TELKOM
NIKA
, Vol.12, No
.3, Septembe
r 2014, pp. 7
33~740
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v12i3.103
733
Re
cei
v
ed Ma
rch 9, 2
014;
Re
vised July
25, 2014; Accepted Augu
st
10, 2014
Diversity Combining Using Maxima Ratio Combining
for All Modulation Mode
Yusuf Kurni
a
w
a
n*
1
, An
d
y
es Fourma
n D.A. Sudirja
2
Dep
a
rtment of Electrical E
ngi
neer
ing, Sch
o
o
l
of Electrical E
ngi
neer
in
g and
Informatics
Institut T
e
khnol
ogi
Ba
nd
ung, Band
un
g,
Indo
nesi
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
:
y
u
s
u
fk@stei.i
t
b.ac.id
1
, and
yes.sudir
j
a@
ya
hoo.com
2
A
b
st
r
a
ct
The destruction caused
by channel
can be seen
by
the
existence of Amplitude and Phase Shift.
By usin
g the 6
W
a
ys Diversity
Co
mbi
n
i
ng
me
thod (6 A
n
te
n
n
a
s/Rece
ivers),
it is ex
pect
ed t
hat the d
i
srupti
o
n
cause
d
by A
m
plitu
de
and
Ph
ase Sh
ift can
be su
ppr
esse
d
as s
m
al
l as
possi
ble. In
a
dditi
on, by
usi
n
g
diversity c
o
mb
inin
g
mod
u
le,
w
e
w
ill g
e
t a
l
a
rge
SN
R
o
u
tput w
h
ich
h
a
s
a v
a
lu
e s
u
m
of SNR
of e
a
c
h
diversity path. T
he
des
ig
n
of Diversi
ty co
mb
inin
g
mo
dul
e b
egi
ns w
i
th MAT
L
AB functio
n
a
l d
e
sig
n
as
a
big
picture of
the w
hole
syste
m
. Subse
q
u
ently, it
w
ill
be made the
har
dw
ar
e b
a
sed
on th
e M
A
TLAB functio
nal.
This architectu
ral desi
gn that
w
ill be the cornerston
e in th
e MATLAB
bit precisio
n man
u
facturin
g. The
n
MAT
L
AB bit
p
r
ecisio
n w
i
l
l
b
e
d
e
sig
n
e
d
as
the fo
un
datio
n of th
e V
HDL
des
ign.
Diver
sity co
mbi
n
i
n
g
t
h
e
output mod
u
le
me
ets the st
andards sp
ecifie
d
by the DVB consorti
u
m
.
In the har
dw
are (F
PGA) test result
s
of divers
ity co
mb
ini
ng, th
e
maxi
mu
m w
o
rki
n
g frequ
enc
y
is
44.5
6
MH
z
w
h
ich
has
s
how
n that is
qu
alifi
ed
w
i
th the stan
da
rd sa
mp
li
ng c
l
o
ck (9.14
2
MH
z). T
h
is des
ig
n
also
ne
eds
4%
of total F
P
GA
Cyclon
e II 4
84I
8
combi
natio
na
l units w
h
ich is
249
9 units a
n
d
it needs al
s
o
3% of total reg
i
ster of F
P
GA
Cyclon
e II 484
I8
w
h
ich is 172
0 register u
n
its.
Ke
y
w
ords
: MRC, selectiv
e combi
n
in
g, diver
s
ity comb
in
ing,
OFDM
1. Introduc
tion
Rapi
d growth
deman
d of
so
ciety to the ele
c
troni
c
device
gives high luxu
ry
cau
s
e
d
improvem
ent
in device technolo
g
y. One
of its impa
ct
s is
Digital
T
e
levisio
n
(DT
V
).There a
r
e
so
many types of DTV
s
u
c
h
as
DVB-H,
DV
B-S and so
fo
rth, yet one
o
f
the mo
st su
ccessful
DTV
is
DVB-T (Digit
al Video Broad
ca
sting T
e
rrestri
a
l)
. DVB-T is a standard of terrestri
a
l digi
tal
television
whi
c
h is
static an
d its re
ceiver
and tran
smitt
e
r are stea
dy or not moving
.
DVB-T
uses OF
DM (Ort
hogo
nal F
r
e
quen
cy
Division Multipl
e
xing) te
ch
niqu
e. Thi
s
modulatio
n t
e
ch
niqu
e u
s
es
multi
sub
-
ca
rri
er whic
h is o
r
thog
o
nal b
e
twe
e
n
ea
ch
othe
r. Thi
s
OFDM meth
o
d
allows ea
ch sub
c
a
rrie
r
to be ov
erlap
ped to each other with
out
any interfere
n
ce
both Inter Symbol Interference (ISI) an
d
ICI (Inter Carrier Interference)[1].
DVB-T utilize
s
an equ
alization modul
e
that
is located betwe
en the input of Cha
nnel
De
cod
e
r a
n
d
the output o
f
FFT. The Equali
z
ation
m
odule i
s
u
s
ed
to comp
en
sate the ch
an
nel
noise from b
o
th noise
cau
s
ed by amplit
ude and p
h
a
s
e noi
se. The
module that comm
only used
to comp
en
sat
e
these n
o
ise
s
is
Diversity Combi
n
ing m
odule. In the
comm
on u
s
e
there i
s
only 2
ways dive
rsit
y but in fact, it is not enou
gh to
redu
ce
the noise bot
h AWG
N
and
Rayleigh Noi
s
e
and
6
ways
Diversity (Re
c
e
i
ved Anten
n
a
)
i
s
the
hig
h
e
s
t nu
mbe
r
of
Antenna
so t
he
re
ceived
d
a
ta
are still correl
ated between
one and a
not
her.
2. Div
e
rsit
y
Co
mbining
This
cha
p
ter
discu
s
ses a
b
out gene
ral
a
l
gorithm of
Di
versity Co
mbi
n
ing.
Thi
s
mo
dule is
use
d
to comb
ine (o
r sel
e
cti
ng) the SNR
of re
ceive
r
s in orde
r to get
the better SNR o
u
tput.
2.1
Maxima Ra
tio Combining
In Maxima Ratio Combi
n
i
ng methods, t
he data output of each
F
FT will be co-phased
due to the influence of phase shift
Θ
beca
u
se of ch
annel
s that is multiplied b
y
the weighting
∝
i
=a
i
.e
-j
θ
with
amplitude a
i
.
Without co-p
hasi
ng, ea
ch
bran
ch i
s
not
summe
d coh
e
rently that st
ill
contai
ns d
a
m
aged d
a
ta du
e to fading. As explain
ed in
[2].
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 3, September 20
14: 73
3 – 740
734
Figure 1. Linier
Com
b
iner Dive
rsity Combi
n
ing
In Figure 1 T
he tran
smitte
d signal
s(t)
whi
c
h ha
s interferen
ce
s from Cha
nnel R
i
e
j
Θ
i
will
be multiplied
with co
mplex
amplitude
∝
i
= a
i
e
-j
θ
So will be get the com
b
ined SNR
With r = e
-j
θ
.
This eq
uation
can be
simpli
fied usin
g Ca
uchy Sch
w
a
r
t
z
Method.
We get that
γΣ
will be
γΣ
∑
∑
(1)
̅
With
̅
is the
mean S
N
R from e
a
ch receiver. From a
bove eq
uatio
n, we
ca
n ge
t the
SNR gai
n of combi
ned S
N
R com
p
a
r
ed
with mean S
NR fro
m
ea
ch receiver i
s
M value. As has
been expl
ain
ed in [2] to get BER from common M
R
C will be used
an equ
ation
2
(2)
In
this case we will assume
t
hat used
channel will
be i.i.d
Rayleigh Fading P
a
ths. S
N
R per bit
per path
γ
i
ha
s an expo
nen
tial PDS with mean SNR p
e
r bit
̅
.
/
(3)
And for SNR
per bit for co
mbined S
NR
has PDF.
!
/
(4)
In the end, we can get the error
probabil
i
ty for BPSK
mode will be
∑
1
(5)
Equ. 1 Error Probability (Pb)
Equation in BPSK mode. With
(6)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Diversity Co
m
b
ining Usi
n
g Maxim
a
Ra
tio Com
b
ining
for All Modulation Mode
(Yusuf Kurni
a
wan
)
735
In the bel
ow
grap
h, BER
grap
h
whi
c
h i
s
cre
a
ted from created DVB s
y
s
t
em to tes
t
the
MRC mod
u
le
is represent
ed with
col
o
red graph
and
the BER g
r
a
ph come
s fro
m
theory
(Pb
)
is
rep
r
e
s
ente
d
with “Bla
ck” g
r
aph.
In Figure 2, it
can
be o
b
se
rved that the
MRC BER g
r
aph from cre
a
ted DVB
system ha
s
a very
simila
r BER g
r
ap
h
with BER fro
m
theo
ry. So
that the
cre
a
ted
DVB System is valid to
test
the MRC mo
deled
and it
can
be u
s
e
d
to test othe
r
method
s too
(Sele
c
tive Co
mbining, Eq
u
a
l
Gain Combini
ng, etc).
Figure 2. MRC BER, from desi
gne
d system (col
or one) VS theo
ry (Pb) with L
is numbe
r of
diversity path
2.2 Normalize
d
MRC
In the pape
r [3], it adopts a metho
d
of MRC i
n
gene
ral b
u
t with addi
tion of
comp
en
satio
n
of am
plitu
de noi
se
ch
annel. It ex
plain
s
a
me
thod
which
descri
bed
th
a
t
conve
n
tional
MRC fou
nd t
hat go
od fo
r
pro
c
e
ssi
ng i
n
QAM m
odul
a
t
ion both
QA
M 16
an
d QA
M64
mode. In equ
ation co
nvent
ional
MRC ou
tput of MRC i
s
(7)
With W is the
weightin
g fro
m
the estimat
ed ch
ann
el wi
th the below
equatio
n
W =
,
,
,
(8)
And Y is the Fast Fou
r
ie
r Tran
sfo
r
m (F
FT) outp
u
t with
k = 0,
Λ
, N-1
(9)
While at MRC algo
rithm, it has a different we
ightin
g
with Weighti
ng in the con
v
entional MRC
with
,
(10
)
It
can be ob
served
th
at
in conve
n
tional
MRC,
there i
s
un
comp
en
sa
ted amplitu
d
e
|H|
2
that
sho
w
n
in belo
w
equ
ation
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 3, September 20
14: 73
3 – 740
736
(11
)
Y =
H X
With
X
is tran
smitted data
Z(k) =
H
H
H X
(12
)
Z(k) = |H|
2
X
It can
be
observed that the output of
the conventional MRC
still l
eaves |H|
2.
Us
ing MRC
method that
propo
se
d by this paper [3], the value of |H|
2
is eliminate
d
or redu
ce
d
by
norm
a
lization
using
which
is equal with
|H|
2
. With the
same equati
on as the MRC, it can
be derived th
at the total
value of SNR
on this pape
r has a simila
r combi
ned S
NR with the
conve
n
tional MRC.
2.3
Selectiv
e Combining (SC)
In SC-type systems [2], it pro
c
e
s
ses
on
ly one
of the diversity bran
che
s
. Spe
c
ifically, in
its conventio
nal form, th
e
SC combin
er ch
ooses
the
bra
n
ch with
the high
est S
NR. In
additi
on,
sin
c
e the out
put of the SC com
b
ine
r
i
s
equ
al
to the sign
al on
only one of the high
est SNR
bran
ch, the sum of the individual bra
n
ch
signal
s is n
o
t requi
red li
ke
in MRC a
nd
EGC.
Figure 3. Selective Co
mbining
Diag
ram
For M bran
ch
diversity, the CDF of com
b
ining SNR
γΣ
is given by
P
γΣ
= p(
γΣ
<
γ
) =
p(m
a
x
[
γ
1,
γ
2,…,
γ
M]<
γ
)
(13
)
We obtai
n the pdf of
γΣ
by differentiatin
g
P
γΣ
(
γ
) rela
tiv
e
to
γ
and the outage p
r
obability
by evaluating
P
γΣ
(
γ
) at
γ
=
γ
0. Assu
m
e
that
we h
a
v
e M b
r
an
ch
es
with u
n
correlated
Rayle
i
gh
fading amplit
ude
s
r
i
. The instantane
o
u
s SNR on the
i
th
branch
is therefore
given by
γ
i
=
.
Definin
g
the
averag
e SNR
on the
i
th
branch a
s
γ
i
the SNR distribution will be
expone
ntial:
p(
γ
i
) =
/
(14
)
The outage probability for a target
γ
0 on
the
i
th
branch in Rayleigh
fading is
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Diversity Co
m
b
ining Usi
n
g Maxim
a
Ra
tio Com
b
ining
for All Modulation Mode
(Yusuf Kurni
a
wan
)
737
Pout (
γ
0
) =
1-
/
(15
)
The outage probability of the se
l
e
ction combiner for target
γ
0 is th
en
Pout(
γ
0
) =
∏
γ
i
γ
0
∏
1
/
(16
)
Relative to
γ
0
yields the pd
f for
γΣ
:
P
γΣ
=
[1-
/
]
M-
1
/
(17
)
We see that the avera
ge S
NR of the
co
mbine
r
in i.i.d Rayleigh fadi
ng is
=
γ
γdγ
=
1
/
γ
=
̅
∑
(18
)
Thus, the av
erag
e SNR g
a
in increa
se
s with
M but not linearly.
The high
er a
m
ount of
diversity the highe
r SNR o
u
tput we get [5].
2.4
Equal Gain Combining
A simpler te
chniqu
e like M
RC i
s
equal
-gain co
mbini
ng whi
c
h co-pha
se
s the si
gnal
s on
each branch
and then
co
mbine
s
t
hem
with equ
al weighting
α
i
=
e
-
θ
i
. The SNR of the combi
ned
output, assu
ming equ
al n
o
ise PSD
N0
in each bra
n
ch is given by [6]
∑
(19
)
3.
Proposed
Div
e
rsit
y
Combining Meth
od
To obtain the
best BER ou
tput, we can
n
o
t just use o
n
e
method of Diversity Co
mbining.
In QPSK and BPSK modul
ation method,
using M
RC
method i
s
the best m
e
thod com
pared
wi
th
other
metho
d
s
. It is
due
to the M
R
C
can
provide t
he g
r
eate
s
t
value of th
e
combi
ned
S
NR
comp
ared to others. While
for QAM 16 and QAM
64
modulatio
n, using the M
R
C method i
s
no
longe
r effecti
v
e beca
u
se it offers SNR gain
with
out any improvem
ent in
amplitude shift
comp
en
satio
n
. For the Q
A
M method,
bothQAM16
and QAM6
4
,
the best m
e
thod is to
use
Selective Co
mbining m
e
th
od. This i
s
b
e
ca
use this
method h
a
s t
he large
s
t instantaneo
us S
N
R
and it
can
repair p
h
a
s
e
shift (su
c
h
a
s
M
R
C) an
d
amplitu
de
shift that a
r
e
not on
the
MRC
method.
By obse
r
ving
these imp
r
o
v
ements, thi
s
p
ape
r p
r
ov
ides
a met
h
od of
weig
hting that
combi
n
e
s
two or mo
re m
e
thod
s of
Di
versity Comb
in
ing. In thi
s
case, it will
take 3
weighting
methods by
combining the
MRC method that
has the
best BER for
BPSK/QPSK modulation
and
Selective Co
mbining fo
r
QAM modul
a
t
ion. The third method i
s
Normali
z
e M
RC
whi
c
h h
a
s
a
good BER val
ue at any modulation mo
d
e
even it is
not the best in every modul
a
t
ion method [7].
The
weig
htin
g is
ba
sed
o
n
the valu
e o
f
SNR fo
r sp
ecific BE
R to
each meth
o
d
[8].For
example i
n
QPSK modul
ation mode i
n
BER
2x10
-4
, It has
been s
i
mulated for BER graph in
Figure 5.
Usi
ng
MRC
method,
we
g
e
t SNR 7
d
B f
o
r BE
R 2x1
0
-4
. For S
e
lective Co
mbining Method
[9], It can be see
n
that req
u
ired S
NR i
s
11 dB and for NMRC, it needs 2
6
dB SNR.
7dB = 2.238
7
Whe
r
ea
s for
SC method (11dB) it is si
mila
r with 3.5
48 linea
r SNR and requi
red linea
r
SNR for Normalize
d
M
RC (2
6dB) a
s
m
u
ch
a
s
19.95
3. By doin
g
weighting, f
r
om
the BE
R g
r
a
ph,
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 3, September 20
14: 73
3 – 740
738
it is obtained that the probabilit
y of using the M
R
C output is
cl
oser to the transmitted data
comp
ared wit
h
SC it is because MR
C BER <
S
C BER <
N
ormaliz
ed MRC BER
1
(20
)
1
2.2387
2.2387
3
.
548
1
9
.
953
0
.913
And weig
htin
g for No
rmali
z
e MRC
1
(22
)
1
19
.953
2.
2387
3
.548
1
9
.
953
0
.2248
With the
re
su
lts of thi
s
wei
ghting, the
val
ue of th
e de
modulato
r
of each meth
od
is
use
d
to get a better approximati
on instea
d of only usi
ng o
ne spe
c
ific di
versity com
b
ining metho
d
. If
the de
mod
u
la
tor o
u
tput of
use
d
two m
e
thod
s h
a
s a
same val
ue th
en the
o
u
tput
of d
e
mod
u
la
tor
in Figure 4 is equ
al to this value. B
y
using
this
method, it is expected to
get the hig
her
probability of getting the same out
put as
the transmitted data.
In the other hand, because in QPSK modulat
ion it doesn’t have amplitude noi
se then it
not need
s a
n
amplitude
compen
sato
r
as in QA
M
1
6
and QAM6
4
.In QAM modulation, with
out
amplitude
co
mpen
sato
r, e
a
ch
point in t
he con
s
tellati
on will
be ov
erlap
p
ing. T
h
erefo
r
e o
ppo
site
than Figu
re 5
,
instead of h
a
ving the
be
st SNR, MRC
has the
worst
SNR.
Figure 5
sho
w
s that p
r
o
c
essed
data
u
s
ing
only
1
method i
s
no
t enou
gh to
handl
e all
modulatio
n mode
s. For the further simulation,
Compa
r
i
s
on
of constell
ation betwee
n
the
prop
osed me
thod and an
o
t
her pap
er
sh
own that con
s
tellation u
s
in
g prop
osed
method is b
e
tter
than usi
ng pa
per in [3].
Figure 4. The 6 way Di
versity Com
b
i
n
ing with
Wei
ghting Probability Combini
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Diversity Co
m
b
ining Usi
n
g Maxim
a
Ra
tio Com
b
ining
for All Modulation Mode
(Yusuf Kurni
a
wan
)
739
Figure 5. BER Graph of MRC, SC and NM
R for
QPSK modul
ation
4. FPGA
Simulation
Becau
s
e
thi
s
new meth
od
use
s
3 m
e
tho
d
s
at o
n
ce,
th
en it n
eed
s to
be
sm
art to
remove
the do
ubled
module
which
the mo
dule
that can
be
sh
ared
for 2 m
o
dule
s
o
r
m
o
re. Figu
re
8 gi
ves
the illustration for the shared modules. Each
method (MRC, NM
RC and
SC)
[10] use phase
comp
en
sato
r. Therefo
r
e, the re
sulted v
a
lue of
pha
se comp
en
sat
o
r ca
n be distributed to ea
ch
method to be
pro
c
e
s
sed furthermo
re in e
a
ch p
a
rt se
pa
rately.
Figure 6. Signal Tap
si
mulation
Figure 6
sho
w
s a
sim
u
lati
on u
s
in
g
CY
CLO
N
E II FP
GA u
s
ing
Sig
nal T
ap
software
from
QUAT
RUS h
a
ve had th
e
same
output
with the o
u
tp
ut of MATLAB. It can be
sho
w
n th
at the
simila
rities b
e
t
ween th
e out
put of MATL
AB, MODEL
S
IM and Simulation in h
a
rdwa
re (FPGA
)
so
that it can be analyzed that the archite
c
t
u
re wh
ich ha
s bee
n create
d
is co
rre
ct a
nd have a go
od
desi
gn. Be
sid
e
s the
de
sign
ed a
r
chitectu
re fulfilled
the
expectatio
n
, it is be
ca
use there
is
no jitte
r
or delay com
b
ination
a
l affecting in the
data pro
c
e
s
sing in 1 cl
o
ck. While the
simulation a
l
so
sho
w
s that hard
w
a
r
e able
to calculate
in t
he high speed mo
de
whi
c
h ha
s 4 times faste
r
(81
MHz) samplin
g rate than th
e stand
ard
sp
eed in ETSI (9.142 MHz).
5.
FPGA Te
st S
cenario
To comp
rehe
nd the true
condition
as i
n
nature,
it ne
eds to
de
sign
the testing
scen
ario
due to the ro
bustn
ess of o
u
r mod
u
le in the real te
st field.
7dB
11
d
B
26
dB
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 12, No. 3, September 20
14: 73
3 – 740
740
Figure 7. FPGA Test Scen
ario
To obtai
n th
e de
sired
si
mulation
re
sults it is
ne
cessary th
at p
r
ope
r te
sting
system i
s
desi
gne
d (Fi
gure
7
)
so t
hat the
outp
u
t of the
FPGA can
rep
r
ese
n
t the
ou
tput of a
Div
e
rsity
Combi
n
ing [1
1] module in the real
DVB system.
At first, VHDL co
de
will b
e
do
wnloa
d
e
d
to the Cy
cl
one II FPGA
module f
r
om
Personal
Comp
uter (P
C). Besi
des
VHDL T
op of Diversity Co
mbining mo
d
u
le, It is download
ed also
2
ROM of data
that contain d
a
ta from the out
put of the FFT and o
u
tp
ut Chan
nel Resp
on
se.
After that the data of FFT
o
u
tput and
Ch
annel
Re
spo
n
se
will be i
n
clud
ed in the
input of
Digital to An
a
l
og Conve
r
ter (DA
C
) th
at a
m
ende
d the
origin
al digita
l sign
al into
a
n
anal
og
sign
al.
FPGA output
will be fed b
a
ck into the
Signal Tap
i
s
one of the fe
ature
s
of the
softwa
r
e Q
u
a
r
tus
to get
results
back f
r
om
FP
GA
output loop. The data
from thes
e
T
ap signal will be
dumped and
the outcom
e
woul
d be che
c
ked in a
c
cordan
ce with th
e desi
r
ed
con
s
tellation o
r
n
o
t.
6. Conclu
sion
This p
ape
r shows that we
need n
o
t to use o
n
ly one
method of di
versity co
mbi
n
ing for
spe
c
ific mo
d
u
lation mod
e
. It gives an explanation
th
at we need to con
s
ide
r
the output of other
method
s eve
n
they don’t have the
best BER
o
u
tput (su
c
h
as Sele
ctive Combini
ng and
Normali
z
ed
MRC in BPSK/QPSK modulation).T
he
proposed method
gives
a breakthrough in
Diversity Co
mbining I
s
su
e. It solved a probl
em of
M
RC that o
n
ly gives go
od p
e
rform
a
n
c
e o
n
ly in
modulation BPSK and QPSK.
Referen
ces
[1]
Unio
n EB. ET
SI EN 300 74
4 V1.6.1.
Europ
e
a
n
Standar
d.
20
09.
[2] Proakis
JG.
Digital C
o
mmun
i
c
ation ov
er F
adin
g
Cha
n
n
e
ls
.
Wille
y
.
[3]
Ning M. A Simple and Effective Post
-F
FT
Beam formin
g
T
e
chn
i
qu
e. Chi
n
a.
IEEE.
2007.
[4]
El-Dea
n SG. Performanc
e
Of S
w
itche
d
D
i
versit
y
W
i
th P
o
st- Exam
ini
n
g
Selecti
on In
Cdma S
y
stem.
Ubiq
uito
us Co
mp
utin
g an
d C
o
mmunic
a
tio
n
Journ
a
l
. 20
10.
[5]
Lio
u
mpas AS. Adaptiv
e Gene
ralize
d
Se
l
e
ctio
n Comb
ini
ng (
A
-GSC) Recei
v
ers.
IEEE.
2008.
[6] Goldsmith
A.
Wireless Co
mmu
n
ic
ations
. C
a
mbrid
ge U
n
iv
ersit
y
Press. 2
005.
[7]
Kee Bo
ng So
ng. A Lo
w
C
o
mpl
e
xit
y
S
p
a
c
e-
Frequenc
y BICM MIMO-OFDM Sy
stem for nex
t
-
Generati
on W
L
ANs.
IEEE.
2003.
[8]
El-Dea
n S. Per
f
ormance of S
w
itc
h
e
d
Divers
i
t
y
w
i
th P
o
st Examini
ng s
e
lecti
on in
CDMA s
ystem.
IEEE.
200
3.
[9]
Martin V. C
l
ar
k. Adaptiv
e F
r
equ
enc
y-D
o
ma
in Eq
ua
lizati
o
n
an
d Div
ersit
y
Comb
ini
n
g
fo
r Broad
ba
nd
W
i
reless Com
m
unic
a
tions.
IEEE,
2002; 20(
2).
[10]
Shifen Ou. Improve
d
a Pr
iori SN
R Estima
tion for Sp
eech E
nha
nce
m
ent Incorp
or
ating S
peec
h
Distortio
n C
o
mpon
ent.
T
E
L
K
OMNIKA Ind
ones
ian
Jo
urn
a
l
of Electric
al
Eng
i
ne.
20
13
; 11(9):
53
59-
536
4.
[11]
Jian
bin
Xue.
T
r
ansmissi
on P
e
rfo
rmanc
e R
e
search
of Di
git
a
l Mo
du
latio
n
Sign
als
in AW
GN Cha
n
n
e
l.
T
E
LKOMNIKA Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
ng.
2013; 1
1
(2): 9
91-9
97.
Evaluation Warning : The document was created with Spire.PDF for Python.