I
nd
o
ne
s
ia
n J
o
urna
l o
f
E
lect
rica
l En
g
ineering
a
nd
Co
m
p
u
t
er
Science
Vo
l.
1
2
,
No
.
2
,
N
o
v
e
m
b
er
201
8
,
p
p
.
6
9
9
~
7
0
5
I
SS
N:
2502
-
4752
,
DOI
: 1
0
.
1
1
5
9
1
/
i
j
ee
cs
.
v
1
2
.i
2
.
p
p
699
-
7
0
5
699
J
o
ur
na
l ho
m
ep
a
g
e
:
h
ttp
:
//ia
e
s
co
r
e.
co
m/jo
u
r
n
a
ls
/in
d
ex
.
p
h
p
/
ijeec
s
I
m
ple
m
e
ntatio
n o
f
Dee
p Lea
rning
i
n Spa
tial M
ult
ipl
ex
ing
M
IM
O
Co
mm
uni
ca
tion
M
a
hd
in Ro
h
m
a
t
illa
h
1
,
H
a
di Su
y
o
no
2
,
Ra
h
m
a
dw
a
t
i
3
,
Sh
o
l
eh
H
a
di P
ra
m
o
no
4
1
De
p
a
rtme
n
t
o
f
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
,
Na
ti
o
n
a
l
S
u
n
Ya
t
-
S
e
n
U
n
iv
e
rsit
y
,
Ka
o
h
siu
n
g
,
8
0
4
,
T
a
iw
a
n
,
R.
O.C
1,
2,
3,
4
El
e
c
tri
c
a
l
En
g
in
e
e
rin
g
De
p
a
r
tm
e
n
t,
Un
iv
e
rsitas
Bra
w
ij
a
y
a
,
Jln
.
M
T
.
Ha
r
y
o
n
o
1
6
7
,
M
a
lan
g
6
5
1
4
5
,
I
n
d
o
n
e
sia
Art
icle
I
nfo
AB
ST
RAC
T
A
r
ticle
his
to
r
y:
R
ec
eiv
ed
A
p
r
9
,
2
0
1
8
R
ev
i
s
ed
Ma
y
2
0
,
2
0
1
8
A
cc
ep
ted
Ju
l
11
,
2
0
1
8
Re
se
a
rc
h
in
M
u
lt
ip
le
In
p
u
t
M
u
lt
i
p
le
Ou
tp
u
t
(M
IM
O)
c
o
m
m
u
n
ica
ti
o
n
sy
ste
m
h
a
s
b
e
e
n
d
e
v
e
lo
p
e
d
ra
p
id
ly
in
o
rd
e
r
t
o
im
p
ro
v
e
th
e
e
ff
e
c
ti
v
e
n
e
ss
o
f
c
o
m
m
u
n
ica
ti
o
n
a
m
o
n
g
u
se
rs.
Ho
w
e
v
e
r,
trad
e
-
o
ff
p
h
e
n
o
m
e
n
o
n
b
e
tw
e
e
n
p
e
rf
o
r
m
a
n
c
e
a
n
d
c
o
m
p
u
tatio
n
a
l
c
o
m
p
lex
it
y
a
l
wa
y
s
b
e
c
o
m
e
th
e
h
u
g
e
st
d
il
e
m
m
a
su
ff
e
r
e
d
b
y
re
s
e
a
rc
h
e
r
s.
A
s
a
n
a
lt
e
rn
a
ti
v
e
so
lu
ti
o
n
,
t
h
is
p
a
p
e
r
p
ro
p
o
se
s
a
n
o
p
ti
m
iza
ti
o
n
in
3
x
3
sp
a
ti
a
l
m
u
lt
ip
lex
in
g
M
IM
O
c
o
m
m
u
n
ica
ti
o
n
s
y
ste
m
u
sin
g
e
n
d
-
to
-
e
n
d
b
a
se
d
l
e
a
rn
in
g
,
sp
e
c
if
ica
ll
y
,
it
a
d
a
p
ts
a
u
to
e
n
c
o
d
e
r
b
a
se
d
m
o
d
e
l
w
it
h
th
e
k
n
o
w
led
g
e
o
f
Ch
a
n
n
e
l
S
tate
In
f
o
rm
a
ti
o
n
(
CS
I)
in
th
e
re
c
e
iv
e
r
sid
e
,
m
a
k
e
it
f
a
irl
y
c
o
m
p
a
re
d
w
it
h
th
e
b
a
se
li
n
e
m
e
th
o
d
.
T
h
e
p
ro
p
o
se
d
m
o
d
e
ls
w
e
re
e
v
a
lu
a
te
d
in
o
n
e
o
f
th
e
m
o
st
c
o
m
m
o
n
c
h
a
n
n
e
l
im
p
a
ir
m
e
n
t
w
h
ich
is
f
a
st
Ra
y
lei
g
h
f
a
d
in
g
w
it
h
a
d
d
it
i
o
n
a
l
A
d
d
i
t
iv
e
W
h
it
e
G
a
u
ss
ian
No
ise
(
AWG
N).
B
y
a
p
p
ro
p
riate
ly
d
e
ter
m
in
in
g
h
y
p
e
r
p
a
ra
m
e
ters
a
n
d
t
h
e
h
e
lp
o
f
P
Re
L
U
(P
a
ra
m
e
t
ric
Re
c
ti
f
ied
L
in
e
a
r
Un
it
),
th
e
re
su
lt
s
sh
o
w
th
a
t
th
is
a
u
t
o
e
n
c
o
d
e
r
b
a
se
d
M
IM
O
c
o
m
m
u
n
ica
ti
o
n
s
y
ste
m
r
e
su
lt
s
in
v
e
r
y
p
ro
m
isin
g
re
su
lt
s
b
y
e
x
c
e
e
d
in
g
th
e
b
a
se
li
n
e
m
e
th
o
d
s
(m
e
th
o
d
s
w
id
e
ly
u
se
d
in
c
o
n
v
e
n
ti
o
n
a
l
M
IM
O
c
o
m
m
u
n
i
c
a
ti
o
n
)
b
y
re
a
c
h
in
g
BER
lo
w
e
r
th
a
n
a
t
S
NR 2
2
.
5
d
B
.
K
ey
w
o
r
d
s
:
Au
to
en
co
d
er
E
n
d
-
to
-
en
d
lear
n
i
n
g
MI
MO
co
m
m
u
n
icatio
n
Sp
atial
Mu
lt
ip
lex
i
n
g
Co
p
y
rig
h
t
©
2
0
1
8
In
stit
u
te o
f
A
d
v
a
n
c
e
d
E
n
g
i
n
e
e
rin
g
a
n
d
S
c
ien
c
e
.
Al
l
rig
h
ts
re
se
rv
e
d
.
C
o
r
r
e
s
p
o
nd
ing
A
uth
o
r
:
Ma
h
d
in
R
o
h
m
atilla
h
,
Dep
ar
t
m
en
t o
f
E
lectr
ical
E
n
g
i
n
ee
r
in
g
,
Natio
n
al
S
u
n
Yat
-
Sen
U
n
i
v
er
s
it
y
,
Kao
h
s
i
u
n
g
,
8
0
4
,
T
ai
w
an
,
R
.
O.
C
.
E
m
ail:
r
o
h
m
atil
lah
m
a
h
d
in
1
9
9
4
@
g
m
ail.
co
m
1.
I
NT
RO
D
UCT
I
O
N
T
h
e
u
tili
za
tio
n
o
f
s
ev
er
al
a
n
t
en
n
a
s
eit
h
er
at
tr
an
s
m
itter
o
r
r
ec
eiv
er
o
r
at
b
o
th
o
f
t
h
e
m
h
as
b
ec
o
m
e
m
o
r
e
p
o
p
u
lar
n
o
w
ad
a
y
s
d
u
e
t
o
its
ab
ilit
y
to
m
ain
tain
a
r
eli
ab
le
co
m
m
u
n
icatio
n
in
a
w
ir
eless
c
h
an
n
el
w
it
h
s
o
m
e
i
m
p
air
m
e
n
t
p
r
ed
o
m
i
n
an
tl
y
b
y
f
ad
i
n
g
.
T
h
is
r
elia
b
le
co
m
m
u
n
icatio
n
ca
n
b
e
m
ai
n
tai
n
ed
b
ec
au
s
e
m
u
ltip
le
an
ten
n
a
s
tech
n
o
lo
g
y
p
r
o
v
id
es
b
en
ef
it
s
i
n
a
co
m
m
u
n
ica
tio
n
s
y
s
te
m
w
h
ich
ar
e
ar
r
a
y
g
ai
n
,
s
p
atial
d
iv
er
s
i
t
y
o
r
s
p
atial
m
u
ltip
le
x
i
n
g
g
ai
n
an
d
i
n
ter
f
er
e
n
ce
r
ed
u
ctio
n
a
n
d
av
o
i
d
an
ce
[
1
]
.
Fo
r
y
ea
r
s
,
r
esear
c
h
er
s
h
a
v
e
b
e
en
d
ev
elo
p
in
g
al
g
o
r
ith
m
s
i
n
m
u
ltip
le
an
te
n
n
as
tec
h
n
o
lo
g
y
in
o
r
d
er
to
i
m
p
r
o
v
e
it
s
p
er
f
o
r
m
an
ce
eith
e
r
in
d
etec
tio
n
ta
s
k
o
r
ch
a
n
n
el
esti
m
atio
n
tas
k
o
r
o
th
er
tas
k
s
.
Ho
w
e
v
er
,
th
e
is
s
u
e
o
f
a
tr
ad
e
-
o
f
f
b
et
w
ee
n
p
er
f
o
r
m
an
ce
i
m
p
r
o
v
e
m
en
t
a
n
d
co
m
p
u
tat
io
n
al
co
m
p
le
x
it
y
al
w
a
y
s
b
ec
o
m
e
a
m
ai
n
r
estrictio
n
a
n
d
co
n
s
id
er
atio
n
.
As
a
s
o
l
u
tio
n
,
m
ac
h
i
n
e
lear
n
i
n
g
,
a
n
ap
p
r
o
ac
h
s
h
in
in
g
n
o
w
ad
ay
s
esp
ec
iall
y
i
n
d
o
m
ai
n
s
s
u
c
h
as
co
m
p
u
ter
v
i
s
io
n
,
is
i
n
tr
o
d
u
ce
d
in
m
u
ltip
l
e
an
ten
n
a
s
co
m
m
u
n
icat
io
n
s
y
s
te
m
.
As
a
r
es
u
lt,
it
p
er
f
o
r
m
s
v
er
y
w
ell
a
n
d
ev
en
b
etter
co
m
p
ar
ed
to
th
e
b
aseli
n
e
m
et
h
o
d
s
.
So
m
e
o
f
t
h
e
m
o
s
t
i
n
ter
es
tin
g
r
esu
l
ts
o
f
m
ac
h
in
e
lear
n
in
g
i
m
p
le
m
e
n
tatio
n
in
a
co
m
m
u
n
icat
io
n
s
y
s
te
m
ar
e
p
ap
er
titl
ed
An
I
n
tr
o
d
u
ctio
n
to
Dee
p
L
ea
r
n
i
n
g
f
o
r
th
e
P
h
y
s
ical
L
a
y
er
[
2
]
an
d
Dee
p
_
L
ea
r
n
in
g
-
B
ased
C
o
m
m
u
n
icat
io
n
o
v
er
t
h
e
A
ir
[
3
]
w
h
ich
in
tr
o
d
u
ce
d
ee
p
lear
n
in
g
a
s
an
en
d
-
to
-
e
n
d
s
y
s
te
m
in
SISO
co
m
m
u
n
icatio
n
.
T
h
is
e
n
d
-
to
-
en
d
m
o
d
el
m
ea
n
s
th
at
tr
an
s
m
itter
,
c
h
a
n
n
el
i
m
p
air
m
e
n
ts
,
an
d
r
ec
ei
v
er
ar
e
r
ep
r
esen
ted
b
y
o
n
e
o
r
s
ev
er
al
n
eu
r
al
n
et
w
o
r
k
la
y
er
(
d
en
s
e
)
th
en
in
ter
p
r
et
t
h
e
w
h
o
le
s
y
s
te
m
as a
n
a
u
to
en
co
d
er
,
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l
.
1
2
,
No
.
2
,
No
v
e
m
b
er
201
8
:
6
9
9
–
7
0
5
700
a
p
o
w
er
f
u
l
m
et
h
o
d
f
o
r
p
er
f
o
r
m
in
g
u
n
s
u
p
er
v
i
s
ed
lear
n
in
g
[
4
]
.
Sin
ce
th
e
y
s
h
o
w
g
o
o
d
r
esu
lts
,
r
esear
ch
e
s
r
elate
d
to
au
to
en
co
d
er
i
m
p
le
m
en
tatio
n
i
n
MI
MO
co
m
m
u
n
ica
tio
n
h
as b
ee
n
d
ev
elo
p
in
g
r
ap
i
d
ly
,
f
o
r
i
n
s
ta
n
ce
i
ts
ap
p
licatio
n
in
ch
a
n
n
el
d
ec
o
d
in
g
[
5
]
an
d
Or
th
o
g
o
n
al
Fre
q
u
en
c
y
Di
v
is
io
n
Mu
l
tip
lex
i
n
g
(
OF
DM
)
[
6
]
.
Ho
w
ev
er
,
th
e
n
ee
d
o
f
i
m
p
r
o
v
e
m
e
n
t
i
n
t
h
is
to
p
ic
is
s
t
ill
r
eq
u
ir
ed
esp
ec
i
all
y
i
n
en
d
-
to
-
e
n
d
lear
n
in
g
b
a
s
ed
m
o
d
el
in
o
r
d
er
to
m
a
k
e
it
f
ea
s
ib
le
to
b
e
i
m
p
le
m
en
ted
in
t
h
e
r
ea
l
w
o
r
ld
co
n
d
i
tio
n
.
I
n
th
i
s
w
o
r
k
,
in
v
est
ig
at
io
n
o
f
en
d
-
to
-
e
n
d
lear
n
i
n
g
in
3
x
3
MI
MO
co
m
m
u
n
icatio
n
s
y
s
te
m
in
s
p
atial
m
u
ltip
lex
in
g
is
d
i
s
cu
s
s
ed
w
it
h
f
air
co
m
p
ar
i
s
o
n
s
to
t
h
e
b
as
elin
e
m
et
h
o
d
s
w
h
er
e
k
n
o
w
led
g
e
o
f
C
h
a
n
n
el
State
I
n
f
o
r
m
a
tio
n
(
C
SI)
is
p
er
f
e
ctl
y
k
n
o
w
n
i
n
th
e
r
ec
eiv
er
s
id
e.
T
h
e
h
ig
h
o
r
ig
in
a
lit
y
,
w
h
ic
h
p
r
o
p
o
s
ed
a
n
e
w
m
et
h
o
d
o
r
alg
o
r
ith
m
,
t
h
e
ad
d
itio
n
al
c
h
ap
ter
af
ter
t
h
e
R
e
s
u
lt
s
s
h
o
w
t
h
at
en
d
-
to
-
e
n
d
l
ea
r
n
in
g
b
ased
d
ee
p
lear
n
in
g
MI
MO
co
m
m
u
n
icati
o
n
r
esu
lt
s
in
b
etter
p
er
f
o
r
m
a
n
c
e
co
m
p
ar
ed
to
th
e
b
as
eli
n
e
m
e
th
o
d
s
.
2.
RE
S
E
ARCH
M
E
T
H
O
D
B
asicall
y
,
t
h
e
m
o
d
el
p
r
o
p
o
s
ed
in
th
i
s
w
o
r
k
is
in
s
p
ir
ed
b
y
t
h
e
m
o
d
el
i
n
a
p
ap
er
titl
ed
Dee
p
L
ea
r
n
i
n
g
-
B
ased
MI
MO
C
o
m
m
u
n
icatio
n
s
[
7
]
.
Ho
w
ev
er
,
th
er
e
ar
e
s
o
m
e
d
i
f
f
er
en
ce
s
t
h
at
w
ill
b
e
ex
p
lain
ed
in
t
h
e
f
o
llo
w
in
g
s
ec
tio
n
.
F
u
r
th
er
m
o
r
e,
th
i
s
s
ec
tio
n
also
b
r
ief
l
y
d
es
cr
ib
es b
aselin
e
m
eth
o
d
s
u
s
ed
f
o
r
co
m
p
ar
is
o
n
w
i
th
d
ee
p
lear
n
in
g
b
ased
m
e
th
o
d
s
2
.
1
.
M
o
del A
rc
hite
ct
ure
A
r
c
h
itect
u
r
e
m
o
d
el
f
o
r
t
h
e
f
ir
s
t
an
d
t
h
e
s
ec
o
n
d
o
f
s
p
atial
m
u
ltip
le
x
i
n
g
ca
s
e
ar
e
d
ep
icted
b
y
Fi
g
u
r
e
1
an
d
Fi
g
u
r
e
2
r
esp
ec
ti
v
el
y
.
T
h
ese
p
r
o
p
o
s
ed
m
o
d
els
co
n
s
is
t
o
f
s
e
v
er
al
d
en
s
e
an
d
la
m
b
d
a
la
y
er
w
h
ic
h
r
ep
r
esen
t
en
d
-
to
-
en
d
lear
n
i
n
g
s
y
s
te
m
.
6
b
it
s
eq
u
en
ce
s
ar
e
r
ep
r
esen
ted
b
y
i
n
te
g
er
s
f
r
o
m
0
u
n
til
6
3
,
s
o
th
at
to
tal
o
f
in
p
u
t
s
eq
u
en
ce
s
ar
e
6
4
d
if
f
er
en
t
i
n
p
u
t
s
(
S
)
.
T
h
o
s
e
in
p
u
ts
ar
e
f
ir
s
t
f
ed
to
e
m
b
ed
d
in
g
la
y
er
to
cr
ea
te
v
ec
to
r
o
f
m
es
s
ag
e
i
n
d
ices.
T
h
en
,
th
e
y
ar
e
en
co
d
ed
b
y
d
en
s
e
la
y
er
i
n
tr
an
s
m
it
ter
b
lo
ck
to
f
o
r
m
p
ar
allel
tr
an
s
m
it
s
tr
ea
m
s
o
f
1
ti
m
e
s
a
m
p
les
(
X
)
w
i
th
t
h
e
te
n
s
o
r
s
h
ap
e
[
b
atch
_
s
ize,
,
2
,
1
]
w
h
er
e
t
h
e
t
h
ir
d
d
i
m
en
s
io
n
r
ep
r
esen
ts
r
ea
l
a
n
d
i
m
a
g
i
n
ar
y
p
ar
t.
T
h
is
p
ar
allel
s
tr
ea
m
s
s
h
a
p
e
is
d
o
n
e
b
y
r
es
h
ap
e
la
y
er
.
Nex
t,
t
h
ese
p
ar
alle
l
tr
an
s
m
itted
s
y
m
b
o
l
s
w
ill
b
e
f
e
d
in
to
s
ev
er
al
la
m
b
d
a
la
y
er
s
r
ep
r
esen
tin
g
c
h
an
n
el
a
n
d
n
o
is
e
ef
f
ec
t
s
i
n
w
ir
eles
s
p
r
o
p
ag
atio
n
r
esu
ltin
g
in
te
n
s
o
r
s
h
ap
e
[
b
atch
_
s
ize,
,
2
,
1
]
.
an
d
d
en
o
tes
n
u
m
b
er
o
f
r
ec
eiv
e
r
an
ten
n
a
an
d
tr
an
s
m
itter
a
n
te
n
n
a
r
esp
ec
tiv
el
y
.
E
v
en
t
u
all
y
,
t
h
e
r
ec
e
iv
er
b
lo
ck
w
h
ich
h
a
s
s
ev
er
al
d
en
s
e
la
y
er
s
w
it
h
s
o
f
t
m
ax
ac
t
iv
atio
n
f
u
n
ctio
n
at
th
e
en
d
w
ill
d
ec
o
d
e
th
e
r
ec
eiv
ed
s
ig
n
al
to
p
r
o
d
u
ce
̂
.
C
o
n
ca
ten
ate
la
y
er
s
b
o
th
in
tr
an
s
m
itter
an
d
r
ec
eiv
er
m
e
an
th
at
t
h
e
in
f
o
r
m
a
tio
n
o
f
c
h
a
n
n
el
r
ep
o
n
s
e
H
is
co
n
ca
ten
ate
d
to
th
e
o
u
tp
u
t
o
f
n
eu
r
al
n
e
t
w
o
r
k
la
y
er
in
o
r
d
er
t
o
h
elp
th
e
w
ei
g
h
t
a
n
d
b
ias
u
p
d
ate
p
r
o
ce
s
s
.
T
h
e
d
if
f
er
en
ce
b
etw
ee
n
th
e
f
i
r
s
t
an
d
th
e
s
ec
o
n
d
m
o
d
el
w
h
ic
h
o
n
l
y
u
s
e
p
er
f
ec
t
C
SI
i
n
t
h
e
r
ec
ei
v
er
s
id
e
is
j
u
s
t
t
h
e
p
o
s
itio
n
o
f
r
esh
ap
e
la
y
er
.
T
h
is
r
esh
ap
e
la
y
er
ac
t
u
all
y
h
as
a
s
i
g
n
i
f
ica
n
t
i
m
p
ac
t
to
t
h
e
p
er
f
o
r
m
an
ce
a
n
d
th
e
s
h
ap
e
o
f
co
n
s
t
ellatio
n
p
o
in
ts
o
f
t
h
e
s
y
s
te
m
.
B
y
c
h
an
g
i
n
g
p
o
s
itio
n
o
f
r
esh
ap
e
la
y
er
,
th
e
n
w
e
m
u
s
t
s
et
th
e
h
y
p
er
p
ar
a
m
eter
s
d
if
f
er
en
tl
y
to
o
b
tain
th
e
b
est r
esu
lt.
T
ab
le
1
an
d
T
ab
le
2
s
h
o
w
la
y
o
u
t o
f
Ne
u
r
al
Net
wo
r
k
u
s
ed
in
t
h
i
s
w
o
r
k
.
Fig
u
r
e
1
.
Au
to
en
co
d
er
b
ased
s
p
atial
m
u
ltip
lex
in
g
p
er
f
ec
t C
S
I
T
an
d
C
SIR
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
I
mp
leme
n
ta
tio
n
o
f D
ee
p
Lea
r
n
in
g
in
S
p
a
tia
l Mu
ltip
lexin
g
MIMO
C
o
mmu
n
ica
tio
n
(
Ma
h
d
in
R
o
h
ma
till
a
h
)
701
Fig
u
r
e
2
.
Au
to
en
co
d
er
b
ased
s
p
atial
m
u
ltip
lex
in
g
C
SIR
T
ab
le
1
.
L
ay
o
u
t o
f
p
er
f
ec
t
C
SI
T
an
d
C
SIR c
ase
T
r
a
n
sm
i
t
t
e
r
(
T
X
)
:
Pa
r
a
m
e
t
e
r
s
Ou
t
p
u
t
D
i
m
e
n
si
o
n
I
n
p
u
t
0
1
Emb
e
d
d
i
n
g
7
6
8
1
,
1
2
D
e
n
se
(
P
R
e
L
U
)
3
1
2
24
L
i
n
e
a
r
1
24
N
o
r
mal
i
z
a
t
i
o
n
0
4
R
e
c
e
i
v
e
r
(
R
X
)
:
Pa
r
a
m
e
t
e
r
s
Ou
t
p
u
t
D
i
m
e
n
si
o
n
I
n
p
u
t
0
24
D
e
n
se
(
P
R
e
L
U
)
6
4
0
0
2
5
6
D
e
n
se
(
P
R
e
L
U
)
3
2
8
9
6
1
2
8
D
e
n
se
(
S
o
f
t
max
)
8
2
5
6
64
T
ab
le
2
.
L
ay
o
u
t o
f
p
er
f
ec
t
C
SI
R
ca
s
e
T
r
a
n
sm
i
t
t
e
r
(
T
X
)
:
Pa
r
a
m
e
t
e
r
s
Ou
t
p
u
t
D
i
m
e
n
si
o
n
I
n
p
u
t
0
1
Emb
e
d
d
i
n
g
7
6
8
1
,
1
2
C
o
n
c
a
t
e
n
a
t
e
0
3
,
2
,
5
D
e
n
se
(
P
R
e
L
U
)
7
4
4
24
L
i
n
e
a
r
1
24
N
o
r
mal
i
z
a
t
i
o
n
0
4
R
e
c
e
i
v
e
r
(
R
X
)
:
Pa
r
a
m
e
t
e
r
s
Ou
t
p
u
t
D
i
m
e
n
si
o
n
I
n
p
u
t
0
24
D
e
n
se
(
P
R
e
L
U
)
6
4
0
0
2
5
6
D
e
n
se
(
P
R
e
L
U
)
3
2
8
9
6
1
2
8
D
e
n
se
(
S
o
f
t
max
)
8
2
5
6
64
C
o
m
p
ar
ed
to
th
e
p
r
ev
io
u
s
m
o
d
el,
m
o
d
els
s
h
o
w
n
b
y
Fig
u
r
e
1
an
d
Fig
u
r
e
2
alr
ea
d
y
s
h
o
w
s
s
ev
er
al
d
if
f
er
e
n
ce
s
b
es
id
e
th
e
d
ep
th
o
f
t
h
e
n
er
w
o
r
k
.
Firs
t,
b
o
th
m
o
d
el
u
s
e
C
h
an
n
el
State
I
n
f
o
r
m
atio
n
i
n
t
h
e
r
ec
eiv
er
s
id
e
s
o
t
h
at
w
e
ca
n
m
ak
e
a
f
air
co
m
p
ar
is
o
n
w
i
th
th
e
b
as
e
lin
e
m
et
h
o
d
w
h
ic
h
i
m
p
le
m
en
ts
p
r
ef
ec
t
C
SI
R
i
n
o
r
d
er
to
d
ec
o
d
e
th
e
r
ec
eiv
ed
s
ig
n
a
l.
Mo
r
eo
v
er
,
th
e
ch
an
n
el
an
d
n
o
is
e
ar
e
r
ep
r
esen
ted
as
in
p
u
t
s
o
f
th
e
m
o
d
el
u
s
i
n
g
“
r
a
n
d
n
”
f
u
n
ctio
n
f
r
o
m
Nu
m
p
y
lib
r
ar
y
r
ath
er
th
a
n
g
e
n
er
ated
b
y
s
e
v
er
al
la
m
b
d
a
lay
er
s
th
at
e
m
er
g
e
a
d
o
u
b
t
w
h
et
h
er
th
e
g
en
er
ated
c
h
an
n
el
r
esp
o
n
s
e
s
u
itab
le
to
th
e
p
r
ed
eter
m
in
ed
s
ta
n
d
ar
d
.
T
h
e
s
ec
o
n
d
,
n
o
n
li
n
ea
r
ac
tiv
atio
n
f
u
n
ctio
n
u
s
ed
is
P
R
eL
U
[
8
]
in
s
tead
o
f
R
e
L
U.
On
e
o
f
th
e
ad
v
a
n
tag
e
s
o
f
u
s
in
g
P
R
e
L
U
is
t
h
e
n
eg
at
iv
e
v
al
u
e
i
n
p
u
t
w
il
l
s
til
l
h
av
e
o
u
tp
u
t
r
at
h
e
r
th
a
n
ze
r
o
.
As
th
e
d
ata
f
lo
w
i
n
g
i
n
t
h
e
m
o
d
el
h
as
a
r
an
g
e
o
f
-
to
,
th
e
P
R
eL
U
p
r
o
p
er
ties
is
v
er
y
b
en
ef
ic
ial
f
o
r
i
m
p
r
o
v
e
th
e
m
o
d
el
ac
cu
r
ac
y
.
T
h
e
o
u
tp
u
t
o
f
P
R
eL
U
ac
tiv
atio
n
f
u
n
ctio
n
f
o
llo
w
s
t
h
e
eq
u
atio
n
(
)
{
(
1)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l
.
1
2
,
No
.
2
,
No
v
e
m
b
er
201
8
:
6
9
9
–
7
0
5
702
is
th
e
in
p
u
t
o
f
n
o
n
l
in
ea
r
ac
t
iv
atio
n
f
u
n
ctio
n
f
o
n
th
e
ch
an
n
el,
w
h
ile
is
a
co
ef
f
icien
t
ad
ap
tiv
el
y
co
n
tr
o
llin
g
th
e
s
lo
p
e
o
f
th
e
n
eg
ati
v
e
p
ar
ts
.
T
h
is
co
ef
f
icie
n
t
is
u
p
d
ated
u
s
in
g
m
o
m
en
tu
m
m
et
h
o
d
w
h
ic
h
is
g
iv
e
n
b
y
(
2
)
w
h
er
e
an
d
d
en
o
tes
th
e
m
o
m
en
tu
m
an
d
lear
n
i
n
g
r
ate
r
esp
ec
tiv
el
y
.
R
e
L
U,
ac
tiv
atio
n
p
r
o
p
o
s
ed
in
th
e
p
r
ev
io
u
s
w
o
r
k
,
h
as b
ee
n
tr
ied
to
b
e
im
p
le
m
e
n
ted
in
th
i
s
m
o
d
el.
Un
f
o
r
tu
n
atel
y
,
t
h
e
tr
ai
n
i
n
g
an
d
v
alid
atio
n
lo
s
s
b
ec
o
m
e
v
er
y
h
ig
h
d
u
e
to
ze
r
o
g
r
ad
ien
t is
s
u
e.
T
h
e
th
ir
d
o
r
th
e
last
,
in
t
h
is
wo
r
k
w
e
s
i
m
u
lated
3
x
3
MI
MO
co
m
m
u
n
icatio
n
s
y
s
te
m
,
n
o
t
2
x
2
MI
MO
co
m
m
u
n
icatio
n
s
y
s
te
m
.
T
h
e
ch
an
n
el
i
s
f
ast
R
a
y
lei
g
h
f
ad
i
n
g
w
h
ich
m
ea
n
s
th
at
th
e
f
ad
in
g
v
ar
ie
s
at
e
v
er
y
tr
an
s
m
itted
s
y
m
b
o
l
w
h
ile
n
o
is
e
is
A
d
ap
ti
v
e
W
h
ite
Ga
u
s
s
ia
n
No
is
e
(
A
W
GN)
.
2
.
2
.
T
ra
ini
ng
P
ha
s
e
I
n
p
u
t
d
ata
u
s
ed
f
o
r
tr
ain
i
n
g
an
d
test
i
n
g
(
b
its
,
ch
a
n
n
e
l
an
d
n
o
is
e)
w
er
e
r
an
d
o
m
l
y
g
e
n
er
ated
b
y
f
u
n
ctio
n
in
th
e
N
u
m
p
y
lib
r
ar
y
.
T
o
tal
am
o
u
n
t
o
f
i
n
p
u
t
d
ata
(
b
its
)
f
o
r
tr
ain
in
g
w
as
8
0
0
0
0
0
0
b
its
.
T
h
is
m
o
d
el
w
a
s
t
h
en
tr
ai
n
ed
i
n
1
0
0
ep
o
ch
s
w
it
h
b
atc
h
s
ize
eq
u
al
to
5
0
0
.
Sev
er
al
h
y
p
er
p
ar
am
eter
s
t
u
n
i
n
g
w
er
e
i
m
p
le
m
en
ted
i
n
ce
r
tain
la
y
er
s
,
f
o
r
in
s
ta
n
ce
w
e
s
et
g
a
m
m
a
co
n
s
tr
ain
t
in
b
atch
n
o
r
m
a
lizati
o
n
la
y
er
in
o
r
d
er
to
g
iv
e
p
o
w
er
co
n
s
tr
ain
t
in
th
e
tr
an
s
m
itter
s
id
e.
Mo
r
eo
v
er
,
th
is
m
o
d
el
w
as
tr
ai
n
ed
in
a
f
i
x
ed
v
alu
e
o
f
⁄
d
B
.
As
w
e
i
n
ter
p
r
et
th
i
s
m
o
d
el
a
s
an
a
u
to
en
co
d
er
b
ased
class
i
f
icatio
n
tas
k
,
a
ca
te
g
o
r
ical
cr
o
s
s
-
en
tr
o
p
y
lo
s
s
f
u
n
ct
io
n
(
)
m
a
y
b
e
an
ap
p
r
o
p
r
iate
l
o
s
s
f
u
n
ctio
n
to
b
e
u
s
ed
f
o
r
o
p
tim
izatio
n
u
s
i
n
g
g
r
ad
ien
t
d
escen
t
to
s
elec
t n
et
w
o
r
k
p
a
r
a
m
eter
s
.
C
a
teg
o
r
ical
cr
o
s
s
-
e
n
tr
o
p
y
lo
s
s
f
u
n
ctio
n
(
)
is
g
i
v
en
b
y
(
̂
)
|
|
∑
(
|
|
(
̂
)
(
)
(
̂
)
(
3
)
Usi
n
g
a
f
o
r
m
o
f
s
to
ch
as
tic
g
r
ad
ien
t
d
escen
t,
A
d
a
m
[
9
]
,
w
e
ig
h
ts
w
er
e
iter
ati
v
el
y
u
p
d
ated
b
ased
o
n
lo
s
s
g
r
ad
ien
t
u
s
i
n
g
b
ac
k
-
p
r
o
p
ag
atio
n
[
1
0
]
.
A
l
th
o
u
g
h
A
d
a
m
ca
n
w
o
r
k
ad
ap
ti
v
el
y
as
it
tak
es
b
en
e
f
its
o
f
A
d
ap
tiv
e
Gr
ad
ien
t
A
l
g
o
r
ith
m
[
1
1
]
an
d
R
o
o
t
Me
an
Sq
u
ar
e
P
r
o
p
ag
atio
n
(
R
MSP
r
o
p
)
[
1
2
]
,
w
e
s
t
ill
s
et
t
h
e
lear
n
in
g
r
ate
to
b
e
d
ec
r
ea
s
in
g
if
t
h
e
v
alid
atio
n
lo
s
s
i
s
n
o
t r
ed
u
cin
g
s
ig
n
i
f
ica
n
tl
y
.
2
.
3
.
T
esting
P
ha
s
e
Si
m
i
lar
w
it
h
t
h
e
tr
ain
i
n
g
p
h
a
s
e,
in
p
u
t
d
ata
f
o
r
te
s
ti
n
g
w
er
e
r
an
d
o
m
l
y
g
en
er
ated
u
s
in
g
f
u
n
ct
io
n
i
n
Nu
m
p
y
,
s
o
th
a
t t
h
e
y
w
er
e
d
i
f
f
er
en
t
w
it
h
d
ata
f
ed
i
n
tr
ain
i
n
g
s
ec
tio
n
.
T
h
e
to
tal
n
u
m
b
er
o
f
b
its
i
n
t
h
i
s
s
ec
tio
n
i
s
1
0
0
0
0
0
0
b
its
an
d
B
it E
r
r
o
r
R
ate
(
B
E
R
)
w
as iter
ati
v
el
y
ca
lcu
lated
in
r
an
g
e
o
f
SN
R
-
4
d
B
u
n
til 2
2
.
5
d
B
.
2
.
3
.
B
a
s
eline
M
e
t
ho
d
I
n
t
h
is
w
o
r
k
,
w
e
co
n
s
id
er
MI
MO
s
p
atial
m
u
ltip
le
x
i
n
g
s
y
s
te
m
i
n
t
w
o
d
if
f
er
en
t
ca
s
e
s
,
f
ir
s
t
is
s
y
s
te
m
u
s
i
n
g
b
o
th
C
SIT
an
d
C
SIR
a
n
d
th
e
s
ec
o
n
d
is
s
y
s
te
m
u
s
i
n
g
o
n
l
y
C
SIR.
T
h
e
co
n
f
ig
u
r
atio
n
o
f
ea
ch
s
y
s
te
m
is
d
is
cu
s
s
ed
in
th
e
f
o
llo
w
i
n
g
p
ar
ag
r
ap
h
s
.
Si
m
ilar
w
it
h
t
h
e
d
ee
p
lear
n
in
g
b
ased
m
et
h
o
d
,
th
es
e
b
aselin
e
m
et
h
o
d
s
w
er
e
also
s
i
m
u
lated
i
n
3
x
3
MI
MO
co
m
m
u
n
icatio
n
s
y
s
te
m
.
Fo
r
th
e
f
ir
s
t
s
y
s
te
m
,
w
e
co
n
s
id
er
a
lin
ea
r
p
r
e
-
eq
u
aliza
tio
n
w
h
ic
h
e
m
p
lo
y
s
p
r
e
-
eq
u
aliza
ti
o
n
o
n
th
e
tr
an
s
m
it
ter
s
id
e
as d
ep
icted
b
y
Fig
u
r
e
3
[
1
3
]
.
Fig
u
r
e
3
.
L
in
ea
r
p
r
e
-
eq
u
aliza
ti
o
n
T
h
e
p
r
ec
o
d
ed
s
y
m
b
o
l
v
ec
to
r
ca
n
b
e
r
ep
r
esen
ted
as
̃
(
4
)
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
I
mp
leme
n
ta
tio
n
o
f D
ee
p
Lea
r
n
in
g
in
S
p
a
tia
l Mu
ltip
lexin
g
MIMO
C
o
mmu
n
ica
tio
n
(
Ma
h
d
in
R
o
h
ma
till
a
h
)
703
W
h
er
e
̃
is
th
e
o
r
ig
in
al
s
y
m
b
o
l
v
ec
to
r
f
o
r
tr
an
s
m
is
s
io
n
an
d
is
a
p
r
e
-
eq
u
alize
r
w
eig
h
t
m
atr
ix
.
A
s
th
e
MM
SE
p
r
e
-
eq
u
a
lizatio
n
was u
s
ed
i
n
th
e
s
i
m
u
la
tio
n
,
t
h
e
w
ei
g
h
t
m
a
tr
ix
i
s
g
i
v
e
n
b
y
*
‖
(
̃
)
̃
‖
(
)
(
5
)
w
h
ile
is
a
co
n
s
ta
n
t to
m
ee
t t
h
e
to
tal
tr
an
s
m
itted
p
o
w
er
co
n
s
tr
ain
t a
f
ter
p
r
e
-
eq
u
aliza
tio
n
an
d
it is
g
i
v
en
as
√
(
(
)
)
(
6
)
w
h
er
e
H
an
d
d
en
o
te
a
ch
an
n
e
l r
esp
o
n
s
e
an
d
n
u
m
b
er
o
f
tr
a
n
s
m
i
tter
an
te
n
n
a
r
esp
ec
ti
v
el
y
.
Fo
r
th
e
s
ec
o
n
d
s
y
s
te
m
,
Ma
x
i
m
u
m
L
i
k
eli
h
o
o
d
(
ML
)
alg
o
r
ith
m
w
a
s
u
s
ed
to
d
etec
t
.
ML
d
etec
tio
n
ca
lcu
late
s
th
e
E
u
cl
id
ea
n
d
is
tan
ce
b
et
w
ee
n
t
h
e
r
ec
eiv
ed
s
ig
n
al
v
ec
to
r
an
d
th
e
p
r
o
d
u
ct
o
f
all
p
o
s
s
ib
le
tr
an
s
m
itted
s
i
g
n
al
v
ec
to
r
s
w
i
t
h
th
e
g
i
v
en
c
h
an
n
el
H
.
M
L
d
etec
t
io
n
d
eter
m
i
n
es t
h
e
tr
an
s
m
i
tted
s
y
m
b
o
l
x
as
̂
‖
‖
(
7
)
3.
RE
SU
L
T
S
A
ND
AN
AL
Y
SI
S
I
n
th
i
s
s
ec
tio
n
,
w
e
tr
ai
n
ed
th
e
en
d
-
to
-
en
d
lear
n
i
n
g
b
ased
MI
MO
co
m
m
u
n
icatio
n
m
o
d
el
d
escr
ib
ed
o
n
th
e
p
r
ev
io
u
s
s
ec
tio
n
w
i
th
t
h
e
h
elp
o
f
Ker
as
w
i
th
te
n
s
o
r
f
lo
w
-
g
p
u
b
ac
k
e
n
d
an
d
ev
al
u
ated
th
e
B
E
R
o
v
er
th
e
r
an
g
e
o
f
SN
R
s
.
T
h
e
r
esu
lts
a
r
e
f
air
l
y
co
m
p
ar
ed
w
it
h
b
aselin
e
m
et
h
o
d
s
s
i
m
u
lated
in
M
AT
L
A
B
w
i
th
QP
SK
m
o
d
u
latio
n
w
a
s
u
s
ed
to
m
o
d
u
late
in
p
u
t
b
its
.
B
o
th
s
y
s
te
m
s
w
er
e
s
i
m
u
lated
in
3
x
3
MI
MO
co
m
m
u
n
icatio
n
s
y
s
te
m.
3
.
1
.
Sp
a
t
ia
l
M
ultiplex
ing
P
er
f
ec
t
CSI
R
a
nd
CSI
T
Fo
r
th
e
f
ir
s
t
m
o
d
el,
w
e
s
i
m
u
la
ted
3
x
3
MI
MO
s
y
s
te
m
w
it
h
p
er
f
ec
t
C
SIT
an
d
C
SIR
s
o
t
h
at
th
er
e
is
n
o
f
ee
d
b
ac
k
f
r
o
m
r
ec
ei
v
er
to
tr
a
n
s
m
itter
.
I
n
b
aseli
n
e
m
et
h
o
d
s
,
th
e
p
o
w
er
o
f
ea
c
h
a
n
te
n
n
a
w
a
s
s
et
to
b
e
eq
u
al,
w
h
ile
i
n
a
u
to
en
co
d
er
b
ased
m
o
d
el
tr
an
s
m
i
t
p
o
w
er
o
f
e
ac
h
an
ten
n
a
i
s
d
if
f
er
en
t
d
u
e
to
d
if
f
er
e
n
t
w
ei
g
h
ts
a
n
d
b
iases
o
f
ea
ch
an
ten
n
a
as
a
r
esu
lt
o
f
tr
ain
i
n
g
s
ec
tio
n
.
Ho
w
e
v
er
,
th
e
av
er
a
g
e
en
er
g
y
is
eq
u
al
to
1
(
r
ea
ch
in
g
it
s
av
er
ag
ed
p
o
w
er
b
y
u
n
ev
e
n
p
o
w
er
d
is
tr
ib
u
tio
n
b
et
w
ee
n
ea
ch
a
n
te
n
n
a)
.
C
o
n
s
tellat
io
n
p
o
in
t
o
f
ea
c
h
au
to
en
co
d
er
b
ased
MI
MO
an
d
its
r
ec
eiv
ed
p
o
in
t
s
is
s
h
o
w
n
b
y
Fi
g
u
r
e
4
.
Me
a
n
w
h
ile,
th
e
p
er
f
o
r
m
a
n
ce
o
f
th
e
p
r
o
p
o
s
ed
s
y
s
te
m
is
d
ep
icted
b
y
F
ig
u
r
e
5
.
P
er
f
o
r
m
a
n
ce
i
s
e
v
alu
ated
i
n
ter
m
s
o
f
B
E
R
w
h
ic
h
i
s
a
n
a
v
er
ag
e
o
f
all
B
E
R
co
m
p
u
ted
b
y
ea
ch
a
n
ten
n
a.
I
t
s
ee
m
s
th
at
t
h
e
au
t
o
en
co
d
er
b
ased
m
o
d
el
o
u
tp
er
f
o
r
m
s
t
h
e
b
aselin
e
m
et
h
o
d
s
i
n
ce
t
h
e
v
al
u
e
o
f
S
NR
i
s
5
d
B
.
A
s
th
e
SN
R
g
et
h
ig
h
er
,
t
h
e
h
u
g
e
g
ap
p
er
f
o
r
m
an
ce
b
et
w
ee
n
ea
c
h
m
et
h
o
d
b
ec
o
m
e
h
ig
h
er
.
T
h
is
p
er
f
o
r
m
a
n
ce
w
a
s
ac
h
ie
v
ed
w
i
th
s
o
m
e
h
y
p
er
p
ar
a
m
eter
s
t
u
n
i
n
g
,
f
o
r
in
s
tan
ce
i
n
th
e
n
o
r
m
aliza
tio
n
la
y
er
,
w
e
s
et
th
e
m
a
x
n
o
r
m
o
f
g
a
m
m
a
c
o
n
s
tr
ain
t
to
an
ap
p
r
o
p
r
iate
v
alu
e
(
1
.
1
)
in
o
r
d
er
to
ef
f
ec
tiv
e
l
y
p
u
t a
p
o
w
er
co
n
s
tr
ain
t i
n
th
e
e
n
co
d
er
b
lo
ck
m
o
d
e
Fig
u
r
e
4
.
L
ea
r
n
ed
co
n
s
te
llatio
n
au
to
e
n
co
d
er
b
ased
MI
MO
p
er
f
ec
t CS
I
T
an
d
C
SIR
Evaluation Warning : The document was created with Spire.PDF for Python.
I
SS
N
:
2502
-
4752
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci,
Vo
l
.
1
2
,
No
.
2
,
No
v
e
m
b
er
201
8
:
6
9
9
–
7
0
5
704
Fig
u
r
e
5
.
B
E
R
co
m
p
ar
is
o
n
b
etw
ee
n
p
r
o
p
o
s
ed
MI
MO
p
e
r
f
ec
t CS
I
T
an
d
C
SIR
m
o
d
el
an
d
b
a
s
elin
e
m
et
h
o
d
3
.
2
.
Sp
a
t
ia
l
M
ultiplex
ing
P
er
f
ec
t
CSI
R
Si
m
ilar
to
th
e
p
r
ev
io
u
s
l
y
d
i
s
cu
s
s
ed
m
o
d
el,
in
th
e
p
er
f
ec
t
C
SIR
ca
s
e,
t
h
e
p
o
w
er
o
f
ea
ch
an
te
n
n
a
is
u
n
e
v
e
n
l
y
d
is
tr
ib
u
ted
,
b
u
t
s
till
ac
h
iev
e
s
i
ts
a
v
er
ag
e
p
o
w
er
tr
a
n
s
m
i
s
s
io
n
.
L
ea
r
n
ed
co
n
s
tellati
o
n
p
o
in
t
i
s
s
h
o
w
n
b
y
Fi
g
u
r
e
6
,
w
h
i
le
s
y
s
te
m
p
er
f
o
r
m
a
n
ce
e
v
alu
a
tio
n
in
ter
m
s
o
f
B
E
R
co
m
p
ar
i
s
o
n
b
et
w
ee
n
au
to
en
co
d
er
b
ased
m
et
h
o
d
an
d
b
aselin
e
m
et
h
o
d
is
s
h
o
w
n
b
y
Fi
g
u
r
e
7
.
I
n
t
h
is
c
ase,
th
e
e
n
d
-
to
-
en
d
b
ased
m
o
d
el
o
u
tp
er
f
o
r
m
s
t
h
e
b
aselin
e
m
e
th
o
d
s
in
ce
n
ea
r
l
y
2
d
B
.
T
h
is
p
er
f
o
r
m
a
n
ce
also
ac
h
iev
ed
w
it
h
s
ev
er
al
h
y
p
er
p
ar
a
m
eter
s
t
u
n
i
n
g
,
f
o
r
in
s
ta
n
ce
th
e
co
n
s
tr
ain
t
in
th
e
b
atch
n
o
r
m
a
lizatio
n
la
y
er
.
W
e
m
u
s
t
s
et
t
h
e
m
ax
n
o
r
m
o
f
g
a
m
m
a
co
n
s
tr
ai
n
t
to
b
e
0
.
8
.
W
e
also
f
o
u
n
d
th
at
th
e
i
n
cr
ea
s
e
o
f
d
ataset
n
u
m
b
er
w
il
l
n
o
t
i
m
p
r
o
v
e
th
e
p
er
f
o
r
m
a
n
ce
d
ir
ec
tl
y
.
B
atch
s
ize
an
d
co
n
s
tr
ai
n
t in
s
ev
er
al
la
y
er
s
m
u
s
t b
e
d
if
f
er
en
tl
y
s
et
to
g
et
th
e
b
est p
er
f
o
r
m
a
n
ce
.
Fig
u
r
e
6
.
L
ea
r
n
ed
co
n
s
te
llatio
n
au
to
e
n
co
d
er
b
ased
MI
MO
p
er
f
ec
t CS
I
R
Evaluation Warning : The document was created with Spire.PDF for Python.
I
n
d
o
n
esia
n
J
E
lec
E
n
g
&
C
o
m
p
Sci
I
SS
N:
2502
-
4752
I
mp
leme
n
ta
tio
n
o
f D
ee
p
Lea
r
n
in
g
in
S
p
a
tia
l Mu
ltip
lexin
g
MIMO
C
o
mmu
n
ica
tio
n
(
Ma
h
d
in
R
o
h
ma
till
a
h
)
705
Fig
u
r
e
7
.
B
E
R
co
m
p
ar
is
o
n
b
etw
ee
n
p
r
o
p
o
s
ed
MI
MO
p
e
r
f
ec
t CS
I
R
m
o
d
el
a
n
d
b
aselin
e
m
et
h
o
d
4.
CO
NCLU
SI
O
N
As a
s
o
lu
tio
n
o
f
tr
ad
e
-
o
f
f
p
h
e
n
o
m
e
n
o
n
i
n
MI
MO
co
m
m
u
n
i
ca
tio
n
o
p
ti
m
izatio
n
,
th
is
p
ap
er
p
r
o
p
o
s
es a
m
et
h
o
d
i
m
p
le
m
e
n
ti
n
g
o
n
e
o
f
t
h
e
m
o
d
el
i
n
d
ee
p
lear
n
in
g
ar
ea
,
en
d
-
to
-
e
n
d
lear
n
i
n
g
au
to
e
n
c
o
d
er
.
T
h
is
m
eth
o
d
s
h
o
w
s
p
r
o
m
i
s
i
n
g
r
es
u
lt
s
co
m
p
ar
e
to
th
e
b
aselin
e
m
et
h
o
d
s
in
ter
m
s
o
f
B
E
R
o
v
er
f
ast
R
a
y
le
ig
h
f
ad
i
n
g
c
h
an
n
el
b
y
r
ea
ch
i
n
g
m
o
r
e
th
a
n
in
ter
m
o
f
B
E
R
.
Mo
r
eo
v
er
,
b
y
u
s
in
g
d
ee
p
lear
n
in
g
b
ased
m
e
th
o
d
,
th
e
co
m
p
u
tatio
n
al
co
m
p
le
x
it
y
ca
n
b
e
r
ed
u
ce
d
b
ec
au
s
e
i
n
d
ee
p
le
ar
n
in
g
f
ield
,
co
m
p
u
tatio
n
al
co
m
p
lex
i
t
y
j
u
s
t ta
k
es
p
la
ce
in
th
e
tr
ai
n
in
g
s
ec
tio
n
.
Ho
w
e
v
er
,
th
er
e
ar
e
s
til
l
s
o
m
e
co
n
s
id
er
atio
n
s
in
o
r
d
er
to
m
a
k
e
th
e
p
r
o
p
o
s
ed
m
o
d
els
to
b
e
f
itted
w
it
h
th
e
r
ea
l
w
o
r
ld
i
m
p
air
m
e
n
t
s
.
On
e
o
f
t
h
e
m
is
b
y
d
o
i
n
g
o
n
li
n
e
lear
n
in
g
i
n
s
tead
o
f
d
o
i
n
g
o
f
f
lin
e
lear
n
i
n
g
u
s
i
n
g
s
y
n
t
h
etica
ll
y
g
en
er
a
ted
d
ata.
Mo
r
eo
v
er
,
th
e
c
h
an
n
el
esti
m
atio
n
m
o
d
el
ca
n
b
e
i
m
p
le
m
en
ted
u
s
i
n
g
d
ee
p
lear
n
in
g
b
a
s
ed
m
e
th
o
d
b
ec
a
u
s
e
s
o
m
e
ti
m
es
i
t
w
i
ll
b
e
h
ar
d
to
o
b
tain
p
er
f
ec
t
C
SI
i
n
t
h
e
r
ea
l
w
o
r
ld
co
m
m
u
n
icatio
n
.
RE
F
E
R
E
NC
E
S
[1
]
Big
li
e
ri
E,
Ca
ld
e
rb
a
n
k
R,
Co
n
sta
n
ti
n
i
d
e
s
A
,
G
o
ld
s
m
it
h
A
,
P
a
u
lraj
A
,
P
o
o
r
HV
.
M
I
M
O
w
irele
s
s
c
o
m
m
u
n
ica
ti
o
n
s.
Ca
m
b
rid
g
e
u
n
iv
e
rsity
p
re
ss
.
2
0
0
7
:
1
-
8.
[2
]
O’Sh
e
a
T
,
Ho
y
d
is
J.
A
n
in
tro
d
u
c
ti
o
n
to
d
e
e
p
lea
rn
in
g
f
o
r
th
e
p
h
y
sic
a
l
la
y
e
r.
IEE
E
T
ra
n
sa
c
ti
o
n
s
o
n
Co
g
n
i
ti
v
e
Co
m
m
u
n
ica
ti
o
n
s an
d
Ne
tw
o
rk
in
g
.
2
0
1
7
;3
(
4
):
5
6
3
-
7
5
.
[3
]
Do
rn
e
r
S
,
Ca
m
m
e
re
r
S
,
Ho
y
d
is
J,
ten
Br
in
k
S
.
On
d
e
e
p
lea
rn
in
g
-
b
a
se
d
c
o
m
m
u
n
ica
ti
o
n
o
v
e
r
t
h
e
a
ir.
S
ig
n
a
ls,
S
y
st
e
m
s,
a
n
d
Co
m
p
u
ters
5
1
st
A
si
lo
m
a
r
Co
n
f
e
re
n
c
e
.
2
0
1
7
:
1
7
9
1
-
1
7
9
5
[4
]
Ba
ld
i
P
.
A
u
to
e
n
c
o
d
e
rs,
u
n
su
p
e
rv
ise
d
lea
rn
in
g
,
a
n
d
d
e
e
p
a
rc
h
it
e
c
tu
re
s.
P
r
o
c
e
e
d
in
g
s
o
f
ICM
L
w
o
rk
sh
o
p
o
n
u
n
su
p
e
rv
ise
d
a
n
d
tran
sf
e
r
lea
rn
in
g
.
2
0
1
2
:3
7
-
4
9
.
[5
]
G
ru
b
e
r
T
,
Ca
m
m
e
re
r
S
,
Ho
y
d
is
J,
ten
Bri
n
k
S
.
O
n
d
e
e
p
lea
rn
in
g
-
b
a
se
d
c
h
a
n
n
e
l
d
e
c
o
d
in
g
.
I
n
In
f
o
r
m
a
ti
o
n
S
c
ien
c
e
s
a
n
d
S
y
ste
m
s (CIS
S
),
2
0
1
7
5
1
st
A
n
n
u
a
l
Co
n
f
e
re
n
c
e
.
2
0
1
7
:1
-
6.
[6
]
Ye
H,
L
i
G
Y,
Ju
a
n
g
BH.
P
o
w
e
r
o
f
d
e
e
p
lea
rn
in
g
f
o
r
c
h
a
n
n
e
l
e
st
im
a
ti
o
n
a
n
d
sig
n
a
l
d
e
tec
ti
o
n
in
OFDM
s
y
ste
m
s.
IEE
E
W
irele
ss
Co
m
m
u
n
ica
ti
o
n
s L
e
tt
e
rs.
2
0
1
8
;
(1
):
1
1
4
-
7.
[7
]
O'
S
h
e
a
T
J,
Erp
e
k
T
,
Clan
c
y
T
C.
De
e
p
le
a
rn
in
g
b
a
se
d
M
IM
O
c
o
m
m
u
n
ica
ti
o
n
s.
a
rX
iv
p
re
p
rin
t
a
rX
iv
:1
7
0
7
.
0
7
9
8
0
.
2
0
1
7
.
[8
]
He
K,
Zh
a
n
g
X
,
Re
n
S
,
S
u
n
J.
De
lv
in
g
d
e
e
p
i
n
to
re
c
ti
f
iers
:
S
u
rp
a
ss
in
g
h
u
m
a
n
-
lev
e
l
p
e
rf
o
r
m
a
n
c
e
o
n
im
a
g
e
n
e
t
c
las
si
f
ica
ti
o
n
.
In
P
ro
c
e
e
d
in
g
s o
f
th
e
IEE
E
in
tern
a
ti
o
n
a
l
c
o
n
f
e
re
n
c
e
o
n
c
o
m
p
u
ter v
isio
n
.
2
0
1
5
:
1
0
2
6
-
1
0
3
4
.
[9
]
Kin
g
m
a
DP
,
Ba
J.
A
d
a
m
:
A
m
e
th
o
d
f
o
r
st
o
c
h
a
stic o
p
ti
m
iza
ti
o
n
.
a
r
X
iv
p
re
p
r
in
t
a
rX
iv
:1
4
1
2
.
6
9
8
0
.
2
0
1
4
.
[1
0
]
Ru
m
e
lh
a
rt
DE,
Hin
to
n
G
E,
W
i
ll
iam
s
R
J.
Lea
rn
in
g
re
p
re
se
n
tatio
n
s
b
y
b
a
c
k
-
p
ro
p
a
g
a
ti
n
g
e
rro
rs.
n
a
tu
re
.
1
9
8
6
;
3
2
3
(6
0
8
8
):
5
3
3
.
[1
1
]
Du
c
h
i
J,
Ha
z
a
n
E,
S
in
g
e
r
Y.
A
d
a
p
ti
v
e
su
b
g
ra
d
ien
t
m
e
th
o
d
s
f
o
r
o
n
li
n
e
lea
rn
in
g
a
n
d
sto
c
h
a
stic
o
p
t
i
m
iz
a
ti
o
n
.
Jo
u
r
n
a
l
o
f
M
a
c
h
in
e
L
e
a
rn
in
g
Re
se
a
rc
h
.
2
0
1
1
;1
2
:2
1
2
1
-
5
9
.
[1
2
]
Da
u
p
h
i
n
YN
,
De
V
ries
H,
Ch
u
n
g
J,
Be
n
g
io
Y.
RM
S
P
ro
p
a
n
d
e
q
u
il
ib
ra
te
d
a
d
a
p
ti
v
e
lea
rn
in
g
ra
tes
f
o
r
n
o
n
-
c
o
n
v
e
x
o
p
ti
m
iza
ti
o
n
.
a
r
X
iv
p
re
p
rin
t
a
rX
i
v
:1
5
0
2
.
0
4
3
9
0
.
.
2
0
1
5
[1
3
]
Ch
o
YS,
Kim
J,
Ya
n
g
W
Y,
Ka
n
g
CG
.
M
IM
O
-
OFDM
w
irele
ss
c
o
m
m
u
n
ica
ti
o
n
s
w
it
h
M
ATLA
B
.
Jo
h
n
W
il
e
y
&
S
o
n
s;
2
0
1
0
:
3
2
7
-
3
3
9
,
3
8
1
-
3
8
3
.
Evaluation Warning : The document was created with Spire.PDF for Python.