Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
6, N
o
. 1
,
Febr
u
a
r
y
201
6,
pp
. 99
~105
I
S
SN
: 208
8-8
7
0
8
,
D
O
I
:
10.115
91
/ij
ece.v6
i
1.9
337
99
Jo
urn
a
l
h
o
me
pa
ge
: h
ttp
://iaesjo
u
r
na
l.com/
o
n
lin
e/ind
e
x.ph
p
/
IJECE
Robust Digital Predistortion in
Saturati
on Region of Power
Amplifiers
So
on-i
l
Ho
ng
, Kw
an
g-P
y
o
L
ee,
E
u
i
-
Ri
m J
e
on
g
*
Department o
f
I
n
formation and
Communication
E
ngineering, HanbatNational University
, 305-71
9
, K
o
r
e
a
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Aug 22, 2015
Rev
i
sed
No
v 5, 201
5
Accepted Nov 29, 2015
This paper pro
poses adigital predis
tortion (D
PD) technique
to improve
line
a
riz
a
tion
per
f
orm
a
nce when
the power
am
pli
f
ier (PA) is dr
iv
en ne
ar th
e
saturation reg
i
on
. Th
e PAis a no
n-linear
d
e
vice in gener
a
l,
and th
e nonlinear
distortion b
ecom
e
s severer
as
the output pow
er in
creases
. However, th
e PA’s
power effi
cien
c
y
in
cre
a
s
e
s
as
the P
A
outp
u
t power in
cr
eas
es
.
The
nonline
a
rit
y
res
u
lts in spe
c
tr
al
regrowth, whi
c
h
leads
to ad
ja
ce
nt chann
e
l
interf
eren
ce,
an
d degrades the
transm
it signal qualit
y. Ac
cor
d
ing to our
sim
u
lation,
the
l
i
near
iza
tion p
e
rf
orm
a
n
ce of DP
D is degrad
ed abruptly
when
the PA operates in its satur
a
tion
region.
To reliev
e
this problem,
we propose
an improvedDPD techniqu
e.
Th
e propos
ed techn
i
que performs o
n
/off contro
l
of the adaptiv
e algorithm
based
on the m
a
gnitude of the transm
itted signa
l.
S
p
ecifi
cal
l
y
, th
e
adapta
tion nor
m
a
ll
y
wo
rks for
small and medium signals
while it stops for large sign
als
.
Th
er
efore, har
m
ful coefficien
t updates
b
y
saturated signals can be
avoid
e
d. A computer
simulation shows that the
proposed method can
improve the lin
ear
ization performance co
mpared
with
the
conventional DPDmethod in
highly
dr
iven P
A
s.
Keyword:
Dig
ital pred
ist
o
rtion
No
n-l
i
near di
st
ort
i
o
n
Power am
p
lifie
r
Recursi
v
e least
square
s (RL
S
)
Satu
ration
Copyright ©
201
6 Institut
e
o
f
Ad
vanced
Engin
eer
ing and S
c
i
e
nce.
All rights re
se
rve
d
.
Co
rresp
ond
i
ng
Autho
r
:
Eui
-
R
i
m
Jeong
,
Depa
rt
m
e
nt
of
In
fo
rm
at
i
on an
d C
o
m
m
uni
cati
on
En
gi
nee
r
i
n
g,
Han
b
at
Nat
i
o
n
a
l
Uni
v
ersi
t
y
,1
25
D
o
ngse
o
da
ero
,
Yuse
o
n
g
-
gu
, Daeje
o
n, 3
0
5
-
71
9,
K
o
rea
Em
a
il: erj
e
on
g@h
a
nb
at.ac.k
r
1.
INTRODUCTION
During the las
t
two deca
des
,
m
a
ny researc
h
es
fo
r im
p
r
ov
ing
th
e
p
e
rform
a
n
ce o
f
power am
p
lifier
(PA
s
)
ha
ve
bee
n
per
f
o
rm
ed. I
n
part
i
c
ul
a
r
, t
h
e l
i
n
eari
t
y
an
d
efficiency
of the PA ca
n
be
us
ed as
a
n
indica
tor
of
the
perform
a
nce of t
h
e
ove
rall comm
unica
tion system
. Th
e PA
is
a non
-lin
ear d
e
v
i
ce in
g
e
n
e
ral, and
th
e
no
nl
i
n
ea
r di
st
ort
i
o
n bec
o
m
e
s severe
r as t
h
e o
u
t
p
ut
po
wer in
creases. Th
e n
o
n
lin
earity resu
lts in
sp
ectral
regrowth, wh
ich
lead
s to
adj
a
cen
t ch
an
n
e
l interferen
ce, and d
e
grad
es th
e t
r
an
sm
itted
sig
n
a
l qu
ality. To av
o
i
d
th
e sev
e
re
n
o
n
l
in
earity, o
n
e
si
m
p
le
so
lu
tio
n
i
s
to
d
r
iv
e
th
e PA in
a lo
w power reg
i
on
,
(i.e., th
e lin
ear reg
i
on
).
Howe
ver, the
powe
r e
fficienc
y
of t
h
e PA is l
o
we
re
d to
10%
.
In
order to i
n
crease th
e
power efficiency, t
h
e PA
n
eeds to
b
e
dri
v
en
a h
i
g
h
power, tog
e
th
er
with
lin
ear
izat
ion tec
hni
ques
to linearize t
h
e PA. Li
nearization
t
echni
q
u
es i
n
c
l
ude
feed
bac
k
,
anal
o
g
pre
d
i
s
t
o
rt
i
o
n (P
D
)
,
and
fee
d
f
o
r
w
ard m
e
t
hods
[
1
]
–
[
4
]
.
Am
ong t
h
ese
t
echni
q
u
es
, t
h
e
di
gi
t
a
l
predi
s
t
o
rt
i
o
n (
D
P
D
) i
s
kn
ow
n t
o
be
t
h
e
m
o
st
cost
and
per
f
o
r
m
a
nce-ef
fect
i
v
e [
5
]
.
DP
D
base
d
on
er
ro
r fee
d
back
c
o
rrect
i
o
n i
s
a
po
we
rf
ul
l
i
n
ea
ri
zat
i
on
st
rat
e
gy
beca
use
i
t
has
t
h
e
nat
u
r
e
o
f
a
man
a
g
eab
le dig
ital o
p
e
ration
,
and
th
e error co
rrection is in
sen
s
itiv
e to
a
m
p
lifier v
a
riatio
n
s
, su
ch
as
t
e
m
p
erat
ure
,
s
u
p
p
l
y
v
o
l
t
a
ge,
and
de
vi
ceva
r
i
a
t
i
ons, as
we
ll as the nonlinea
r cha
r
acteristics of t
h
e PA.By using
t
h
e i
n
put
a
n
d
out
put
si
gn
al
s
of t
h
e P
A
,
t
h
e
DP
D i
m
pl
em
ent
s
t
h
e i
nve
rse
fu
nct
i
o
n
o
f t
h
e
PA t
o
l
i
n
ea
ri
ze t
h
e
n
on-
lin
ear
PA
. Th
e in
v
e
r
s
e fu
n
c
tion
can
b
e
r
ealized
eith
er b
y
Lo
oku
p
Tab
l
e o
r
p
o
l
ynomial. Be
tw
een th
e
m
,
polynom
i
al PD is prefe
r
able
because it e
xhi
bits bette
r linearization pe
rformance an
d faster
converge
nce
.
The
coefficients
of the Polynom
ial PD
a
r
efound
via ada
p
tive
algorithm
s
, su
ch as Least
Mean Squa
re
(LMS)
orRec
u
rsi
v
e
Least Square
(RLS)
[6]
–
[9]. Howe
ve
r, according t
o
our
sim
u
la
tion, the line
a
rization
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE Vo
l. 6
,
N
o
. 1
,
Febru
a
ry
2
016
: 9
9
–
10
5
10
0
per
f
o
r
m
a
nce d
e
gra
d
es se
vere
l
y
when
t
h
e PA op
erates i
n
its satu
ratio
n
reg
i
on
. To
reliev
e
th
is pro
b
l
em
,
th
is
pape
r p
r
op
ose
s
a new
DP
D
t
echni
q
u
e. T
h
e p
r
o
p
o
se
d t
echni
que
pe
rf
orm
s
on/
o
ff c
ont
rol
o
f
t
h
e
adapt
i
v
e
alg
o
rith
m
b
a
sed
o
n
t
h
e tran
sm
it
ted
sig
n
al’s m
a
g
n
itu
de. Specifically, fo
r sm
all si
g
n
a
ls, th
e adap
tiv
e
p
r
ed
istortio
n
al
g
o
rith
m
wo
rks n
o
rm
al
ly. In
co
n
t
rast, th
e ad
ap
tiv
e alg
o
rithm sto
p
s
for larg
e sign
als u
n
til
s
m
a
l
l
si
gnal
s
occ
u
r
agai
n.
T
h
ere
f
o
r
e,
ha
rm
ful
coef
fi
ci
ent
up
d
a
t
e
s by
se
ve
r
e
n
onl
i
n
eari
t
y
can
be
a
voi
ded
.
A
com
put
er si
m
u
l
a
t
i
on sh
ow
s t
h
at
t
h
e
propos
e
d technique linearize the PA be
tter th
an
t
h
e co
nv
en
tio
n
a
l
PD in
h
i
gh
ly dr
iv
en
PA
s.
2.
PROP
OSE
D
DPD
TECHNIQUE
We c
o
nsider a
DPD structure
base
d
on the i
n
direct learning
architect
ur
e as
sh
own
in Fi
g
u
re 1[
10
].
The
DPD c
o
efficients are
found at
t
h
e
p
o
st
di
st
ort
e
r
by
l
i
n
ea
ri
zi
ng
t
h
e
PA
-
p
o
st
di
st
o
r
t
e
r c
h
ai
n.
T
h
e
coefficients a
r
e
copied t
o
t
h
e
PD
bloc
k.
Fi
gu
re
1.
B
l
oc
k
di
ag
ram
of di
gi
t
a
l
pre
d
i
s
t
o
rt
i
o
n
Th
e ad
ap
tiv
e alg
o
rith
m
f
o
r
find
ing
th
e PD
param
e
ters at
th
e p
o
s
t
d
isto
rter
n
eeds two
inpu
ts: Th
e pred
isto
rter
out
put
v(
n
)
an
d t
h
e
fee
dbac
k
si
gnal
a
(
n
)
.
T
h
e
post
d
i
s
t
o
rt
e
r
an
d
PD
ha
ve
t
h
e sam
e
st
ru
ct
ure, a
n
d t
h
i
s
pa
per
e
m
p
l
o
y
s a
p
o
l
y
n
o
m
ial
m
o
d
e
l.
Th
e
ou
tpu
t
v(n)
o
f
t
h
e PD is
written
a
s
∗
|
|
∗
,
∗
,…,
∗
|
|
⋮
|
|
(1
)
wh
ere
2
Q
+1
is th
e m
a
x
i
m
u
m p
o
l
yno
m
i
al o
r
d
e
r
for th
e
pred
isto
rtion
,
,
,…,
is the coefficient
vector of
t
h
e predistortion,
and
,
|
|
,…,
|
|
.T
he
p
o
st
di
st
ort
e
r
i
s
m
odel
e
d
b
y
th
e
sam
e
p
o
lyn
o
m
ia
l
m
o
d
e
l. Th
e ou
tpu
t
z(n
)
of th
e po
st
d
i
sto
r
tion
can
b
e
written
as
1
∗
|
|
∗
,
∗
,…,
∗
|
|
⋮
|
|
(2
)
whe
r
e
2
Q
+1
i
s
t
h
e m
a
xim
u
m
pol
y
n
o
m
i
al
order
f
o
r
t
h
e
p
o
st
di
st
ort
i
o
n,
,
,…,
is t
h
e coe
fficient
vect
o
r
of
t
h
e p
o
st
di
st
o
r
t
i
o
n,
a
n
,
and
,
|
|
,…,
|
|
.
Fo
r t
h
e ad
ap
tiv
e algo
rith
m
fo
r
find
ing
th
e
p
o
s
t
d
istort
er c
o
efficients, we
consi
d
er the
recursi
v
e least
sq
uares
(RLS) alg
o
rith
m
.
Th
e RLS algo
rith
m
u
p
d
a
tes
the post
distort
e
r coefficientsat every sam
p
le to
minimize the s
qua
re
d error
between
v(n)
a
n
d z(
n
)
. T
h
us,
t
h
e
post
d
i
s
t
o
rt
er
be
com
e
s the
inve
rse system of the
PA.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Rob
u
s
t
Dig
ita
l
Pred
isto
rtion
i
n
S
a
t
u
r
a
tio
n Reg
i
on
o
f
Po
wer Amp
lifiers (Eui-Rim Jeo
n
g
)
10
1
Tab
l
e
1
.
Th
e d
i
g
ital p
r
ed
istortio
n algo
rith
m
of ind
i
rect
learning a
r
c
h
itecture
In
itializatio
n
:
Ad
ap
tiv
e al
go
rith
m
fo
r calcu
l
a
tin
g
th
e pred
isto
rtion
co
efficien
ts:
Th
e p
r
op
osed RLS
algo
r
ith
m
fo
r
up
d
a
ti
n
g
w
is su
mm
ariz
ed
in Tab
l
e
1
.
Th
e
(Q+1)-b
y
-(Q+1
)
m
a
trix
P
re
prese
n
t
s
t
h
e i
nve
rse c
o
rre
l
a
t
i
on m
a
t
r
i
x
. (
1
) a
n
d
(2
) i
n
T
a
bl
e 1
are t
h
e
up
dat
e
s
of
w
an
d
P
to m
i
nimize the
squ
a
re
d er
r
o
r |
e
(n
)|
2
at each time “n”, re
spec
tively. In t
h
e c
o
nve
n
tional R
L
S algorithm
,
the
updates
of
w
and
P
always work at every sam
p
le. Howe
ver, the
pr
opose
d
m
e
thod in Ta
ble 1 occasionally updates
w
and
P
onl
y
whe
n
|
|
i
s
sm
al
ler t
h
a
n
a t
h
res
hol
d val
u
e,
. Specifically, if
|
|
,
w
and
P
ar
e n
o
r
m
a
l
l
y
u
p
d
a
ted,
o
t
h
e
rw
ise th
e upd
at
e do
es
no
t
o
c
cu
r, an
d
w
an
d
P
are no
t ch
ang
e
d un
tilsm
a
ll sig
n
a
l app
ears
ag
ain
.
There
f
ore,
har
m
ful
up
dat
e
s
b
y
hi
g
h
l
y
sat
u
ra
t
e
d sam
p
l
e
s can
be a
v
oi
de
d.
The m
a
gni
t
u
de
t
h
re
sh
ol
d
shou
ld
be ap
pr
op
ri
at
el
y
det
e
r
m
i
n
ed by
consi
d
e
r
i
ng t
h
e t
y
pes
of P
A
s bei
n
g
used an
d t
h
e PA i
n
p
u
t
si
gnal
’
s
characte
r
istics. After t
h
e operati
o
n
of th
e adap
tiv
e alg
o
rith
m
for
∈
1,
, th
e fin
a
l
w
is co
p
i
ed
in
to
th
e
PD
bl
oc
k.
3.
COMPUTER SIMULA
TION RESULTS
The
per
f
o
rm
ance o
f
t
h
e
p
r
o
p
o
se
d D
P
D
t
e
c
hni
que
i
s
e
x
a
m
i
n
ed t
h
ro
u
g
h
com
put
er si
m
u
l
a
t
i
on.
T
h
e
sim
u
lation environm
entsare as follows. T
h
e
input signa
l of
the PAis the
LTE uplink si
gnal. T
h
e m
odulation
schem
e
i
s
64
-
QAM
(
Q
ua
d
r
a
t
ure
Am
pl
i
t
ude M
o
dul
at
i
o
n)
,
an
d t
h
e
ban
d
w
i
d
t
h
i
s
20M
Hz.
As
a P
A
m
odel
,
we
u
s
e t
h
e Saleh
m
o
d
e
l [11
]
. The d
e
tailed
m
o
del is as fo
llows.
1
|
|
|
|
|
|
1
.
1
,
0
.
3
,
1
1
(3
)
Th
e id
eal
g
a
in
o
f
t
h
e PA is assu
m
e
d
to
b
e
“1”, and
th
e sat
u
ratio
n
po
in
t is |
y
(n
)|=1
. Th
is
mean
s th
at if
th
e PA i
n
pu
t m
a
g
n
itud
e
ex
ceed
s
“1
”, th
e PA
o
u
t
p
u
t
falls in
t
o
t
h
e sat
u
ration
reg
i
o
n
.
In the sim
u
latio
n
,
to
d
r
i
v
e
th
e am
p
lifier o
v
e
rt
h
e
satu
rati
o
n
reg
i
o
n
, th
e
max
i
m
u
m
v
a
lu
e o
f
th
e tran
smit
ted
sig
n
a
l’s
m
a
g
n
itu
d
e
is
set to
“1.6”.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE Vo
l. 6
,
N
o
. 1
,
Febru
a
ry
2
016
: 9
9
–
10
5
10
2
Fig
u
re
2
.
Ch
aracteristics o
f
po
wer am
p
lifier (AM
–
AM ch
aracteristics)
Figure
3. Cha
r
acteristic of t
h
e powe
r am
p
lifier (AM–
PM
ch
aracteristics)
Figures
2 a
n
d
3 s
h
ow t
h
e AM–AM a
n
d AM–PM charact
eristics of t
h
e
PA m
odel. T
h
e red curve (a
)
shows
the i
d
ea
l characte
r
istics and t
h
e blue curve (b)
s
h
ows
the
c
h
aracte
r
istics o
f
th
e
PA
m
o
d
e
l in
(3
).
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Rob
u
s
t
Dig
ita
l
Pred
isto
rtion
i
n
S
a
t
u
r
a
tio
n Reg
i
on
o
f
Po
wer Amp
lifiers (Eui-Rim Jeo
n
g
)
10
3
Fi
gu
re
4.
S
p
ect
rum
s
at
po
we
r
am
pl
i
f
i
e
r o
u
t
p
ut
Fi
gu
re
5.
S
p
ect
rum
s
at
po
we
r
am
pl
i
f
i
e
r o
u
t
p
ut
Fig
u
re 4
sh
ows th
e p
o
wer sp
ectru
m
o
f
th
e PA ou
tpu
t
sig
n
a
ls wh
en
the PA is d
r
iv
en a relativ
el
y
lin
ear
r
e
g
i
on
(
f
ar
fr
o
m
th
e sat
u
r
a
tion
po
in
t)
.
I
n
t
h
e sim
u
lati
o
n
, th
e PA
inpu
t sign
al’
s
m
a
g
n
itud
e
is under
“1
”
so t
h
at
al
l
t
h
e si
gnal
s
are am
pli
f
i
e
d bel
o
w t
h
e
sat
u
rat
i
on
poi
nt
. The m
a
xim
u
m
pol
y
nom
i
a
l orde
r o
f
t
h
e D
P
D i
s
15
, an
dt
he t
h
r
e
sh
ol
d
is “1
”. Th
e red
cu
rve (a) is th
e tran
sm
it
ted
sig
n
a
l’s sp
ectru
m
.
Th
is is th
e id
eal
spectrum
.
The blue c
u
rve (b)
is the
PA
o
u
t
pu
t sp
ectru
m
with
ou
t pred
istort
i
o
n
.
Sign
ificant sp
ectral reg
r
owth
is
o
b
s
erv
e
d
du
e to
th
e PA’s nonlin
earity.
The
black c
u
rve
(c) is the PA
out
put
s
p
ect
r
u
m
wi
t
h
t
h
e c
o
nve
nt
i
ona
l
DPD. Th
e sp
ectral reg
r
owth
is red
u
ced
by
m
o
re t
h
an 3
0
dB
. Th
e gre
e
n
curve (d) is the PA output spectrum
wi
t
h
t
h
e
p
r
op
o
s
ed
DP
D.
The
con
v
e
n
t
i
onal
a
n
d
p
r
op
ose
d
D
P
D t
e
c
hni
que
s
sho
w
al
m
o
st
t
h
e sam
e
perf
or
m
a
nce.
Fi
gu
re
5 s
h
ow
s t
h
e
PA
o
u
t
p
u
t
spect
ru
m
s
when
th
e PA is
driv
en
ov
ert
h
e satu
ration
reg
i
o
n
. Here, the
m
a
xim
u
m
PA i
nput
si
g
n
al
’
s
m
a
gni
t
ude i
s
‘1.
6
’ s
o
t
h
a
t
so
m
e
port
i
o
n of t
h
e si
gna
l
fal
l
s
i
n
t
o
t
h
e PA’s
satu
ration
reg
i
o
n
.
W
ith
t
h
e co
nv
en
tio
n
a
l
DPD,
on
ly a
10
dB
re
duct
i
o
n
o
f
t
h
e s
p
ect
ral
r
e
gr
o
w
t
h
i
s
o
b
s
e
rve
d
.
The l
i
n
ea
ri
zat
i
o
n
pe
rf
orm
a
nc
e de
gra
d
es s
e
verel
y
at
t
h
e
PA’
s
sat
u
rat
i
o
n re
gi
on
. H
o
weve
r, t
h
e
pr
op
ose
d
m
e
t
hod f
u
rt
h
e
r
re
duces
t
h
e sp
ect
ral
re
gr
owt
h
by
3 – 4 dB
co
m
p
ared
t
o
t
h
e con
v
e
n
t
i
onal
D
P
D.
In
sum
m
ary
,
t
h
e
pr
o
pose
d
D
P
D
pe
rf
orm
a
nce i
s
com
p
arable to
th
at
of the con
v
en
tion
a
l
DPD at
th
e
PA’s lin
ear regio
n
wh
ile th
e fo
rm
er p
e
rfo
rms b
e
tter th
an
t
h
e latter at th
e PA’s
satu
ratio
n
reg
i
o
n
. Th
ese resu
lts
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
I
J
ECE Vo
l. 6
,
N
o
. 1
,
Febru
a
ry
2
016
: 9
9
–
10
5
10
4
i
ndi
cat
e t
h
at
t
h
e p
r
op
ose
d
t
echni
que
c
a
n i
n
crease the
PA’s powe
r efficien
cy to
g
e
t
h
er
with
its lin
eari
zatio
n
perform
a
nce com
p
ared
with
t
h
e co
nv
en
tion
a
l DPD.
4.
CO
NCL
USI
O
N
In t
h
i
s
pa
pe
r,
we p
r
op
ose
d
a
new
D
P
D t
e
c
hni
que
t
h
at
i
m
pr
o
v
es t
h
e l
i
n
e
a
ri
zat
i
on
per
f
o
rm
ance whe
n
t
h
e PA i
s
dri
v
e
n
nea
r
t
h
e sat
u
rat
i
on
regi
on
b
y
t
h
e go/
st
op
co
n
t
ro
l of th
e ad
ap
tiv
e
al
gorithm
.
Specifically, for
sm
a
ll sig
n
a
ls, th
e ad
ap
tiv
e
pred
istortion
alg
o
rith
m
wo
rk
s n
o
rm
all
y
wh
ile th
e ad
ap
tiv
e alg
o
rith
m
sto
p
s
for
larg
e si
g
n
a
ls un
til sm
a
ll sig
n
als o
c
cu
r
ag
ai
n. Th
u
s
,
h
a
rm
fu
l co
efficien
t
u
p
d
a
tes
b
y
severe no
n
lin
earity can
b
e
avoi
ded
.
T
h
e s
i
m
u
l
a
t
i
on res
u
l
t
s i
ndi
cat
e t
h
at
t
h
e
pr
o
pose
d
pre
d
i
s
t
o
rt
i
o
n
t
echni
que
i
s
m
o
re
ef
fect
i
v
e t
h
an
t
h
e
co
nv
en
tio
n
a
l
DPD
wh
en th
e
PA is
d
r
i
v
en n
e
ar th
e saturatio
n reg
i
on
.
ACKNOWLE
DGE
M
ENTS
Thi
s
re
searc
h
was s
u
pp
ort
e
d
by
t
h
e
Wo
rl
d C
l
ass
300
R&D
projects funded by t
h
e Sm
all and
M
e
di
um
B
u
si
n
e
ss A
d
m
i
ni
st
rat
i
on
of
K
o
rea
(
S
2
3
1
8
1
0
9
)
.
REFERE
NC
ES
[1]
S.C. Cripps
, "RF power
amplifie
r
s
for wireless co
mmunications",
Artech
House M
i
crowave Library
, 2006
.
[2]
T. Nojima
and
T. Konno, “Cu
b
er predistor
tio
n lineari
zer for
relay
e
quipment in 800 MHz b
a
nd land
mobile
telephone s
y
stem”,
IE
EET
rans. Ve
h. Techno
l
., v
o
l. VT-34
,
no
. 6
,
pp. 169–177, Nov. 1985
.
[3]
J. Cha,
J. Yi
, J.
Kim
,
and B.
Ki
m
,
“
O
ptim
um
de
sign of
a p
r
edist
o
rtionRF power
am
plifier
for m
u
ltic
arri
er W
C
DMA
applications”,
IE
EET
r
ans
. Micr
o
w
. T
h
eor
y
T
ech
., vol. 52, no. 2, p
p
. 655–663
, Feb
.
2004.
[4]
Y.
Y.
Woo,
Y.
Y
a
ng,
J. Yi,
J
.
Nam,
J.
Cha, and
B.
Ki
m
,
“
A
new adapt
i
vef
eedfor
w
ard am
plifi
e
r f
o
r W
C
DM
A base
stations using
imperfe
ctsignal can
cellation”,
Microw. J
., vol. 46
, n
o
. 4
,
pp
. 22–44
,
Apr. 2003.
[5]
R. Marsalek, P.
Jardin,
and G. B
a
udoin, “From p
o
st-distor
tion to predistortion
for
power
amplifier
s
linear
ization
”
,
IEEE
Commun.
L
e
tters
, vo
l. 7, p
p
. 308-310
, July
2003.
[6]
S. Takab
a
y
a
shi, M. Orihashi,
T.
Matsuoka,
and M. Sagawa, “Adaptivepre
d
i
stortion lin
ear
izer with dig
i
tal
quadratur
e modem”,
IEEE51st Veh.Techno
l.
Con
f
, vo
l.3
,
no
.3, pp. 2237–2241, 200
0.
[7]
Eui-Rim Jeong, “A New Poly
nomial Digi
tal
Predistortion M
e
thod B
a
sed on
Direct Learn
i
n
g
for Linear
izin
g
Nonlinear Powe
r Am
plifier”,
Th
e Journal of the
Korean Institute of
Information and Communication Engin
eering
,
pp. 2382–2390
,
vol. 11
, no
.12, 2
007.
[8]
Chang B
y
ong-
Kun, Jeon Chang-Dae, Song
Dong-Hy
uk
, “P
erformance Improvement
in
Alternate Mainbeam
Nulling b
y
Adap
tive
Estimation
of Convergen
ce
Paramete
rs in
Linearly
Constr
ain
e
d Adaptiv
e Arr
a
y
s
”,
Journal of
information
and
communication
convergen
c
e
eng
i
neering
, pp
. 39
2–398, vol.7, no. 3, 2009
.
[9]
Sungho Choi, Eui-Rim Jeong, Lee. Y.H., "A Direct
Lear
ning Structure
for Adaptive Poly
nomial-Based
Predistortion
for
Power Am
plifie
r Lin
eari
z
a
tion",
Vehi
cular
T
echn
o
logy Con
f
er
enc
e
,
2007,
VTC2007-Spring.
IEEE
65th, pp
.1791–1
795, April. 2007
.
[10]
Young-Doo Ki
m, Eui-Rim Jeong, and Y
ong H. Lee, "Adaptive compensation
f
o
r power amplif
ier nonlin
ear
ity
in
the pr
esence of
quadrature modulation
/
demodu
lation
errors",
S
i
gnal Processin
g
, IEEE Transactions on
, 55(9)
(2007): 4717-47
21.
[11]
A.
A.
M.
Saleh,
“Fre
quency
-
independent and fr
eque
ncy
-
dep
e
ndent nonlinear mode
ls of TW
T am
plifiers”,
IE
EE
Trans. Commun
., vol. COM-29,
no. 11
, pp
. 1715
-1720, Nov. 198
1.
BIOGRAP
HI
ES OF
AUTH
ORS
Soon-il Hong receiv
e
d B.S. degr
ees in
the Dep
a
r
t
ment of Rad
i
o
Wave
Engin
eering from Hanbat
National Univer
sity
, Daejeon
,
K
o
rea, in 2013
. H
e
is curr
ently
pu
rsuing a Master’
s
degree
in th
e
department of R
a
dio Wave Engi
neering
from
Hanbat Nat
i
ona
l Universit
y
.
His re
search
inter
e
sts
are
in
the
are
a
s
o
f
digi
tal
s
i
gna
l p
r
o
cessing, p
r
edis
tortion
,
and
modem design.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Rob
u
s
t
Dig
ita
l
Pred
isto
rtion
i
n
S
a
t
u
r
a
tio
n Reg
i
on
o
f
Po
wer Amp
lifiers (Eui-Rim Jeo
n
g
)
10
5
Kwang-Py
o
Lee receiv
ed B
.
S.
degrees
in
the
Department o
f
Radio
Wave En
gineer
ing from
Hanbat National University
,
Daejeon, Korea,
in 2
013. He is cu
rr
ently
pursuing a
Master’s degree
in the depar
t
ment of Rad
i
o Wave Engin
eer
ing
from Hanbat National Univ
e
r
sity
. His re
sea
r
c
h
inter
e
s
t
s
ar
e in
t
h
e ar
eas
of
dig
i
t
a
l s
i
gn
al p
r
oces
s
i
ng, pr
edis
tort
io
n, and
m
odem
d
e
s
i
gn.
Eui-Rim
Jeong receiv
e
d B.S., M.S., and Ph.D. de
grees in
the Depar
t
m
e
nt
of Elec
tric
al
Engineering fro
m the Korea Ad
vanced Institute
of
Science and
Techno
log
y
(KAIST), Daejeon,
Korea, in 1995, 1997, and 2001, respectively
.
He
is currently an associate pr
ofessor in the
Department o
f
I
n
formation and
Communication
E
ngineering, Hanbat Nati
onal U
n
iversity
, since
2009. His research inter
e
sts are in the areas of
communication signal processing, predistortion
,
and modem design.
Evaluation Warning : The document was created with Spire.PDF for Python.