
700
,
Mar.
2015
.
[19]
N.
Zha
ng,
Z
.
Ba
i,
B.
Zh
ang,
Y.
Su,
T.
Han
and
K.
S.
Kw
ak,
“
G
ene
t
ic
Algorit
hm
base
d
Pilot
Allo
ca
t
ion
Scheme
for
Mass
ive
MIM
O
S
y
stem,”
2016
IEE
E
Inte
rnat
i
onal
Confe
ren
c
e
on
Comm
unication
Syst
ems
(
I
CCS)
,
Shenzhe
n
,
pp.
1
-
5
,
2016
.
[20]
A.
M.
Girgis,
B
.
Abdelha
m
id
an
d
S.
El
r
aml
y
,
“
Im
prove
d
Pilot
Sequences
Alloca
ti
on
in
Mass
ive
MIM
O
Sy
stems
,
”
2017
IEE
E
Wir
el
ess
Comm
unic
ati
ons
and
Networ
ki
ng
Confe
r
enc
e
Workshops
(
WCNCW
)
,
Sa
n
F
ran
ci
sco,
C
A
,
pp.
1
-
6
,
2017
.
[21]
Sri
nu
Py
l
a,
K.
R
aj
u,
and
N.
Subrahm
an
y
am,
“
Perform
anc
e
Anal
ysis
of
Adapti
ve
Filt
er
Channel
E
stim
at
ed
MIMO
OF
DM
Co
mmu
nic
a
ti
on
S
y
st
em,
”
Inte
rnat
ional
Journal
of
Elec
t
rical
and
Computer
Engi
n
ee
rin
g
(
IJE
CE)
,
vol.
8,
no.
5
,
pp
.
3829
-
3828,
Oct
.
2018
.
[22]
R.
Zhou,
Y.
Fu
and
H.
W
ang,
“
Uplink
As
y
nchr
onous
Frac
ti
ona
l
Pilot
s
Sch
eduli
ng
in
Mass
ive
MIM
O
Sy
stem,
”
2018
IEEE
18
th Inte
rnational
Co
nfe
renc
e
on
Co
mm
unic
ati
on
Te
chnol
ogy
(
ICCT
)
,
Chongqing
,
pp
.
402
-
406
,
2018
.
[23]
La
m
ia
e
Ha
cha
d
,
Fati
ha
Mr
abti,
Hicha
m
Ghenni
oui,
Mohcin
e
Zo
uak,
“
Pilot
Dec
o
nta
m
ina
ti
on
ov
e
r
Ti
m
e
Freque
n
c
y
andSpac
e
Dom
ai
ns
in
Multi
-
Cel
l
Mass
ive
MIM
OS
y
stem,
”
Inte
rnational
Jour
nal
of
El
ectric
a
l
and
Computer
Engi
ne
ering
(
IJ
ECE
)
,
vol
.
8
,
n
o.
5,
pp.
3609
-
361
9,
Jan
.
2018
.
[24]
Y.
Li
u,
M.
De
rak
hshani,
S.
Parsae
ef
ar
d,
S.
La
m
botha
ran
a
nd
K.
W
ong,
“
Antenna
Allocati
on
and
Pric
i
ng
inVirt
ualiz
ed
Mass
ive
MIM
O
Networks
via
Stac
k
el
ber
g
Gam
e
,”
in
IE
EE
Tr
ansacti
ons
on
Comm
unic
ati
ons
,
vol.
66
,
no
.
11
,
p
p.
5220
-
5234
,
N
ov.
2018
.
[25]
H.
Gao
,
T.
Lv,
X.
Su,
H.
Yan
g
and
J.
M.
Ci
offi,
“
Ene
rg
y
-
Ef
fic
i
ent
Resourc
e
Alloc
at
ion
for
Mass
ive
MIM
O
Am
p
li
f
y
-
and
-
Fo
rward
Rela
y
S
y
s
te
m
s,”
in
IE
EE
Ac
c
ess
,
vol
.
4
,
p
p.
2771
-
2787
,
2
016.
[26]
H.
Huh,
G.
Cai
re,
H.
C.
Papado
poulos,
and
S.
A.
Ramprashad,
“
Achie
ving
“m
assive
MIM
O”
s
pe
ct
r
al
eff
i
cie
n
c
y
with
a
not
-
so
-
l
ar
ge
NoA
,
”
IE
EE
Tr
ans.
Wirel
ess
Comm
un
.
,
vol
.
1
1,
no
.
9
,
pp
.
322
6
–
3239,
2012
.
[27]
R.
Couil
let
and
M.
Debba
h,
“
Random
Matri
x
Methods
for
W
ire
le
ss
Com
muni
cations
,
”
Ca
m
bridge
Univer
sit
y
Press
,
2011.
[28]
H.
T.
Dao
and
S
.
Kim
,
“
Vert
ex
Gr
aph
-
Colori
ng
-
Based
Pilot
As
signm
ent
W
it
h
Lo
ca
t
ion
-
Based
Ch
anne
l
Esti
m
ation
for
Mass
ive
MI
MO
Sy
st
ems
,
”
i
n
IEEE Access
,
vol.
6
,
pp
.
4599
-
4607,
2018
.
Evaluation Warning : The document was created with Spire.PDF for Python.