Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
5, N
o
. 2
,
A
p
r
il
201
5, p
p
.
35
5
~
36
0
I
S
SN
: 208
8-8
7
0
8
3
55
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
Multiple Processes for Le
ast Mean Square Adaptive
Algorithm on Roadwa
y Noise Cancelling
Sri Ar
ttini
Dwi Prasetyowati
*,
Adhi
Sus
anto**
*Dep. of
Electrical
Engin
eer
ing,
Industrial Techn
o
lo
g
y
Facu
lty
,
S
u
ltan
Agung Islamic
University
, Sema
rang, Indo
nesia
** Dep. of
Electrical Eng
i
neerin
g, Facu
lty
of
En
gin
eer
ing, Gadjah Mada Univer
sity
, Yog
y
ak
arta, I
ndonesia
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Ja
n
4, 2014
Rev
i
sed
Feb
22
, 20
15
Accepted
Mar 10, 2015
Noise is a prob
lem often found
in daily
life. No
ise also mak
e
p
e
ople
could
not concentr
ate to do their
work.
Efforts to r
e
duce noise hav
e
been
proposed,
but, due to variety
of the noise’s charac
te
ri
st
ic
s,
eve
r
y
noi
se problem requires
differen
t
s
o
lutio
n. This
res
e
arc
h
aim
to canc
e
l
the veh
i
cl
e’s
nois
e
whil
e
maintaining th
e information heard. Th
ese co
nditions happen
e
d in the
hospitals classrooms, or work ro
om n
ear the ro
a
d
wa
y. The
vehi
cle’s
nois
e
changes ver
y
f
a
st, so the adap
tive s
y
stem is the good solution candidate for
solving this pro
b
lem. On th
e b
e
ginning
, th
e simulation pro
ces
s had the
trouble with th
e iter
a
tions
. Matlab softwar
e
on
ly
can
ex
ecute
the certain
range of i
t
er
ati
on. It cou
l
d no
t can
cel
the no
is
e, ev
en th
e i
n
form
ation
becom
e
s
cripti
c.
The problem
is how to
cancell the vehic
l
e’s noi
se with the
restric
tion software and sti
ll m
a
n
a
ge the
im
portan
t
inform
ation
.
T
h
is research
will m
odif
y
the
LMS adaptive algorithm
so that the iteration co
uld be done
b
y
th
e s
y
s
t
em
and the m
a
in g
o
al of th
e res
e
a
r
ch could b
e
re
ached
. Th
e
modification of the algorithm is
based
on the filter length (L) used to adapt
with the no
ise.
Therefor
e,
this r
e
search
conducted simulation o
f
the
adaptive
noise c
a
nce
lling
with two p
r
oc
ess steps. Th
e
output of
the
fi
rst adap
tiv
e
proces
s
have the
.
s
a
m
e
chara
c
t
e
ri
s
tics
with the no
is
e that would b
e
can
cel
led
,
thus
the firs
t a
d
aptiv
e proces
s
have the error
near to zero
.
The s
econd
adaptive process
changes the inp
u
t b
y
the outpu
t of the first process and
mix
the inform
ation
into the noise
.
Error occur
e
d in the fin
a
l pro
cess is the
inform
ation h
ear
d as
th
e dominan
t output.
Keyword:
Ad
ap
tiv
e pro
c
ess
Filter len
g
t
h
LMS ada
p
tive
No
ise can
cellin
g
Vehicle’s
nois
e
Copyright ©
201
5 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
:
Sri Arttin
i Dwi
Prasetyowati,
Depa
rtem
ent.of Electrical E
n
ginee
r
ing,
In
d
u
st
ri
al
Tec
h
nol
ogy
Facul
t
y
Su
ltan
Ag
ung
Isla
m
i
c
Un
iv
ersity,
Se
m
a
ran
g
, Indo
n
e
sia
Em
a
il: arttin
i
@
un
issu
la.ac.i
d
1.
INTRODUCTION
Th
is research
i
s
in
sp
ired
b
y
actu
a
l situ
atio
n
in
a h
o
sp
ital roo
m
in
Se
m
a
ran
g
con
t
am
in
ate
d
with
n
o
i
ses
fr
om
t
h
e nearb
y
road
. It
i
s
ve
ry
di
ffi
c
u
l
t
t
o
m
a
ke con
v
e
r
sa
t
i
ons bet
w
ee
n adm
i
ni
st
rat
i
on of
fi
cers a
nd
g
u
e
st
s i
n
the presence
of roa
d
way
noise
s, thus th
e
of
fi
cers nee
d
t
o
fr
eque
nt
l
y
repea
t
th
e in
fo
rm
ati
o
n. Therefore,
it is o
f
g
r
eat im
p
o
r
tance to
find
a tech
n
i
q
u
e
to
can
c
el th
e no
ise
while th
e in
form
a
tio
n
can
still b
e
h
e
ard.
The
vehicle
noise cha
n
ges fast, he
nce the
syste
m
m
u
st a
d
apt s
p
ee
dily and
accurately. Least Mea
n
Squ
a
re (LMS) Ad
ap
tiv
e is
a pro
s
p
ectiv
e
so
lu
tion
for
t
h
is p
r
ob
lem
d
u
e
to
its rob
u
stn
e
ss; m
o
reov
er th
e
al
go
ri
t
h
m
does
n
’t
nee
d
m
a
ny
k
n
o
w
n
vari
a
b
l
e
s. A
d
a
p
t
i
v
e syste
m
is syste
m
th
at can
ad
ap
t (witho
u
t
operato
r)
so that it can be optim
al. The
syste
m
always does the
pro
cess so that the variable is very close to the target
[1
]. Ad
ap
tiv
e
No
ise Can
c
ellatio
n
(ANC
) app
licatio
n
is u
s
ef
ul in wi
de ra
nge of scena
r
ios
,
suc
h
as confe
r
ence
room
[2]. It m
eans t
h
at the a
d
aptive al
gorithm
is ve
ry
s
u
i
t
a
bl
e f
o
r
resea
r
c
h
on
r
o
a
d
way
noi
se
cancel
l
a
t
i
on.
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E
V
o
l
.
5,
No
. 2, A
p
ri
l
20
15
:
35
5 – 3
6
0
35
6
In fact, if t
h
e t
a
rget
fluct
u
ates over tim
e (not sta
tic), t
h
e
proces
s s
h
ould
be
done
as s
oon as
possi
ble
wh
ile th
e erro
r
m
u
st
be
kept
m
i
nim
a
l
.
A
d
apt
i
v
e
sy
st
em
s have se
veral
c
h
aract
e
r
i
s
t
i
c
s:
can a
d
apt
au
to
m
a
tical
ly, can
b
e
t
r
ain
e
d to
do
filtering
and
m
a
k
e
decisio
n
,
h
a
v
e
si
m
p
ler step
s th
an th
e non
-ad
a
p
tiv
e
sy
st
em
s alt
h
o
u
g
h
ha
ve a c
o
m
p
l
e
x anal
y
s
i
s
, and ca
n
devel
op t
h
e m
odel
[1]
.
LM
S al
gori
t
hm
s have
p
e
rform
a
n
ce i
ssu
es related to
in
su
fficien
t
ex
citati
on,
no
nst
a
t
i
ona
ry
refere
nce i
n
put
s
,
fi
ni
t
e
p
r
eci
si
o
n
arithm
e
tic, quantization noi
s
e, and m
easurem
ent noi
se
. Suc
h
factors
cause weight drift and potentia
l
in
stab
ility in
th
e con
v
e
n
tional LMS algorith
m
[3
]. Besid
e
s LM
S algorith
m
s
, th
ere
are sev
e
ral
adap
tive
algorithm
s
, e.g. NLMS (Normalized
Least
Mean Square
), RLS (Recursi
v
e Least Squa
re).
One
of possible
schem
e
s is to
place a
n
ide
n
tical filter in t
h
e
refe
re
nce si
gnal path t
o
t
h
e
weight update
of the LM
S al
gorithm
,
also
called
Filtered-x LMS
(Fx
L
MS) algo
rit
h
m
[4
].
Usi
n
g t
h
e si
m
p
l
e
st
st
ruct
ure
t
h
at
have a
n
easy
expl
an
at
i
on f
o
r t
h
e a
n
al
y
t
i
c
descri
pt
i
on,
Li
nea
r
Com
b
iner, the
LMS Ada
p
tive
can
be e
x
ec
ute
d
sim
p
ly. The
form
ula can be
shown a
s
:
L
l
l
k
lk
L
k
Lk
k
k
k
k
k
ok
k
x
w
x
w
x
w
x
w
x
w
y
0
2
2
1
1
...
(1
)
Out
put
y
k
can
be calculated from
the lin
ear com
b
ination
of the i
n
put
x
k
and t
h
e
wei
g
ht
w
lk
. LM
S
Ad
ap
tiv
e is one of th
e sim
p
le
st Ad
ap
tiv
e Al
g
o
rith
m
th
at
can
so
lv
e th
e co
m
p
lex
prob
lem
o
f
v
e
h
i
cle’s no
ise.
Th
e LM
S
Ad
ap
tiv
e
Algo
rithm
can
b
e
sh
own
o
n
th
e equ
a
tio
n (2
).
k
X
W
W
k
1
k
2
k
(2
)
Equ
a
tio
n
(2) is u
s
ed
to
fi
n
d
th
e co
rrect weig
h
t
u
s
ed
in
Equ
a
tio
n
(1). In
add
ition
,
the Mean
Sq
u
a
re Error
(MSE) n
e
ed
s to
b
e
d
e
term
in
ed
, m
a
in
ly from
th
e d
i
fference of
n
o
i
se to
b
e
can
celled
an
d th
e ou
tpu
t
o
f
the
syste
m
. Eq
u
a
tio
n
s
(1
) an
d
(2) are u
s
ed
in
the co
nfigu
r
ation d
e
scri
b
e
d
i
n
Fig
u
re 1
for v
e
hicle n
o
i
se can
cellin
g
in
th
is wo
rk
.
Fig
u
re 1
showed
con
f
i
g
u
r
ation
with
t
w
o
inpu
t, th
e fi
rst in
pu
t con
t
ain
s
sign
al
k
s
and noise
k
n
, whil
e
t
h
e ot
he
r i
n
put
cont
ai
ns
o
n
l
y
noi
se si
gnal
k
n
'
, un
de
r t
h
e co
n
d
i
t
i
on
k
n
'
and
k
n
c
o
m
e
f
r
o
m s
a
me
n
o
i
s
e
b
u
t
taken
from
differe
nt places.
In
bloc
k “S.A”, the input
a
nd
output are
processe
d wit
h
Linea
r
Com
b
ine
r
,
wh
ereas in
b
l
ock
“Algo
r
itm
a
”
th
ere is p
r
o
c
ess for fi
nd
ing th
e weig
h
t
.
Ad
ap
tiv
e algorith
m
ad
j
u
sts th
e filter
coefficient include
d in the ve
ctor
W
n
. Th
e ad
ap
tiv
e
filter ai
m
s
to
eq
u
a
te i
t
s o
u
t
pu
t y(n) to
th
e d
e
sired
ou
tput
d(
n)
. F
o
r
eac
h
i
t
e
rat
i
on, t
h
e e
r
ro
r si
gnal
i
s
gi
ven
by
:
ɛ
(
n
)
= d(
n)
- y(
n)
(3)
whe
r
e
ɛ
or e
r
r
o
r i
s
di
ffe
re
nc
e bet
w
ee
n de
si
red i
n
p
u
t
d a
n
d o
u
t
p
ut
y
.
T
h
e er
ro
r si
g
n
al
i
s
fed
back i
n
t
o
t
h
e
filter, whe
r
e the filter c
h
aract
eristics are alte
re
d accordi
ngl
y as shown in
Figure
2
[5].
Fig
u
re 1
.
Con
f
i
g
uration
o
f
Noise
Can
cellin
g
Fig
u
re
2
.
Ad
aptiv
e Filter Configu
r
ation
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Title o
f
ma
nu
scrip
t
is sh
o
r
t
an
d clea
r, imp
lies resea
r
ch resu
lts (First Au
tho
r
)
35
7
2.
SEAR
CH
IN
G
OPTI
MU
M
DELA
Y
VAL
UE
AN
D TW
O P
R
O
C
ESS
LMS
A
D
A
P
T
IVE
NOIS
E
CANCELLING
Data Co
llection
was
d
o
n
e
b
y
reco
rd
ing
v
e
hicle n
o
i
se
from
th
e d
i
fferen
t
lo
catio
n in
t
h
e sam
e
ti
me.
The fi
rst n
o
ise
(
k
n
) i
n
Fi
g
u
re
1
i
s
vehi
cl
e
noi
se hear
d i
n
t
h
e
ro
om
desi
gnat
e
d f
o
r
t
h
e
noi
s
e
cancel
i
n
g. T
h
e
second noise is the roadway noise (
n
’
k
) from the source of the noise. T
h
e noise com
e
s
from
m
o
torcycle and
cars
wi
t
h
di
ffe
rent
fuel
t
y
pes
.
F
u
rt
he
rm
ore,
i
n
f
o
rm
at
i
on s
i
gnal
was al
s
o
reco
r
d
ed
an
d
sim
u
l
a
t
e
d com
b
i
n
e
d
with
th
e
first
ro
adway no
ise (
s
k
.
)
.
I
t
sh
ou
l
d
be no
ted that th
e f
i
r
s
t and
th
e
second
r
ecor
d
i
n
g w
e
re do
n
e
sim
u
l
t
a
neousl
y
. Si
m
p
l
e
t
ool
s
t
h
at
co
ul
d
hel
p
t
h
e rec
o
rdi
n
g
as a di
vi
de
r
of
t
h
e t
w
o
dat
a
re
cor
d
i
n
g
were
a
dde
d
on
com
put
er
as
sh
ow
n i
n
Fi
g
u
r
e
3.
Fi
gu
re
3.
Si
m
p
l
e
t
ool
di
vi
der t
h
e t
w
o
n
o
i
s
e si
gnal
Searc
h
i
n
g t
h
e
opt
i
m
u
m
del
a
y val
u
e
was
do
ne f
o
r
fi
ve
ve
h
i
cl
e noi
se s
o
u
r
ces, i
.
e. t
r
uc
k,
bus
, m
o
t
o
rcy
c
l
e
, ca
r
wi
t
h
sol
a
r
fuel
,
car wi
t
h
n
on s
o
l
a
r f
u
el
, an
d
m
i
x of seve
ral
vehi
cl
es. B
eca
use o
f
t
h
e di
f
f
i
c
ul
t
y
of com
p
u
t
i
ng,
th
e op
ti
m
u
m
it
eratio
n
was sel
ected
fo
r
n
o
i
se can
cellin
g
.
The in
pu
t o
f
Ad
ap
tiv
e LMS is th
e no
ise h
e
ard
in
th
e
ro
om
(the first noise
) an
d the
refe
rence
of th
e Ada
p
tive
LM
S is noise from
the source (t
he second noise
)
. The
first step
was lo
ok
ing
fo
r th
e
o
p
tim
u
m
del
a
y
with
= 0,0
01
and
det
e
rm
i
n
i
ng t
h
e am
ount
of i
t
e
rat
i
on d
u
e t
o
num
ber o
f
sam
p
l
e
20.
00
0.
Searc
h
i
n
g of t
h
e o
p
t
i
m
u
m
del
a
y
based on
m
ovi
ng aver
ag
e from
t
h
e square err
o
r
(
2
).
B
y
obser
vi
n
g
t
h
e val
u
e of t
h
e
del
a
y
,
i
n
creasi
ng i
t
an
d cha
n
gi
n
g
L val
u
e
p
e
ri
o
d
i
cal
l
y
, opt
im
u
m
val
u
e
of
del
a
y
a
n
d
L co
ul
d
be
m
e
t
.
T
h
e l
a
r
g
e
r
t
h
e
L
val
u
e
t
a
ken
,
t
h
e m
ean
fr
om
m
ovi
ng a
v
e
r
age
t
e
nd
t
o
conve
r
ge
nt and less value ca
n be
obtaine
d.
Ideally, each
ve
hicle’s noise
has di
ffe
rent L
and
delay. In
fact the
v
a
lu
e t
h
at m
o
st
in
flu
e
n
tial in
LMS Ad
ap
tiv
e p
r
o
cess is t
h
e
v
a
lu
e
o
f
. Tab
l
e 1
sh
ow
th
e
valu
e of
D
e
lay and
L with
th
e sam
e
fo
r al
l
vehi
cl
es’ n
o
i
s
e.
Acc
o
r
d
i
n
g t
o
t
h
e o
b
t
a
i
n
ed
dat
a
,
D
e
l
a
y
and L
val
u
e can
be a
dde
d o
r
subt
ract
ed a
c
c
o
r
d
i
n
g t
o
t
h
e
re
qui
red
com
put
at
i
on
[6]
.
It
can be see
n
t
h
at
t
h
e
m
i
nim
u
m
val
u
e fo
r
L i
s
150, a
n
d
t
h
e
m
a
xim
u
m
val
u
e f
o
r L i
s
45
0. F
o
r t
h
e
weight iteration take
n 20.000, it will be hard for the
c
o
m
putation,
beca
us
e algorithm
will look for the
out
put
with 150 itera
tion each whi
c
h woul
d be
processe
d untill 20.000 iterati
ons for sea
r
chi
ng the
we
ights.
Co
n
s
equ
e
n
tly, it tak
e
s co
n
s
iderab
ly lo
ng
time. Fro
m
th
e
di
ffi
c
u
l
t
y
of searchi
ng L val
u
e and t
h
e bi
g i
t
e
rat
i
o
n
s
fo
r fi
ndi
ng t
h
e
wei
g
ht
s, t
h
i
s
re
search
di
vi
des
t
h
e LM
S ada
p
t
i
ve n
o
i
s
e cance
l
l
i
ng i
n
t
o
t
w
o s
t
at
es i
n
ho
pe t
o
get
fast p
r
ocess a
n
d m
i
nim
a
l erro
r in
res
u
lt.
Fi
gu
re
4 s
h
ow
t
h
e t
w
o
pr
oces
ses LM
S a
d
a
p
t
i
v
e
noise ca
nce
lling. T
h
e
proc
ess stages
are:
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E
V
o
l
.
5,
No
. 2, A
p
ri
l
20
15
:
35
5 – 3
6
0
35
8
-
First Process:
d
is in
doo
r
Si
gn
al (
S
D
)
,
x
i
s
t
h
e re
fere
nce o
f
t
h
e adapt
i
ve L
M
S or
out
do
o
r
si
gnal
(S
L), t
o
be e
xpect
e
d
t
h
e o
u
t
p
ut
cl
ose
t
o
SD, so that t
h
e e
r
ror val
u
e
(
error
) cl
ose t
o
zero.
-
Seco
nd P
r
oce
s
s:
d
i
s
SD
m
i
xed
by
i
n
fo
rm
at
i
on si
gn
al
, w
h
ereas
x
i
s
t
a
ken fr
om
out
pu
t
from
t
h
e fi
rst
pr
ocess
.
T
o
be expect
e
d
e
r
r
o
r
val
u
e
close t
o
i
n
form
ation signal [7].
(a)
(b
)
Fig
u
re
4
.
Th
e two processes
LMS ad
ap
tiv
e
n
o
i
se can
celling
;
(a) Th
e
first
p
r
o
cess of LM
S
ada
p
tive al
gorith
m
,
(b) T
h
e s
econd
proces
s
of LMS a
d
a
p
tive algorithm
The M
ean
Sq
uare E
r
r
o
r (M
SE) f
o
r LM
S
Ada
p
tiv
e
wi
t
h
onl
y
o
n
e
pr
o
cess an
d t
w
o pr
ocess
were
obt
ai
ne
d by
de
t
e
rm
i
n
i
ng t
h
e
L val
u
e o
n
t
h
e
st
at
e posi
t
i
on,
t
h
at
event
u
al
l
y
and
del
a
y
c
h
ange. T
h
e L
value
is o
p
t
i
m
u
m
v
a
l
u
e th
at can
b
e
tak
e
n
in
th
e LMS Ad
ap
tiv
e Process, so
th
at th
e co
m
p
u
t
atio
n
a
l sim
u
latio
n
can
be
sol
v
e
d
.
Ho
we
ver
,
i
f
m
o
re p
o
we
rf
ul
l
com
put
i
ng
res
o
u
r
ce
are pre
s
ent
,
L val
u
e
doe
s not
need t
o
b
e
t
a
ken
minim
a
l, but at any
value t
h
at
coul
d ac
hieve
the sm
allest MSE.
3.
RESULTS
A
N
D
DI
SC
US
S
I
ON
The Process
of calculating
L and
del
a
y
wi
t
h
t
h
e di
ese
l
-base
d
vehi
cl
es
as
a noi
se si
gnal
usi
n
g
“M
ovi
ng
A
v
er
age” m
e
t
hods
and
0
,
001
, is show
n
in
Figur
e
5
an
d Figu
r
e
6.
Fi
gu
re
5.
The
Pro
cess
of
cal
cul
a
t
i
ng
L a
n
d
d
e
l
a
y
;
L=19
0
de
l
a
y
220
(
r
e
d
)
, L=20
0
del
a
y
23
0 (b
lu
e)
,
L=22
0
del
a
y
23
0
(g
r
e
en
).
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
Title o
f
ma
nu
scrip
t
is sh
o
r
t
an
d clea
r, imp
lies resea
r
ch resu
lts (First Au
tho
r
)
35
9
Fi
gu
re
6.
The
Pro
cess
of
cal
cul
a
t
i
ng
L a
n
d
d
e
l
a
y
;
enl
a
rge
d
part
i
n
t
h
e
fi
r
s
t
pea
k
of
t
h
e
Fi
gu
re
1.
It
coul
d be co
ncl
u
ded
fr
om
f
i
gu
re 5 an
d 6 t
h
at
t
h
e bi
g
g
er
t
h
e val
u
e
of L
t
h
e sq
uare m
ovi
n
g
ave
r
ge i
s
m
o
re conve
rgent. Howe
ver, as a consequence, the
com
putation is also getti
ng
harder, and the process
becom
e
s l
o
n
g
e
r
.
Neve
rt
hel
e
ss
, i
t
d
o
es
not
g
u
a
rant
ee t
h
at
t
h
e com
put
at
i
on
gi
ves a
ccu
rat
e
resul
t
s
.
The
O
p
t
i
m
a
l
d
e
lay w
ith
230 sa
m
p
les o
f
0
.
00
52
seco
nd
is
a g
ood
v
a
l
u
e.
H
o
w
e
v
e
r
to
p
r
ev
en
t th
e m
i
ssi
n
g
o
f
d
e
lay, the d
e
lay
was t
a
ke
n
o
n
2
0
0
sam
p
l
e
s. Th
e res
u
l
t
fo
r t
h
e
ot
he
r ve
hi
cl
es
fo
r
20
.0
0
0
i
t
e
r
a
t
i
ons a
n
d
0
,001
, are
show
n in
Tabel
1. F
r
om
Tabel
1 s
h
o
w
e
d
t
h
at
L val
u
e
very
hi
gh
fo
r s
o
m
e
vehi
cl
es. The bi
gge
r val
u
e o
f
L t
h
e n
o
i
s
e wi
l
l
be e
r
ase
d
c
o
m
p
le
tely because t
h
e
value
of m
oving
a
v
erage
convergent to certain
num
ber. However the
pr
ocess
bec
o
m
e
s so
har
d
.
H
o
weve
r, t
o
a
n
t
i
c
i
p
at
e t
h
e bi
g
value of L, LM
S ada
p
tive
alg
o
rithm
s
were
m
odified
in
to
two
pro
c
esses.
Tabel
1. T
h
e
v
a
l
u
e o
f
L
,
,
an
d
del
a
y
fo
r se
ve
ral
ve
hi
cl
es
Vehicles
L
(Itera
tio
n
)
(Fix
ed)
Delay
(sa
m
ple
)
Diesel 150
0.
001
200
Bus
250
0.
001
200
T
r
uck 350
0.
001
200
Car 450
0.
001
200
M
o
tor
250
0.
001
200
Car
with Solar
Fuel
300
0.
001
200
The MSE
val
u
e
obtaine
d
from
LMS ada
p
tive algor
ithm
on the fi
rs
t and second
processes
was
i
nvest
i
g
at
e
d
an
d su
bse
q
uent
l
y
com
p
ared wi
t
h
t
h
e L
val
u
e i
n
o
n
e p
r
ocess.
Tabl
e 2 a
nd
3
sho
w
t
h
e M
S
E
val
u
es
fo
r
seve
ral
ve
hicles in
t
h
e
first a
n
d
seco
nd
process
of LMS
Ada
p
tive, re
spectivel
y. T
h
e
val
u
e
w
a
s
d
e
term
in
ed
con
s
tan
t
and
sm
a
ll en
ou
gh
, so
t
h
at th
e
pro
cess is ru
nn
ing
slowly bu
t qu
ite t
h
oro
ugh
.
Tabel
2.
T
h
e
M
S
E
val
u
es of
car wi
t
h
s
o
l
a
r
fu
el i
n
th
e first pro
cess LMS
Ad
ap
tiv
e
L
(Itera
tio
n
)
Delay
(Sa
m
pl
e)
(
fix
e
d
)
MSE1
100
230
0,
001
10.
235
7
230
110
0,
001
0,
1131
230
100
0,
001
0,
1200
230
120
0,
001
0,
1041
230
130
0,
001
0,
1041
230
140
0,
001
0,
1049
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E
V
o
l
.
5,
No
. 2, A
p
ri
l
20
15
:
35
5 – 3
6
0
36
0
Tabl
e 3.
T
h
e
M
S
E
val
u
es of
car wi
t
h
s
o
l
a
r
f
u
e
l
in
th
e s
e
con
d
pro
c
e
s
s
L
M
S
Ad
ap
tiv
e
L
(Itera
tio
n
)
Delay
(Sa
m
pl
e)
(Fix
ed)
MSE1 MSE2
100
230
0,
001
0,
2400
16,
294
4
230
100
0,
001
0,
0998
3,
9117
230
110
0,
001
0,
0990
11,
603
3
230
120
0,
001
0,
0995
7,
0464
230
130
0,
001
0,
1008
32,
710
2
Table 2 and
3
reveal t
h
at
is selected
with
co
nstan
t
v
a
l
u
e
wh
ile L and
delay v
a
ry. Th
e L v
a
l
u
e is
th
e op
ti
m
a
l v
a
l
u
e th
at can
reach
in
th
e LMS
Ad
ap
tiv
e
p
r
oc
ess, beca
use t
h
e com
put
er has
m
a
ny
l
i
m
i
t
a
t
i
ons
i
n
its calculation
[6]. T
h
e fi
rst
process
has M
S
E1 less t
h
an
1, a
nd t
h
is is
a sm
a
ll value because in the
first
pr
ocess t
h
ere i
s
no i
n
f
o
rm
at
ion m
i
x i
n
t
h
e
i
n
p
u
t
.
T
h
e
sec
o
nd process has
MSE2
m
o
re than
1 beca
use the
MSE2 i
n
clude
inform
ation that
m
u
s
t
b
e
o
c
cu
rr
e
d
a
s
an
er
ro
r.
4.
CO
NCL
USI
O
N
From
t
h
e resea
r
ch e
x
peri
m
e
nt
s, t
h
e r
o
a
d
way
noi
se ca
ncel
l
i
ng c
o
ul
d
be re
al
i
zed wi
t
h
L
M
S Ada
p
t
i
v
e
Al
g
o
ri
t
h
m
wi
t
h
m
odi
fi
cat
i
on
i
n
t
o
t
w
o
pr
oce
sses, wi
t
h
L =
23
0,
0
,
001
, and
dela
y as
m
u
ch as
100
sam
p
le
s
with linea
r com
b
iner struct
ure.
W
i
t
h
one
proces
s of LM
S ada
p
tive, the L val
u
e is st
ill cannot ca
nc
ell the
vehi
cl
e’s
noi
se
beca
use t
h
e
p
r
oces
s
need
bi
g L
val
u
e,
b
u
t
wi
t
h
t
w
o
p
r
oc
ess o
f
LM
S
A
d
apt
i
v
e
,
t
h
e
bi
gge
r
L
can be reache
d
,
the
r
efore
the noise
cancelling
c
oul
d be realized
well.
REFERE
NC
ES
[1]
Widrow, B., an
d S.D. Stearns,
“Adaptive Signal Pro
cessing”, 1
985, Prentice-H
a
ll,
Inc., Englewood Clifts, New
Je
rsey
.
[2]
S
carpinit
i
, M
.
, D
a
niel
, C., Raf
f
ael
e, P
., A
u
rel
i
o,
U, “
A
Collabara
tive Fil
t
er
Approach to Adaptive Noise
Cancellation”, S
m
art Innovation,
S
y
stems,
and
Technolog
ies, Volume 19, 2013
, p
p
101-109.
[3]
David A. Car
t
es, Laura R. Ray
,
and Robert D.Collier
,
“Ly
a
punov
Tuning
of Th
e Leak
y
LMS Algor
ithm For Single
S
ource”, S
i
ngl
e-
P
o
int-Nois
e Can
cel
lat
i
on, P
r
oce
e
d
ings
of
the Am
erican Control
Conferenc
e
Arlington, VA J
une
25-27, 2001
[4]
Sanaullah, Kh
an
, M. Arif, T. Ma
jeed
, “Comparison of LMS, RLS and Notc
h B
a
se
d Adaptiv
e Algo
rithm
s
for Noise
Cancellation o
f
a ty
p
i
cal Indus
tr
ial Workroom”, I
EEE, 2004
[5]
Akingbade KF, Isiaka AJA, “Separation of Di
gital Audio
Signal Using Leas
t-Mean-Square (LMS) Adaptiv
e
Algorithm”,
International Journal
of
Electrical and Computer
Engineering
(
I
JECE)
, Vol.4, No.4, August 2014,
pp.557-560, ISSN: 2088-8708
[6]
P
r
as
et
y
o
w
a
ti
, S
.
A.D, “
E
xplorati
on of The Adaptive Vehi
c
l
e’s Noise Cance
lla
tion
in The W
o
rk Room
”, Disertatio
n
,
Gadjah Mad
a
U
n
iversity
, Yog
y
akarta, Indon
esia, 2010
[7]
P
r
as
et
y
o
w
a
ti
, S
.
A
.
D
., Bus
t
anul
, A
., Eka N
.
B.S
.
,
“
S
olu
tion for
Vehicles Noise Cance
llation with Modification of
LMS Adaptive Algorithm
”
,
International Journal on Com
puter Science and Engineering
, Vol: 4 Issue: 5, Engg
Journal Publications, May
2012
Evaluation Warning : The document was created with Spire.PDF for Python.