Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
an
d
Comp
ut
er
Scie
nce
Vo
l.
1
3
,
No.
2
,
Febr
uar
y
201
9
, pp.
447
~
452
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v1
3
.i
2
.pp
44
7
-
452
447
Journ
al h
om
e
page
:
http:
//
ia
es
core.c
om/j
ourn
als/i
ndex.
ph
p/ij
eecs
Converg
ence
analysis
of t
he unp
unctured
turbo
t
rellis
-
cod
ed
modu
lati
on (UTT
CM)
Ah
med
H
aff
ane,
Ab
delh
af
i
d H
as
ni
Must
ap
h
a Khel
ifi,
Bouf
el
d
j
a
K
adri
La
bora
tor
y
of
In
form
at
ion
Proc
e
ss
ing
and
T
elec
o
m
m
unic
at
ions (
LT
IT)
,
Ta
hri
Mo
hamm
ed
Univer
sit
y
of
B
ec
h
ar,
Alge
ria
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
S
ep
9
, 2
018
Re
vised
N
ov
2
1
, 2
018
Accepte
d
Dec
1
5,
2018
In
thi
s
pape
r
,
t
he
per
form
ance
of
the
Unpunc
ture
d
Turbo
Tr
el
li
s
-
Code
d
Modulat
ion
(U
TT
CM)
over
Additi
ve
W
hite
Gauss
ia
n
Noise
(AW
GN)
cha
nne
l
is
a
naly
z
ed
using
the
non
-
bina
r
y
extrinsic
information
tra
nsfer
(EXIT
)
cha
rt
.
T
he
exc
hang
e
of
the
ext
rinsic
info
rm
at
ion
bet
wee
n
the
dec
ode
r
components
is
tr
ac
ked
,
a
ll
owing
the
gen
era
t
ion
o
f
an
EXIT
ch
art,
which
is
a
powerful
tool
for
ana
l
y
z
ing
the
conve
rge
n
ce
be
havi
or
of
it
er
at
i
ve
dec
odin
g
and
pre
di
ct
ion
o
f
conve
rge
n
ce
p
ositi
on.
The
Sim
ula
ti
on
result
s
ar
e
compar
e
d
with
th
e
turbo
c
l
iff
posit
ions o
n
t
he
BER
cur
v
es.
Ke
yw
or
d
s
:
C
hannel c
od
i
ng
Conver
ge
nce a
naly
sis
EXIT c
har
ts
UTTCM
Copyright
©
201
8
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
:
Ah
m
ed
Ha
ff
a
ne
,
Lab
or
at
ory
of
I
nfor
m
at
ion
P
rocessi
n
g an
d Te
le
com
m
un
ic
at
i
on
s
(LTIT
)
,
Tahr
i M
oham
m
ed
U
niv
e
rsity
o
f
Becha
r,
A
lgeria
.
Em
a
il
:
haf
fanh
m
ed@
gm
ai
l.com
1.
INTROD
U
CTION
Turb
o
co
de
as
pr
ese
nted
by
B
errou
a
nd
al
[
1]
was
the
fir
st
error
c
orrecti
ng
co
de
to
cl
os
e
ly
app
r
oac
h
the
cha
nnel
c
apacit
y.
H
ow
e
ver,
li
ke
oth
e
r
Er
r
or
c
orrec
ti
ng
c
od
e
s
the
y
i
m
pr
ove
the
reli
abili
ty
of
data
transm
issi
on
at
the
ex
pen
s
e
of
band
width
e
xpan
sio
n.
T
he
r
ed
un
dan
cy
inc
reases
the
nu
m
ber
of
bits
within
th
e
transm
itted
sym
bo
l,
wh
ic
h
c
an
le
ad
to
a
red
uctio
n
of
inf
orm
ation
trans
m
issi
on
rate
or
to
an
increase
of
the
band
width.
T
r
el
li
s
cod
e
d
m
od
ulati
on
(TCM)
[
2]
is
a
ba
ndwidth
ef
fici
ent
co
ding
te
c
hn
i
qu
e
s
that
i
nteg
r
at
es
a
conv
olu
ti
onal
cod
e
w
it
h a
ba
ndwidt
h
-
e
ff
ic
i
ent m
od
ulati
on
sch
em
e.
Rob
e
rtson and
Wörz in 199
5
[
3]
m
erg
e TCM
an
d
T
urb
o
co
de
s tog
et
he
r
in
order
t
o
be
nef
i
t fr
om
their
bo
t
h
i
m
po
rta
nt
structu
res
pro
per
ti
es
an
d
ad
van
ta
ges,
a
nd
pro
po
se
d
the
“
Turb
o
Trell
is
C
od
e
d
Mo
du
la
ti
on”
(TTCM
).
The
encode
r
str
uc
ture
of
th
e
T
TCM
is
m
ade
up
of
tw
o
id
entic
al
TCM
encode
rs
in
a
par
al
le
l
con
cat
e
natio
n
li
nk
ed
by
a
sy
m
bo
l
interl
eaver.
T
he
c
od
e
d
sym
bo
ls
of
eac
h
co
m
po
nen
t
enc
oder
a
re
al
te
rn
at
ively
punct
ur
e
d
i
n
or
der
t
o
inc
reas
e
the
band
widt
h
ef
fici
ency.
The
dec
od
i
ng
of
TTCM
is
done
it
erati
vely
us
ing
a
sy
m
bo
l
-
by
-
sy
m
bo
l
Ma
xim
um
A
Po
ste
rio
ri
(MAP)
com
pone
nt
dec
od
e
rs.
De
rive
d
f
rom
the
TTCM
bu
t
wit
hout
pu
nctuati
on,
the
U
npun
ct
ur
e
d
Tu
rbo
Trell
is
Cod
e
d
Modula
ti
on
(UTTC
M)
[
4],
[5]
is
a
m
or
e
recent c
oded
m
od
ulati
on tech
nique,
which t
ran
sm
it
s tog
et
he
r: the
in
pu
t i
nfor
m
at
ion
bits an
d
the
2 par
it
y
check bit
s
fro
m
the two com
pone
nt code
rs
i
n
the
sam
e codewor
d.
Extrinsi
c
I
nfo
r
m
at
ion
Tra
nsf
er
(
EX
IT)
c
ha
rt
[
6]
is
powe
rful
to
ol
f
or
a
naly
zi
ng
the
c
onv
e
rg
e
nce
pro
per
ti
es
of
it
erati
ve
decodin
g.
E
XI
T
c
har
t
was
or
i
gin
al
ly
dev
el
op
ed
f
or
c
on
cat
enated
bin
ary
cod
i
ng
schem
es
and
exp
a
nded
la
te
r
to
non
-
bi
nar
y
cod
i
ng
sc
hem
es
[7
]
,
[8
]
.
Ba
se
d
on
the
appr
oa
ch
pr
e
sente
d
in
[8
]
,
the
pur
po
s
e
of
this
pap
e
r
is
to
ap
ply
the
EXI
T
c
har
t
to
a
naly
ze
the
perf
or
m
ance
of
the
UTTCM.
The
rest
of
the
pa
pe
r
is
orga
nized
as
fol
lows
:
Sect
io
n
II
de
scri
bes
t
he
UTTCM,
S
ect
ion
III
pr
es
ents
the
no
n
-
bi
nar
y
extrinsic
i
nform
at
ion
trans
f
er
c
har
act
erist
ic
s.
Sim
ulati
o
n
re
su
lt
s
a
nd
inter
pr
et
at
io
ns
a
re
pr
e
sent
ed
i
n
Sect
ion
I
V.
T
he
conclusi
on is re
ported
i
n
Se
ct
ion
V.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
2
,
Fe
bru
ary
201
9
:
447
–
452
448
2.
UN
P
U
NC
T
U
RED
T
RELL
I
S
-
C
ODE
D M
ODULATIO
N
The
UTTCM
t
urb
o
enc
oder
a
s
sho
wn
i
n
Fi
gure
1
,
as
pr
es
e
nted
i
n
[
4],
is
a
paral
le
l
con
c
at
enati
on
of
tow
m
/
m+1
r
at
e
RSC
(Recu
rsive
Syst
em
atic
Con
voluti
on)
enc
od
e
rs
se
pa
rated
by
a
n
in
te
rleaver
wh
ic
h
can
be
an
S
-
ra
ndom
interl
eaver
or
a
ps
eu
dora
nd
om
interl
eaver
[9
]
.
Eac
h
RSC
produces
one
par
it
y
bit
fo
r
e
ach
m
syst
e
m
atic
bits.
The
m
app
er
ta
kes
the
m
s
yst
e
m
atic
bits,
on
e
par
it
y
bi
t
fr
om
the
upper
e
nc
od
e
r
a
nd
t
he
deinterlea
ved pa
rity
b
it
f
r
om
t
he bo
tt
om
en
c
od
e
r
a
nd m
aps
them
to
2
m
+2
co
ns
te
ll
at
ion
po
i
nts.
Tw
o
m
app
in
g
te
chn
iq
ues
w
er
e
pr
op
os
e
d.
Wh
en
the
fi
rst
m
app
i
ng
te
ch
nique
ad
op
ts
G
r
ay
m
app
in
g,
the seco
nd
c
om
bin
es b
oth
n
a
tural m
app
ing
and
Gra
y
m
app
in
g,
it
is cal
led
U
ng
er
boeck
-
Gra
y
m
app
in
g [5
]
and
it
sh
ow
good
pe
rfor
m
ances al
so
wh
e
n use
d
i
n
TCM
,
as s
ho
wn in Fi
gure
2
[10].
Figure
1.
Str
uc
ture
of the
unpunct
ur
e
d
t
urbo
trel
li
s
-
cod
e
d
m
odulati
on e
ncoder [
4]
Figure
2
.
Dec
odin
g
sc
hem
e o
f
the
unpunct
ured tur
bo trell
is
-
co
de
d
m
od
ula
ti
on
[4
]
The
str
uctu
re
of
UTTCM
re
cei
ver
is
sim
ilar
to
t
hat
of
T
TCM
;
bo
th
are
serial
co
ncate
nation
of
tw
o
const
it
uen
t
sy
m
bo
l
-
by
-
sym
bo
l
MAP
dec
oders
[
4]
Th
e
obj
ect
ive
of
t
he
decode
r
is
to
determ
ine
the
A
Po
ste
ri
or
i
P
r
obabili
ti
es
(A
PP
)
of
the
tra
ns
m
i
tt
ed
sy
m
bo
ls
o
ver
a
n
it
erati
ve
dec
od
i
ng
pro
cess.
T
he
c
ombin
e
d
extrinsic
an
d
s
yst
e
m
atic
info
rm
ation
(L
e&s
)
is
extracte
d
fr
om
the
ou
tput
of
on
e
decod
er
and
passe
d
to
th
e
seco
nd
decode
r
as
a
pr
io
ri
inf
or
m
at
ion
(L
a
)
in
orde
r
to
i
m
prov
e
the
reli
abili
ty
of
the
secon
d
decode
r’
s
outp
ut
and
vice
ver
s
a.
In
this
it
er
at
ive
decodin
g
pr
oc
ess,
the
apr
i
or
i
inf
or
m
at
ion
is
al
ways
avail
able
for
the
const
it
uen
t
de
cod
e
rs
e
xce
pt
for
the
first
decodin
g
sta
ge
wer
e
t
he
only
inform
ation
avail
able
is
that
th
e
pro
bab
il
it
y
of
occurre
nce
of
on
e
of
t
h
e
2
m
com
bin
at
ion
s
of
th
e
enc
oded
sy
m
bo
l
is
equi
prob
a
ble.
T
hus,
th
e
apr
i
or
i i
nf
or
m
at
ion
will
b
e i
niti
al
iz
ed
with
[5
]
:
Pr
(
=
)
=
(
1
2
)
(1)
Af
te
r
a
few
it
erati
on
s
,
a
dec
isi
on
is
co
nduc
te
d
on
the
fi
nal
APP
to
ev
al
uate
the
sy
m
bo
l
that
w
a
s
recei
ve
d.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
C
on
ver
ge
nce
analysis
of the
unpu
ncture
d
t
urbo Trel
li
s
-
c
od
e
d mod
ula
ti
on (
UT
TCM)
... (
Ah
med H
affa
ne
)
449
At each
tim
e k
(
k
= 1
, 2, .
..
,
L
); the MAP
de
cod
e
r
com
pu
te
s the A
Po
ste
ri
or
i Pr
obabili
ty
A
k,i
that t
he
inf
or
m
at
ion
sym
bo
l dk
was
i
(i =
0,
1, . . . ,
2
m
-
1
)
giv
e
n
the
receive
d
se
que
nce
y
an
d
sel
ect
the sym
bo
l wit
h
the
la
rg
est
AP
P
. T
hu
s
, eac
h dec
oder
com
pu
te
s
Pr
{
=
|
}
=
∑
∑
(
′
,
,
′
)
.
−
1
(
′
)
.
(
)
(2)
Wh
e
re,
the te
r
m
(
con
st) is a
c
on
sta
nt a
nd it
is cancele
d ou
t
by the
norm
al
i
zat
ion
of (2)
over
all
i to
un
it
y,
S
k
=
s
is t
he pre
sent
sta
te
o
f
t
he
e
nc
od
e
r
tre
ll
is a
nd
S
k
−
1
=
s
′
is t
he pre
vious stat
e
α
k
(
s
′
)
is the
forwar
d m
e
tric
s co
m
pu
te
d
a
s
(
)
=
∑
(
,
′
)
.
−
1
(
′
)
′
(3)
with the
init
ia
l condit
ion
0
(
0
)
=
1
0
(
)
=
0
f
or
all
s
≠0
(4)
β
k
(
s
)
is t
he bac
kw
a
r
d
m
et
rics compu
te
d
as
−
1
(
)
=
∑
(
,
′
)
.
(
′
)
′
(5)
with the
init
ia
l condit
ion
(
0
)
=
1
(
)
=
0
f
or
all
M≠
0
(6)
γ
k
(
y
k
,
s
,
s
′
)
is t
he bra
nc
h
t
ran
sit
io
n
m
et
ri
cs g
i
ven b
y
[5
]
(
,
,
′
)
=
(
|
=
,
=
,
−
1
=
′
)
.
(
=
|
=
,
−
1
=
′
)
.
(
=
,
−
1
=
′
)
(7)
The
te
rm
q
(
d
k
=
i
|
S
k
=
s
,
S
k
−
1
=
s
′
)
is
equ
al
to
ze
ro
or
on
e
,
de
pe
ndin
g
on
w
het
her
the
i
nput
sy
m
bo
l
d
k
=
i
induc
e
s
a
transiti
on
from
sta
te
s
′
to
s
or
no
t
a
nd
Pr
(
S
k
=
s
,
S
k
−
1
=
s
′
)
is
set
to
the
a
pri
or
i
pro
bab
il
it
y of t
hat sym
bo
l.
In
the
fi
rst
te
rm
of
(
7)
,
y
k
is
the
no
isy
ve
rs
ion
of
the
tra
ns
m
itted
sym
b
ol
c
onsti
tuted
f
ro
m
m
syst
e
m
atic b
it
s,
the
pa
rity
b
it
gen
e
rated
b
y t
he on
e
en
c
oder
and
par
it
y bit
gen
e
rated
b
y t
he othe
r
e
ncod
er.
If
c
k
2
∈
{0
,
1}
is
the
oth
e
r
enc
oder
par
it
y
bit
gen
e
rated
at
ste
p
k,
the
pro
bab
il
it
y
p
(
y
k
|
d
k
=
i
,
s
k
=
s
,
s
k
−
1
=
s
′
)
is set
to
(
|
=
,
=
,
−
1
=
′
)
=
∑
(
,
2
=
|
=
,
=
,
−
1
=
′
)
=
∈
{
0
,
1
}
1
2
∑
(
|
=
,
=
,
−
1
=
′
,
2
=
)
∈
{
0
,
1
}
(8)
wh
e
re it i
s ass
um
ed
that
(
2
=
|
=
,
=
,
−
1
=
′
)
=
(
2
=
)
=
1
2
(9)
i.e.: t
he value
of
c
k
2
is st
at
ist
ic
al
l
y i
nd
e
pende
nt
of
d
k
,
s
k
an
d
s
k
−
1
an
d
ca
n
ta
ke
eit
her ze
ro or
on
e
.
As
in
t
he
case
of
TTCM
the
de
cod
i
ng
al
gorit
hm
can
op
e
rat
e
in
the
l
og
a
rithm
ic
do
m
a
in
and
the
L
og
-
MAP
decode
r
i
s u
se
d or i
ts i
m
pro
ved alg
or
it
hm
[
11
]
,
[
12]
3.
CONVE
RGE
NC
E
PRE
DI
CTIO
N USI
N
G E
X
IT
CHA
RT
I
n
an
it
erati
ve
decodin
g
proc
ess,
base
d
on
SI
S
O
dec
od
e
rs
,
the
ext
rinsic
i
nfor
m
at
ion
ge
ner
at
e
d
at
t
he
ou
t
pu
t
of
each
decoder
is
pa
ssed
to
the
in
pu
t
of
the
ot
he
r
one
as
apr
i
or
i
inf
orm
ation
.
O
ver
it
erati
on
s
,
the
extrinsic
inf
or
m
at
ion
at
the
i
nput
of
the
de
cod
e
r
e
nhance
s
the
knowle
dge
a
bout
t
he
tr
ansm
itted
sym
bo
ls
.
T
o
qu
a
ntify
the
inf
or
m
at
ion
that
te
l
ls
the
extrinsic
inf
or
m
at
ion
ab
ou
t
th
e
transm
i
tt
ed
sy
m
bo
ls,
the
m
utu
al
inf
or
m
at
ion
is
us
ed.
T
he
refo
re
for
the
dec
od
i
ng
process
to
conve
rg
e
th
ere
m
us
t
be
a
n
increase
in
m
utu
al
inf
or
m
at
ion
acro
ss
it
erati
ons.
Th
us
,
the
m
utu
al
info
rm
at
ion
“I
E
”
com
pu
te
d
for
the
extri
ns
i
c
inform
at
ion
at
the
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
2
,
Fe
bru
ary
201
9
:
447
–
452
450
ou
t
pu
t
of
eac
h
com
po
ne
nt
dec
od
e
r
m
us
t
be
gre
at
er
tha
n
the
m
utu
al
info
rm
at
ion
“I
A
”
c
ompu
te
d
f
or
t
he
a
pr
i
or
i
inf
or
m
at
ion
at
it
s input.
Ten
Brin
k
[
6]
pro
po
se
d
to
tr
ack
the
ev
olu
ti
on
of
the
e
xtri
ns
ic
m
utu
al
in
form
ation
f
or
an
it
erati
ve
decodin
g
proc
ess
on
a
si
ng
l
e
com
ponen
t
de
cod
e
r
by
s
ub
sti
tuti
ng
the
a
pr
i
or
i
i
nfor
m
ation
-
wh
ic
h
s
houl
d
be
pro
du
ce
d
at
t
he
out
pu
t
of
the
ot
her
dec
od
e
r
-
by
a
ve
ct
or
gen
e
rated
base
d
on
a
fixe
d
ap
rio
ri
m
utu
al
inf
or
m
at
ion
“I
A
”
and
to
com
pu
te
m
utu
al
inform
ation
“I
E
”
of
the
e
xtrin
sic
inform
ation
ge
ner
at
e
d
at
it
s
ou
t
pu
t.
This
operat
io
n
is
rep
eat
ed
f
or
diff
e
re
nt
value
s
of
“I
A
”.
T
he
resu
lt
in
g
“I
E
”
values
are
repr
esented
as
a
functi
on
of
“I
A
”
on
a
curve
cal
le
d
the
extrinsic
inf
orm
at
ion
transf
e
r
char
act
e
risti
c.
In
orde
r
to
fl
ow
the
e
xch
a
nge
of
extrinsic
in
for
m
at
ion
in
an
it
erati
ve
dec
od
i
ng
process
,
bo
th
decode
r
cha
racteri
sti
cs
are
plo
tt
ed
into
a
sing
le
diag
ram
in
whic
h
the
axes
of
the
transf
e
r
c
har
act
erist
ic
s
of
the
seco
nd
de
cod
e
r
are
s
wa
pp
e
d.
T
he
res
ul
ti
ng
diag
ram
is the Ex
trin
sic
Infor
m
at
ion
Tr
a
nsfe
r
(
EX
IT) cha
rt.
3.1.
Exit ch
ar
t
f
or UTT
C
M
In
the
ca
se
of
the
UTTCM
we
fo
ll
ow
t
he
m
et
ho
d
descri
be
in
[
13
]
t
o
gen
e
rate
the
ve
ct
or
of
th
e
apr
i
or
i sy
m
bo
l
pro
bab
il
it
ie
s.
Assum
ing
that
the
bin
a
ry
bits
const
it
uting
each
tra
ns
m
i
tt
e
d
sym
bo
l
are
m
utu
al
l
y
ind
ep
end
e
nt,
the
a
pr
i
or
i
sym
bo
l
prob
a
bili
ti
es
are
ge
ne
rate
d
via
the
obser
vatio
n
of
the
inform
a
ti
on
sym
bo
l
ov
e
r
a
m
ul
ti
di
m
ension
al
Gaussi
an
channel.
Nam
el
y,
ever
y
bit
from
each
sy
m
bo
l
is
BPSK
m
od
ulate
d
a
nd
then
transm
itted
thr
ough
an
in
de
pe
nd
e
nt
A
WG
N
cha
nnel
.
So
the
a
pr
io
ri
pro
bab
il
it
y
of
each
bit
withi
n
ea
ch
sy
m
bo
l
is
ge
ne
rated
as
f
or
the
bin
a
ry
case
acco
rd
i
ng
to
[6
]
,
t
hen
the
ob
ta
ine
d
l
og
-
li
kelihoo
d
value
s
ar
e
conve
rted
t
o
sy
m
bo
l prob
a
bili
ty
, acco
r
ding t
o:
(
=
(
1
,
2
,
…
,
)
)
=
∏
1
+
=
1
(10)
Wh
e
re
d
is
th
e
encode
d
sy
m
bo
l
fo
rm
ed
f
ro
m
the
m
info
rm
at
ion
bits
b
j
.
L
j
is
the
apr
i
or
i
lo
g
-
li
kelihood val
ue
s of the
b
it
b
j
The
e
xtri
ns
ic
pro
bab
il
it
ie
s
m
us
t
be
e
valuat
ed
at
t
he
ou
t
put
of
t
he
decoder
.
But,
in
th
e
case
of
the
UTTCM
the
s
yst
e
m
atic
bits
and
the
t
wo
pa
rity
bits
are
jointl
y
transm
itted
thr
ou
gh
t
he
cha
nn
el
,
th
us
the
extrinsic
an
d
the
syst
e
m
at
ic
i
nfor
m
at
ion
cannot
be
se
par
a
te
d
[3
]
.
H
owe
ver,
in
[7
]
it
is
assum
ed
t
hat
li
ke
the
case
of
the
bi
na
ry
turbo
co
di
ng,
the
syst
e
m
at
ic
info
rm
at
ion
re
pr
ese
nts
th
e
inform
at
ion
relat
ive
to
the
channel
ob
s
er
vation
of
the
syst
e
m
ati
c
bits.
Ther
e
f
or
e
,
it
can
be
evaluated
a
nd
subtract
e
d
al
ong
with
the
apr
io
ri
inf
or
m
at
ion
from
the
a
po
st
erior
i
sym
bo
l
pro
bab
il
it
y
at
t
he
ou
t
pu
t
of
t
he
decoder
to
com
pu
te
the
e
xtrin
si
c
par
t.
Yet, in
ou
r
cas
e the c
hannel
obser
vatio
n
is a
sso
ci
at
ed
al
s
o wit
h
the
tw
o p
arit
y bit
s.
So
,
s
upposin
g
that
the
transm
it
te
d
data
sy
m
bo
ls
a
re
eq
ually
li
kely
,
the
sy
stem
at
ic
infor
m
at
ion
about
the sym
bo
l
d
k
gi
ven b
y
[
7]
{
P
(
d
k
=
i
|
y
k
)
}
i
=
0
2
m
−
1
.
is
com
pu
te
d
as
{
(
=
|
)
}
=
∑
∑
(
|
=
(
=
.
1
.
2
)
)
.
1
2
=
0
1
1
=
0
(11)
Wh
e
re
s
k
is t
he c
od
e
w
ord
a
nd
c
k
1
,
c
k
2
are the
tw
o parit
y bit
s.
The
c
orres
pondin
g
a
ver
a
ge
e
xtrin
sic
m
utu
al
infor
m
at
ion
wi
ll
b
e cal
culat
ed
as
[8]:
(
)
=
+
1
∑
[
∑
(
(
)
)
.
log
2
(
(
(
)
)
)
2
−
1
=
0
]
=
1
(12)
Wh
e
re
e
D
(
.
)
is
the
extrinsic
pr
ob
a
bili
ty
and
E
[
.]
is
the
e
xpect
at
ion,
w
hich
can
be
a
ppr
ox
im
at
ed
by
si
m
ple tim
e
-
aver
agi
ng of the
extrinsic
pro
ba
bili
ti
es o
f
t
he
i
nfor
m
at
ion
sy
m
bo
l.
Eq
uation (
12)
i
s also use
d
t
o
c
al
culat
e the a
ve
rag
e
a
pr
io
ri i
nfor
m
at
ion
I
A
(
d
)
.
4.
RESU
LT
S
A
ND AN
ALYSIS
In
this secti
on, t
he
exit chart i
s u
se
d
to in
ves
ti
gate the p
er
fo
rm
ance o
f
UT
TCM
o
ve
r
A
W
GN
c
ha
nn
el
for
2
an
d
3
bp
s
sp
ect
ral
eff
ic
ie
ncies.
The
m
a
in
adv
a
ntage
of
the
Exit
char
t
is
that
the
anal
ysi
s
is
per
form
ed
on
a
sing
le
decod
er.
N
o
BER
si
m
ulati
on
of
the
it
erati
ve
de
cod
i
ng
sc
hem
e
it
sel
f
is
req
uire
d.
H
ow
e
ve
r,
this
si
m
ula
ti
on
is
carried
out
for
each
case
in
orde
r
to
ve
rify
the
accu
racy
of
the
exit
char
t
to
pr
e
dict
the
E
b
/N
0
conve
rg
e
nce
thres
hold.
Fi
gure
4
show
s
the
resu
lt
ant
EX
I
T
char
ts
com
pu
te
d
f
or
2
bp
s
UTTCM
syste
m
for
var
i
ou
s
E
b
/N
0
values
.
T
he
U
TTCM
enc
od
e
r
is
m
ade
u
p
of
tw
o
ide
ntica
l
8
-
sta
te
rate
2/
3
rec
ursive
syst
e
m
at
ic
conv
olu
ti
on
c
ode
us
in
g
(
13,
11,
05)
[
5]
as
ge
ner
at
or
poly
nom
ials,
a
bl
oc
k
le
ngt
h
of
16
384
sym
bo
ls
and
16
-
QA
M
c
onste
ll
at
ion
s
wit
h
G
ray
m
app
ing
.
At
the
recei
ve
r
side,
a
sym
bo
l
-
by
-
sym
bo
l
log
-
M
AP
dec
od
i
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci
IS
S
N:
25
02
-
4752
C
on
ver
ge
nce
analysis
of the
unpu
ncture
d
t
urbo Trel
li
s
-
c
od
e
d mod
ula
ti
on (
UT
TCM)
... (
Ah
med H
affa
ne
)
451
al
gorithm
is
u
sed.
F
r
om
this
fig
ur
e
it
can
be
see
n
that
f
or
E
b
/N
0
=
5.4
dB,
t
he
tw
o
e
xtrin
sic
in
f
or
m
at
ion
trans
fer
c
har
a
ct
erist
ic
of
th
e
two
decode
rs
intersect
ea
rly
.
From
this
po
i
nt
of
i
ntersecti
on,
the
m
utu
al
inf
or
m
at
ion
at
the
outp
ut
of
both
decode
rs
w
il
l
be
poor
e
r
th
en
the
m
utu
al
inf
or
m
at
ion
at
their
in
put,
m
ea
ning
that,
afte
r
s
ome
it
erati
on
t
he
dec
od
e
r
was
no
m
or
e
able
to
pr
oduce
e
xt
rinsic
in
f
or
m
ation
with
e
nha
nce
d
inf
or
m
at
ion
on
the
tra
ns
m
itte
d
bits.
Th
ere
fore
the
it
erati
ve
process
will
be
unable
t
o
co
nver
ge
wh
at
e
ve
r
is
the
nu
m
ber
of
it
er
at
ion
s. Fo
r
E
b
/N
0
=
5.6
dB th
e two
c
urves
have
j
us
t a
voide
d
the i
ntersecti
on b
e
fore
reac
hing the
po
i
nt
(1,1),
m
eanin
g
that,
for
this
E
b
/N
0
an
d
above
the
de
cod
e
r
pro
duc
e
an
increase
d
extrinsic
in
for
m
at
ion
and
th
us
the
it
erati
ve
process
will
con
verge
to
correct
trans
m
itted
sequ
e
nc
e
with
high
prob
a
bili
ty
.
Hence
,
the
conve
rg
e
nce
t
hr
es
hold
is
ap
pro
xim
a
te
ly
5.
6
dB.
H
ow
e
ve
r,
it
can
be
cl
early
seen
that
the
co
nve
rg
e
nce
thres
ho
l
d
in
B
ER
curve
of
t
he
Fig
ure
3.
is
at
E
b
/N
0
=
6
dB.
Me
ani
ng
t
hat,
the
re
is
a
def
e
ren
ce
of
0.4
dB
betwee
n
the
two
res
ults,
w
hich
c
ould
be
ex
plain
by
the
li
m
it
validit
y
of
the
ass
um
pt
ion
that
e
xtrin
sic
inf
or
m
at
ion
and the
syst
em
ati
c inform
at
ion
c
an be se
par
at
e
d [8
]
.
Figure
3. Bi
t er
ror rat
e (BE
R)
curves
f
or
UT
TCM
syst
e
m
(
2 b
ps
with
16
-
QA
M
a b
loc
k
le
ng
t
h of 1
6384
sy
m
bo
ls) at
dif
fer
e
nt d
ec
odin
g
it
erati
ons
Figure
4
.
EXIT
ch
a
rts for U
T
TCM
syst
e
m
(
2 bp
s
with
16
-
QA
M
and b
l
ock len
gt
h
of
1638
4
sy
m
bo
ls)
at
v
ari
ou
s
E
b
/N
0
values
In
Fig
ur
e
6
,
th
e
res
u
lt
ant
E
XIT
cha
rts
com
pute
d
for
3
bps
UTTCM
is
depi
ct
ed.
T
he
e
nc
od
e
r
is
m
ade
up
of
tw
o
ide
ntica
l
8
-
sta
te
rate
3/4
recursi
ve
syst
e
m
at
ic
conv
olu
ti
on
cod
e
us
in
g
(
13,
11
,
05,
13)
[
5]
as
gen
e
rato
r
poly
no
m
ia
ls,
32
-
Q
AM
co
ns
te
ll
at
i
on
s
with
Gr
ay
m
app
in
g,
a
n
in
te
rlea
ver
le
ng
t
h
of
16
384
sy
m
bo
ls
and
a
lo
g
-
M
A
P
dec
oder
at
t
he
receive
r
si
de.
The
c
onve
r
ge
nce
is
ac
hieve
d
f
or
E
b
/N
0
=
10.
4
dB,
with
a
0.2
dB
m
is
m
a
tc
h
fro
m
the
co
nver
ge
nce
t
hr
e
shol
d
in
BER
c
urve
of
Fig
ur
e
5
,
w
hich
m
ay
be
again
relat
e
d
to
th
e
assum
ption
tha
t ex
trinsic i
nfo
rm
ation
a
nd th
e syst
e
m
at
ic
inf
orm
ation
ca
n be se
par
at
e
d [8]
.
Figure
5. Bi
t er
ror rat
e (BER)
curves
f
or
UT
TCM
syst
e
m
(
3 b
ps
with
16
-
QA
M
a b
loc
k
le
ng
t
h of
16384 sy
m
bo
ls) at
diff
e
ren
t
dec
od
i
ng it
erati
on
s
Figure
6. EX
IT
ch
a
rts for U
T
TCM
s
yst
e
m
(
3 bp
s
with
16
-
QA
M
and a
blo
c
k
le
ngth
of
16384 s
ym
bo
ls)
at
v
ari
ou
s
E
b
/N
0
values
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
1
3
, N
o.
2
,
Fe
bru
ary
201
9
:
447
–
452
452
5.
CONCL
US
I
O
N
In
t
his
pa
per,
t
he
co
nver
ge
nc
e
behavi
or
o
f
it
erati
ve
dec
ode
r
of
the Unpu
nc
ture
d
T
urbo
T
rell
is
Cod
e
d
Modula
ti
on
(
U
TTCM
)
us
in
g
the
non
-
bi
nar
y
EXI
T
c
har
t
w
as
inv
est
igate
d.
The
accu
racy
of
the
exit
ch
art
to
pr
e
dict
the
dec
od
i
ng
c
onve
r
ge
nce
th
reshold
has
bee
n
ver
if
ie
d
thr
ough
ex
a
m
ples.
The
re
su
lt
s
wer
e
c
ompar
e
d
to
th
os
e
of
BE
R
curve
a
nd
ha
ve
s
how
n
a
sl
igh
t
m
is
m
at
ch,
wh
ic
h
is
e
xp
l
ai
ne
d
by
the
li
m
it
ed
validit
y
of
t
he
ind
e
pende
nce
of
t
he
e
xtrinsi
c
inf
or
m
at
ion
of
eac
h
U
TTC
M
e
m
it
te
d
sy
m
bo
l
fr
om
it
s
syst
e
m
atic
inf
or
m
at
ion
.
Nev
e
rtheless
,
the
relat
ively
less
com
plica
te
d
m
et
ho
d
of
generati
on
m
akes
the
EXI
T
c
har
t
a
m
or
e
co
m
petit
ive
te
chn
iq
ue f
o
r
t
he
c
om
par
ison
of d
if
fer
e
nt sc
hem
es as w
el
l a
s for
t
he
sea
rc
h o
f new
d
esi
gns.
REFERE
NCE
S
[1]
Berr
ou
C,
Glavieux
A
and
Thi
tim
aj
shim
a
P.
Near
Shannon
li
mit
error
-
correc
ti
n
g
codi
ng
and
dec
oding:
Tur
bo
-
code
s.
Proce
edi
n
gs
of
ICC
'
93
-
I
EE
E
Int
ern
ation
al
Confer
en
ce
o
n
Comm
unic
at
io
ns
,
Gene
va,
Sw
i
tz
er
la
nd,
1993
;
2:
1064
-
1070.
[2]
Ungerboe
ck
G.
Channe
l
cod
ing
with
m
ult
il
evel
/pha
se
signal
s.
I
EE
E
Tr
ansacti
o
ns
on
Informati
on
Theory
;
1982.
28(1):
55
-
67
.
[3]
Robert
son
P
a
nd
W
orz
T
.
B
andwidt
h
-
eff
ic
i
e
nt
turbo
trelli
s
-
code
d
m
od
ulat
ion
using
pun
c
ture
d
component
code
s.
IE
EE J
o
urnal
on
Se
lected A
reas i
n
Com
municat
ions
.
19
98;
16(2):
206
-
2
18.
[4]
Bassou
A,
Djebba
ri
A.
Eff
i
cient
turbo
enc
oding
sche
m
e
using
unpunct
ure
d
tr
elli
s
-
code
d
m
odula
t
ion
code
s.
A
EU
-
Inte
rnational
Jo
urnal
of El
e
ct
ro
n
ic
s and
Comm
u
nic
ati
ons
.
2007;
61(9):
621
-
626
.
[5]
Bassou
A,
Djeb
bar
i
A,
B
enaiss
a
M.
Design
of
unpunct
ure
d
tu
rbo
tre
l
li
s
-
cod
ed
m
odula
ti
on
.
A
EU
-
Inte
rnationa
l
Journal
of
Elec
t
ronics
and
Com
municat
ions.
20
13;
67(3):
223
-
2
32.
[6]
Brink
S.
t
en.
C
onver
gence
beh
avi
or
of
it
er
at
iv
ely
de
code
d
p
ar
al
l
el
con
cate
na
t
ed
code
s
.
IE
E
E
Tr
ansacti
ons
on
Comm
unic
ati
ons
.
2001;
49(10)
: 1727
–
1737.
[7]
Chen
H,
Haimovich
A.
EXI
T
ch
art
s
for
turbo
tr
e
ll
is
-
cod
ed
m
odula
ti
on
.
I
EE
E
Co
mm
unic
ati
ons
Letters
.
2004
;
8(1
1):
668
-
670.
[8]
Klie
wer
J,
Ng
S
X
,
Hanz
o
L
.
Eff
icient
Com
pu
ta
ti
on
of
EXIT
Functi
ons
for
Nonbinar
y
It
erati
v
e
Dec
oding
.
IE
EE
Tra
nsac
ti
ons on
Com
m
unic
at
ions.
2006;
54
(12): 2133
-
2136.
[9]
Sali
m
M,
Yada
v
R
P,
Narwal
K
,
Sharm
a
A.
A
New
Bloc
k
S
-
R
andom
Inte
rle
av
er
for
Shorter
Le
ngth
Fram
es
fo
r
Turbo
Codes.
Bu
ll
etin
o
f
E
lectrical
Eng
ine
ering
a
nd
Informatic
s (
BE
EI)
,
2013
;
2
(
4):
293
-
298
.
[10]
Rekka
l
K,
Bassou
A.
Im
provin
g
the
Perform
anc
e
of
Vite
rbi
De
code
r
using
W
indow
Sy
stem.
Int
ernati
onal
Journ
al
of
E
le
c
tric
al
and
Computer
Eng
i
nee
ring (
IJE
C
E)
,
2018
,
8(1):
611
-
621.
[11]
W
ang
J,
Li
J,
Cai
C.
Two
n
ovel
de
codi
ng
a
lgori
thms
for
tu
rbo
code
s
base
d
on
tay
lor
seri
es
in
3GP
P
LT
E
s
y
stem.
Indone
s
ian
Journal
of
Elec
tri
cal
Engi
n
eer
ing
and
Compu
te
r Sc
ie
nc
e
(
IJEECS)
.
2014;
12(
5):3467
–
3475
.
[12]
W
ang
J,
Li
J,
C
ai
C.
Novel
D
ec
oding
Algor
ithm
for
BICM
-
I
D
Embedde
d
Turbo
Codes.
Ind
onesian
Journal
of
El
e
ct
rica
l
Eng
in
ee
ring a
nd
Computer
Sc
ie
nc
e
(
IJ
EE
CS)
,
2014
;
1
2(5):
3460
-
3466
.
[13]
Grant
A.
Con
v
erge
nce
of
non
-
binary
iterati
v
e
de
codi
ng
.
GL
OBECOM
'
01.
I
EE
E
Glob
al
Telec
om
m
unic
atio
ns
Con
fer
ence, San
Antonio, T
X.
2
001;
2:
1058
-
10
62
.
Evaluation Warning : The document was created with Spire.PDF for Python.