Int
ern
at
i
onal
Journ
al of Ele
ctrical
an
d
Co
mput
er
En
gin
eeri
ng
(IJ
E
C
E)
Vo
l.
9
, No
.
5
,
Octo
ber
201
9
, pp.
4010
~4
019
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v9
i
5
.
pp4010
-
40
19
4010
Journ
al h
om
e
page
:
http:
//
ia
es
core
.c
om/
journa
ls
/i
ndex.
ph
p/IJECE
A n
ovel
and
inte
grated
a
rc
hit
ecture for
id
entific
ation and
c
an
ce
ll
ation of
n
oise fr
om GS
M si
gn
al
Rekha
N
1
,
Fa
t
hima
Jabeen
2
1
Depa
rtment of
El
e
ct
ron
i
cs
and
Com
m
unic
at
ion Engi
ne
eri
ng
,
K.
S.
Inst
it
u
te of
T
e
chnol
og
y
,
Indi
a
2
Islamiah
Inst
i
tu
te
of
T
ec
hnolog
y
,
Indi
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Des
20
, 201
8
Re
vised
A
pr
18
, 2
01
9
Accepte
d
Apr
28
, 201
9
The
re
are
m
ult
ip
le
re
asons
for
the
evol
ut
ion
as
w
el
l
as
th
e
pre
sen
ce
of
nois
e
over
tra
nsm
it
t
ed
GS
M
signal
.
In
spite
of
var
ious
appr
oac
h
es
towar
ds
noise
ca
nc
el
l
at
ion
t
echnique
s,
the
r
e
a
re
l
ess
appl
i
ca
b
l
e
t
ec
hniqu
es
for
cont
ro
ll
ing
noise
in
ac
ousti
c
GS
M
sign
al
.
The
ref
or
e,
th
e
proposed
m
anusc
ript
pre
sents
an
in
te
gra
te
d
m
odel
li
ng
which
per
form
s
m
odel
li
ng
of
noise
id
ent
ifica
ti
on
tha
t
coul
d
signif
ic
an
tly
assist
in
succ
essful
noise
ca
ncellation
.
T
he
proposed
s
y
stem
uses
thre
e
different
app
roa
ch
viz.
i)
sto
cha
sti
c
base
d
a
pproa
ch
for
noise
m
odel
li
ng
,
ii
)
ana
l
y
t
ical
-
b
a
sed
appr
oa
ch
wh
ere
al
lo
ca
t
ed
po
wer
acts
as
one
of
the
prom
ine
nt
fa
ct
ors
of
noise,
and
ii
i)
wave
let
-
base
d
a
pproa
ch
for
eff
ective
de
compos
it
ion
of
GS
M
signal
for
assist
ing
bet
te
r
noise
ca
nc
el
l
at
io
n
te
chn
iq
ue
fo
ll
o
wed
b
y
be
tt
e
r
retent
ion
of
s
igna
l
qualit
y
.
Sim
ula
te
d
in
MA
TL
AB,
the
stud
y
ou
tc
om
e
show
s
tha
t
it
offe
rs
a
co
st
-
eff
ectiv
e
implementa
t
ion,
A
Prac
ti
cal
Approac
h
for
Noise
ide
nti
f
icati
on
,
an
d
Eff
ec
t
ive
Noise
Canc
e
ll
a
tion
with
Signal
q
ual
ity
retent
ion
.
T
he
proposed
s
ystem
offe
rs
appr
oximatel
y
2
4%
of
enh
ance
m
ent
in
nois
e
re
duct
ion
as
comp
are
d
to
a
n
y
exi
sting
digital f
il
te
rs
with
1
.
6
se
conds
faste
r
in
p
roc
essing
spee
d
.
Ke
yw
or
d
s
:
Acousti
c
Denoisi
ng
Fil
te
r
GS
M
s
i
gn
al
Wh
it
e
n
oise
Copyright
©
201
9
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
:
Re
kh
a
N
,
Re
search
Sc
ho
la
r,
Dep
a
rtm
ent o
f
Elec
tro
nics
and
C
omm
un
i
cat
ion
E
nginee
rin
g
,
K.
S
. Instit
ute
of Tec
hnology,
B
eng
al
uru
, I
ndia
.
Em
a
il
:
rek
ha
phd2014@
gm
ail.co
m
1.
INTROD
U
CTION
Since
la
st
few
ye
ars,
the
G
SM
-
base
d
wi
r
el
ess
com
m
un
ic
at
ion
has
be
en
acco
un
te
d
for
em
erg
ing
grow
t
h
in
the
te
le
com
ind
us
t
ries.
The
GSM
was
introd
uc
ed
as
a
pro
gressi
on
of
sec
ond
-
gen
e
rati
on
cel
lular
te
chnolo
gy
sp
e
ci
fied
with
digi
ta
l
m
od
ulati
on
ser
vice
.
At
present,
t
he
devel
op
m
ent
of
th
e
GS
M
sta
nda
r
d
ha
s
reache
d
the
le
ve
l
of
m
eet
ing
da
il
y
need
s
of users
a
nd
e
nter
pri
ses
by p
r
ov
i
di
ng
c
os
t
-
e
ff
ect
i
ve
voic
e
se
rv
ic
es
as
well
as
eff
ic
ie
nt
data
ser
vices
w
hich
can
be
acce
s
sed
24
x7
ir
r
es
pecti
ve
of
us
er'
s
locat
io
n
[
1].
GS
M
te
c
hnology
s
upports
va
rio
us
feature
s
f
or
it
s
global
ac
ceptance
an
d
rich
popula
rity
[
2].
S
uc
h
feat
ures
ar
e
li
ke
it
has
eff
ic
ie
nt
sp
ect
ru
m
,
good
voic
e
qual
it
y
serv
ic
e
su
pport
s
low
-
cos
t
cel
lular
dev
ic
es,
com
patible
wit
h
IS
D
N
a
nd
ne
w
serv
ic
es
a
nd
pro
vid
es
ro
am
ing
ser
vices
gl
obal
ly
.
W
it
h
the
evo
l
ution
of
G
SM,
there
a
re
m
any
adv
a
nces
m
ade
in
di
gital
de
vices,
s
uc
h
as
per
s
onal
di
gital
assist
ants,
P
Cs,
m
ob
il
e
phon
e
s,
wireless
LA
N
s,
et
c
[3
]
.
T
hese
dev
ic
es
a
re
en
abled
with
the
su
pp
or
t
of
ce
ll
ular
com
m
un
ic
at
ion
m
od
ule
in
orde
r
to
de
li
ver
on
-
dem
and
ser
vices
a
nd
ente
rtai
nm
ent
in
va
rio
us
fiel
ds
of
ap
plica
ti
on
s
uch
as
sc
hools
,
office,
healt
hc
are,
trans
port,
Ind
ust
rial
area,
a
nd
m
any
m
or
e [
4]
.
In
a
cel
lular
com
m
un
ic
at
ion
syst
e
m
,
the
sp
eech
a
nd
data
inform
ation
transm
itted
via
a
rad
io
li
nk
com
m
un
ic
at
ion
c
hannel
wh
e
re
the
qual
it
y
of
t
ran
sm
it
te
d
data
s
uffer
s
from
m
any
deg
r
adati
on
fact
or
s
su
c
h
a
s
backg
rou
nd
noise
a
nd
c
hannel
inter
fere
nc
es
[
5].
T
he
‘t
erm
'
no
ise
a
nd
i
nterf
e
ren
c
e
basical
ly
re
f
ers
t
o
unwa
nted
destr
uctive
si
gn
al
s
intr
oduce
d
int
o
us
e
-
f
ull
sp
eec
h
an
d
data
sig
na
ls.
The
s
ources
of
no
ise
are
va
rie
d
in
natu
re
it
can
be
ge
ner
at
e
d
f
ro
m
an
env
i
ronm
ental
factor
su
c
h
as
aco
us
ti
c
disturba
nc
e
f
or
m
traff
ic
,
bl
owin
g
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A novel
and
i
ntegr
ated arc
hitec
ture for
ide
ntif
ic
ation
and ca
ncell
atio
n of noise f
rom
G
SM
sig
na
l (
Rek
ha
N
)
4011
the
en
gin
e
,
wi
nd,
lo
ud
m
us
ic
,
cr
ow
e
d,
et
c.
as
well
as
f
rom
m
echan
ic
al
syst
e
m
s
su
ch
a
s
quantiz
at
ion,
channel
interfe
ren
ce
,
hum
m
ing
,
a
nd
hand
off.
The
occurre
nces
of
noise
i
n
us
e
-
fu
ll
si
gn
al
s
af
fect
the
qual
it
y
an
d
intel
li
gib
il
it
y
of
t
he
s
peec
h
and
le
a
d
to
ca
us
e
cal
l
drop
factor,
instabil
it
y
in
vo
ic
e
and
data
ser
vic
es
and
m
aking
data
re
cepti
on
dif
ficu
lt
.
It
has
been
fou
nd
that
s
om
e
of
the
rese
arch
works
ha
ve
ad
dr
e
ssed
di
ff
ere
nt
so
urces
and le
vels
of
noise
in
which
so
m
e o
f
them
are
av
oid
able a
nd s
ome
are
t
hem
are
un
a
voida
ble [6
]
.
The
rese
arc
hers
hav
e
present
ed
va
rio
us
te
c
hn
i
qu
e
s
to
ha
ndle
su
c
h
ki
nds
of
prob
le
m
in
the
cel
lular
syst
e
m
.
The
a
vo
i
dab
le
ty
pe
no
ise
can
be
reco
gniz
e
d
and
el
im
inate
d
by
util
iz
ing
var
i
ou
s
a
nd
di
ff
ere
nt
te
chn
iq
ues
s
uc
h
as
s
peech
pr
ocessin
g
al
gori
thm
,
no
ise
filt
erin
g
a
nd
cl
ass
ific
at
ion
te
c
hn
i
qu
e
[
7
-
9].
Wh
ereas,
un
a
voida
ble
ty
pe
noise
(si
gn
a
l
-
fluctuati
on)
c
an
be
c
on
t
ro
ll
e
d
by
ba
ndwi
dth
ad
justm
ents
t
echn
i
qu
e
a
nd
s
ign
al
aver
a
ging
m
echan
ism
.
But
st
il
l,
it
is
ver
y
c
halle
ng
i
ng
f
or
the
resea
rc
her
s
an
d
the
pract
it
ion
ers
to
de
pl
oy
an
eff
ic
ie
nt
a
nd
r
obus
t
te
c
hn
i
que
to
m
eet
the
use
r
sat
isfact
io
n
an
d
qu
al
it
y
of
ex
per
ie
nce.
It
is
al
so
bee
n
se
en
tha
t
as
the
nu
m
bers
of
use
rs
inc
reases,
the
de
m
and
of
qual
it
y
of
vo
ic
e
a
nd
data
ser
vice
al
so
increase
s
and
therefo
re
due
to
the
co
ns
trai
nt
of
the
sp
ect
r
um
and
network
re
source
th
e
te
le
co
m
m
un
ic
at
ion
operat
ors
al
so
felt
a ch
al
le
nge
in
te
rm
s o
f
m
a
rk
et
c
om
petency and u
ser
e
xp
ect
at
ion
fo
rm
t
heir
se
r
vices [
10]
.
Ther
e
f
or
e,
ther
e
is
need
of
ef
fici
en
t
noise
re
m
ov
al
te
chn
iq
ue
f
or
m
the
re
searche
rs
i
n
or
der
t
o
m
eet
the
us
er
e
xpe
ct
at
ion
an
d
upcom
ing
dem
and
of
s
upply
becau
se
wire
le
ss
com
m
un
i
cat
ion
m
ark
et
is
sti
l
l
grow
i
ng
ve
ry
f
ast
wh
e
re
dif
fe
ren
t
wireless
-
c
omm
un
ic
at
ion
base
d
ap
plica
ti
on
s
will
re
quir
e
div
e
r
s
featu
r
e
an
d
char
act
e
risti
cs. I
n t
he fut
ur
e
, t
he 5G
will
intr
oduce as
a
new com
m
un
ic
at
ion
tech
nolo
gy, whic
h wil
l t
ransform
the
patte
r
n
of
existi
ng
c
omm
un
ic
at
ion
syst
e
m
s
in
m
any
adv
a
nced
a
pp
li
c
at
io
ns
are
a
(h
eal
thca
re
-
s
yst
e
m
,
real
-
ti
m
e
syste
m
s,
autom
at
i
on
i
ndust
ries,
et
c).
T
he
refo
re,
t
his
pa
pe
r
has
prese
nted
a
s
ol
ution
towa
r
ds
denoisin
g
of
th
e
noise
s
a
pp
ea
rin
g
in
GS
M
sign
al
.
Disc
us
si
on
of
the
li
te
ratur
e
has
bee
n
c
arr
ie
d
out
in
S
ect
io
n
1.1
fo
ll
owe
d
by
br
ie
f
highli
ghts
of
ide
ntifie
d
researc
h
pro
blem
s
in
Sect
io
n
1.2
.
Propos
ed
m
et
ho
dolo
gy
of
denoisin
g
GSM
sign
al
is
br
i
efed
i
n
Sect
io
n
1.3
.
Sect
io
n
2
il
lustrate
s
th
e
pro
po
s
ed
al
gorithm
desig
n
and
it
s
i
m
ple
m
entat
io
n
w
hile
res
ult
discuss
i
on
is
c
arr
ie
d
out
in
se
ct
ion
3.
Fi
nally
,
the
su
m
m
ary
of
the
pa
pe
r
is
giv
e
n
in Sec
ti
on
4.
This
par
t
of
t
he
stud
y
is
a
c
onti
nu
at
io
n
of
our
pr
i
or
re
view
[11].
T
he
wor
k
car
ried
out
by
Norholm
et
al
.
[
12
]
,
ha
ve
e
xp
l
or
e
d
noise
el
im
inati
on
iss
ue
i
n
the
tim
e
do
m
ai
n
and
use
d
co
va
riance
m
at
rices
a
nd
op
ti
m
al
filt
erin
g
ap
proac
h
for
sing
le
-
cha
nnel
no
ise
m
ini
m
izati
on
.
T
he
perf
or
m
ance
analy
sis
of
t
he
prese
nte
d
syst
e
m
is
com
par
e
d
with
the
W
ie
ne
r
filt
er
in
te
rm
s
of
S
N
R.
Be
rtra
nd
et
al
.
[
13
]
a
da
ptive
noise
m
ini
m
iz
at
ion
al
gori
thm
based
on
an
overla
p
-
a
dd
m
echan
ism
fo
r
s
peech
enh
a
ncem
ent
and
delay
reduc
ti
on
in
the
ac
ousti
c
wireless
sen
sor
syst
e
m
.
Sayo
ud
et
al
.
[14]
fo
c
us
es
on
the
prob
le
m
of
aco
us
ti
c
no
ise
a
nd
pr
e
sente
d
du
al
-
cha
nnel
le
ast
m
ean
sq
ua
re
(LMS
)
bas
ed
noise
re
du
c
ti
on
p
ro
ce
dure
fo
r
s
peec
h
en
han
cem
ent.
Th
e
study
ou
tc
om
es
resu
lt
in
the
superi
or
pe
rfor
m
ance
of
t
he
prese
nted
te
ch
nique
in
te
rm
s
of
SN
R,
MSE,
an
d
ce
pst
ral
distance
w
hen
com
par
ed
t
o
tr
aditi
on
al
sim
ilar
a
ppro
ac
hes
.
Si
m
il
arly
,
the
work
of
Ra
him
a
et
al
.
[15]
use
d
a
joint
a
ppr
oach
base
d
on
blind
sig
nal
separ
at
io
n
a
nd
adap
ti
ve
tra
ns
ve
rsal filt
erin
g
t
echn
i
qu
e
for
t
he
r
ed
uctio
n of
acoust
ic
n
oise
and sp
eec
h
e
nh
ancem
ent. Lehm
ann
et
al. [
16]
h
a
ve
d
esi
gne
d
a m
od
ifie
d ve
rsion
of
Ed
ge recei
ve
r
in
orde
r
to
r
edu
ce
co
-
ch
an
nel in
te
r
fer
e
nc
e eff
ect
for
the
ra
ndom
li
near
m
odulati
on
.
The
w
ork
of
Ham
ida
and
Am
ro
uc
he
[
17
]
ha
s
stu
died
the
pe
r
form
a
nce
of
Ech
o
ca
ncell
er
syst
e
m
in
pr
esence
of
nois
y
channel.
Sa
dok
et
al
.
[
18
]
hav
e
intr
oduc
ed
im
pr
oved
no
ise
cancel
le
r
syst
em
fo
r
si
ng
le
a
nten
na
inter
fere
nce
us
in
g
V
ol
te
rr
a
filt
ers
up
-
t
o
3
rd
or
der
.
The
st
ud
y
of
Viha
ri
et
al
.
[
19]
has
car
ried
a
n
as
sessm
ent
of
va
rio
us
noise
r
edu
ct
io
n
ap
proach
e
s
a
nd
pe
rfor
m
ed
pe
rfo
r
m
ance
analy
sis wit
h r
espect t
o dif
fe
r
ent noise cat
e
gory a
nd S
NR.
Kalam
ani
et
al.
[
20]
us
e
d
im
pro
ved
LMS
de
pende
nt
no
ise
m
ini
m
iz
ation
te
chn
iq
ue
f
or
s
peech
sig
nal
enh
a
ncem
ent.
Thro
ugh
pe
rfo
rm
ance
analy
sis
presente
d
m
e
chan
ism
found
to
be
ef
fecti
ve
in
te
rm
s
of
PSNR
and
MS
E.
U
pa
dh
ya
y
an
d
Ja
iswal
[21]
us
e
an
it
erati
ve
noise
assessm
e
nt
proce
dure
w
it
h
W
ie
ner
f
il
te
ring
appr
oach
f
or
e
nh
a
ncin
g
sin
gle
cha
nn
el
sp
ee
ch.
Pr
em
anand
a
an
d
Um
a
[2
2]
hav
e
intr
oduc
ed
a
ps
yc
ho
ac
ou
sti
c
-
base
d
gai
n
re
gu
la
to
r
for
ba
ckgr
ound
nois
e
rem
ov
al
an
d
sp
eec
h
si
gn
al
i
m
pr
ovem
ent
in
m
ob
il
e
te
le
phony
com
m
un
ic
at
ion
.
A
nothe
r
w
ork
car
r
ie
d
by
S
hukla
et
al
.
[23
]
hav
e
desi
gn
e
d
an
ec
ho
ca
nc
el
la
ti
on
syst
e
m
base
d
on th
reshold
f
il
te
r
f
or im
pr
ov
i
ng voice si
gn
al
in
a
ha
ndsf
ree
com
m
un
ic
at
ion
e
nv
i
ronm
ent.
The
wor
k
of
Afro
z
et
al
.
[
24
]
a
nd
G
upta
et
al
[25]
ha
ve
c
onduct
ed
a
com
par
at
iv
e
analy
sis
of
diff
e
re
nt
adap
t
ive
filt
er
fo
r
e
valuati
ng
it
s
per
f
or
m
ance
for
sp
eech
en
ha
nc
e
m
ent
in
te
r
m
s
of
PS
NR,
S
NR,
an
d
MSE.
Gb
a
da
m
os
i
et
al
.
[2
6]
desig
ne
d
sign
al
-
de
nois
ing
fr
am
ewor
k
based
on
Four
ie
r
tra
ns
f
or
m
and
non
-
pa
ram
et
ric
m
od
el
ing
f
or
no
ise
el
i
m
inati
on
in
G
SM
s
peec
h
signa
l.
Bi
tz
er
and
Ra
dem
acher
[27
]
hav
e
introd
uced
t
w
o
-
way
ap
proa
ch
i.e
fi
ng
e
r
pri
nt
m
echan
ism
fo
r
detect
ion
a
nd
inter
pola
ti
on
al
gorithm
s
fo
r
cancel
la
ti
on
of
bum
blebee
no
ise
.
Ma
hbub
an
d
Fatt
a
h
[28]
presente
d
gr
a
dient
a
nd
adap
ti
ve
LM
S
base
d
acoust
ic
ec
ho c
ancell
at
ion
sys
tem
f
or
voic
e s
ign
al
e
nh
a
nce
m
ent.
Wang
et
al
.
[
29]
an
d
C
hen
[
30]
ha
ve
al
s
o
c
on
ce
ntrate
d
on
bac
kgr
ound
noise
prob
le
m
and
prese
nted
an
inte
gr
at
e
d
sp
eec
h
e
nh
a
nc
e
m
ent
fr
am
ewo
r
k
base
d
on
du
al
m
ic
ro
ph
one
a
rr
ay
,
H
2
e
stim
at
or
,
an
d
sp
eec
h
m
od
el
i
ng
syst
e
m
for
pro
vid
in
g
a
cl
ean
sig
nal
in
a
m
ob
il
e
co
m
m
un
ic
at
ion
dev
ic
e.
Sh
a
kee
b
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
5
,
Oct
ober
20
19
:
4
0
1
0
-
4
0
1
9
4012
Sayi
dm
arie
[31
]
hav
e
pr
ese
nted
the
c
onfigurati
on
of
a
cel
lular
base
s
ta
ti
on
anten
na
fo
r
e
rr
at
ic
co
ver
a
ge
.
Ko
ll
em
et
a
l.
[32
]
ha
ve
il
lustrate
d
a
vie
wpoint
wh
ic
h
is
m
od
ifie
d
par
am
et
er
in
S
-
G
ra
dient
Hi
stogra
m
protect
ion
denoisin
g
te
chn
i
que.
A
wa
d
[
33
]
hav
e
dem
onstrat
ed
a
nove
l
app
r
oac
h
f
or
renov
at
in
g
im
ages
defor
m
ed
by
fi
xed
-
value
d
im
pu
lse
noise.
The
si
gn
i
ficant
r
esea
rch p
robl
e
m
s ar
e as
fo
ll
ow
s:
-
Existi
ng
re
sear
ch
te
c
hn
i
qu
es
are
m
or
e
f
oc
use
d
on
dev
el
op
i
ng
a
s
ophisti
ca
te
d
an
d
ne
w
fi
lt
er
w
hich
is
not
on
ly
e
xp
e
ns
i
ve
but also
ef
fecti
ve
f
or a s
pecif
ic
set of noises
.
-
Ther
e
a
re
fe
w
er
resea
rch
work
s
bein
g
car
ried
out
co
ns
ide
rin
g
pote
ntial
no
ise
prob
le
m
s
in
GS
M
sig
na
l
wh
ic
h
is c
har
a
ct
erized
by d
i
fferent
form
s o
f no
ise
s
.
-
The
i
nclusi
on
of
stoc
hastic
c
har
act
erist
ic
s
over
a
no
ise
is
s
om
et
hin
g
that
has
ne
ver
eve
r
bee
n
c
onside
r
ed
in the e
xisti
ng
syst
e
m
f
or
w
hich reaso
n
t
he
s
olu
ti
ons a
re
not app
li
ca
ble for
r
eal
-
ti
m
e app
li
cat
ion
.
-
Existi
ng
a
ppr
oa
ches
are
m
or
e
sp
eci
fic
to
ta
r
geted
no
ise
on
ly
wh
ereas
i
n
r
eal
it
y
there
are
var
i
ou
s
ty
pe
s
of
no
ise
s
that a
ff
e
ct
the GSM si
gnal
.
Ther
e
f
or
e,
the
pro
blem
sta
teme
nt
of
t
he
pro
posed
stu
dy
can
be
sta
te
d
as
“De
velo
pi
ng
a
noise
filt
e
rin
g
m
echan
ism
that
can
perform
rob
us
t
ide
ntific
at
ion
of
the
dy
nam
ic
beh
a
vi
or
of
no
ise
pre
sent
in
GS
M
s
ign
al
and offe
r
a
co
s
t
-
eff
ect
ive
s
olut
ion
to
m
itigate i
t”
.
The
propose
d
work
is
a
c
onti
nu
at
io
n
of
our
pri
or
im
ple
m
entat
ion
[3
4
,
3
5
]
wh
e
re
a
n
e
nh
a
ncem
ent
has
bee
n
car
rie
d
ou
t
by
c
on
st
r
ucting
an
integ
rated
m
od
el
in
orde
r
to
em
ph
asi
ze
on
the
no
ise
-
relat
ed
pro
blem
ov
e
r GSM si
gnal
. Th
e
over
al
l archite
ct
ure
of
the pr
opos
e
d
s
yst
e
m
is a
s f
ollow
s
in Fi
gure
1
.
N
o
i
s
e
c
a
n
c
e
l
l
a
t
i
o
n
u
s
i
n
g
S
t
o
c
h
a
s
t
i
c
A
p
p
r
o
a
c
h
W
a
v
e
l
e
t
-
b
a
s
e
d
D
e
n
o
i
s
i
n
g
T
e
c
h
n
i
q
u
e
D
a
t
a
S
e
g
m
e
n
t
a
t
i
o
n
L
o
c
a
l
/
g
l
o
b
a
l
o
p
t
i
m
a
m
a
t
r
i
x
N
e
w
d
i
g
i
t
a
l
l
o
w
p
a
s
s
f
i
l
t
e
r
H
e
n
k
e
l
m
a
t
r
i
x
R
e
p
r
e
s
e
n
t
o
r
W
a
v
e
P
o
w
e
r
f
a
c
t
o
r
(
g
o
o
d
/
b
a
d
)
N
o
i
s
e
I
d
e
n
t
i
f
i
c
a
t
i
o
n
D
e
c
o
m
p
o
s
i
t
i
o
n
o
f
G
S
M
s
i
g
n
a
l
(
w
a
v
e
l
e
t
)
T
h
r
e
s
h
o
l
d
i
n
g
C
o
m
p
r
e
h
e
n
s
i
v
e
D
e
t
e
c
t
i
o
n
/
c
a
n
c
e
l
l
a
t
i
o
n
o
f
N
o
i
s
e
I
n
v
e
r
s
i
n
g
T
r
a
n
s
f
o
r
m
a
t
i
o
n
Figure
1
.
Pro
pose
d
A
rc
hitec
ture
of
GS
M Si
gn
al
De
no
isi
ng
The
pri
m
e
des
ign
co
nce
pt
of
the
propose
d
syst
e
m
is
that
el
i
m
inati
on
of
no
ise
is
on
e
of
the
m
os
t
chall
eng
i
ng
ta
s
ks
es
pecial
ly
if
it
s
GS
M
au
di
o
sig
nal
as
t
here
is
al
ways
a
presence
of
no
is
e
on
the
pro
gr
e
ss
of
t
i
m
e
of
com
m
un
ic
at
ion
from
so
ur
ce
to
the
destinat
i
on
de
vice.
He
nce,
the
pro
pose
d
syst
em
offer
s
a
n
integrate
d
arc
hi
te
ct
ur
e that progressi
vely
r
ed
uces th
e
pr
ese
nc
e o
f diffe
re
nt ty
pes
of
noise
s
p
rese
nt in
the
GS
M
sign
al
.
Th
e
upper
bl
ock
of
t
he
arc
hitec
ture
i
nt
r
oduces
a
st
och
a
sti
c
m
od
el
ing
a
ppr
oac
h
wh
e
re
t
he
no
is
es
are
offer
e
d
a
sim
pl
e
featu
re
on
th
e
basis
of
local
an
d
global
op
t
i
m
a
in
orde
r
to
assist
s
in
ide
nt
ific
at
ion
.
T
he
log
i
c
beh
i
nd
t
his
is
if
the
identifi
cat
ion
proc
ess
is
recti
fied
than
ca
ncell
at
ion
proces
s
ha
s
to
pr
eci
se
a
nyhow
.
The
m
idd
le
bl
ock
of
ar
chite
ct
ur
e
us
es
a
novel
c
oncept
of
re
pr
ese
ntin
g
wav
e
w
her
e
the
G
SM
sig
nal
is
char
act
e
rized
with
res
pect
to
power
fact
or
com
pu
te
d
fro
m
i
ts
base
sta
t
ion
.
Hen
ce
the
no
ise
ide
ntific
at
ion
is
carried
out
with
resp
e
ct
t
o power
f
eat
ure.
The
final
/
bo
t
tom
blo
ck
of
the
arc
hitec
ture
i
m
ple
m
ents
wav
el
et
trans
f
or
m
at
ion
sche
m
e
as
it
can
deco
m
po
se
t
he
sign
al
with
ou
t
act
ually
aff
ect
ing
t
he
qu
al
it
y
of
the
si
gn
al
.
The
c
on
t
rib
ution
of
the
pro
pose
d
schem
e
is
that
the
uppe
r
la
ye
r
ad
dr
es
ses
r
andom
no
ise
,
m
idd
le
la
ye
r
addresses
Ga
ussi
an
noise
,
w
hi
le
the
bo
tt
om
la
ye
r
addresses
TD
MA
noise
.
T
he
refor
e
,
the
propose
d
syst
em
offer
s
t
he
en
han
ce
d
ca
pab
i
li
ty
of
identify
in
g
di
fferent
var
ia
nts
of
noise
in
G
S
M
aud
i
o
sig
nals
an
d
is
capa
bl
e
enou
gh
t
o
el
im
inate
it
at
the
en
d
of the
de
no
isi
ng
process
. T
he next
sect
ion di
scusses
alg
or
it
hm
i
m
ple
m
entat
ion
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A novel
and
i
ntegr
ated arc
hitec
ture for
ide
ntif
ic
ation
and ca
ncell
atio
n of noise f
rom
G
SM
sig
na
l (
Rek
ha
N
)
4013
2.
SY
STE
M DESIGN
The
c
ore
inte
nt
ion
of
the
pr
opose
d
syst
em
desig
n
is
to
ex
plore
the
e
xtent
of
possi
ble
no
ise
in
the
GS
M
si
gn
al
f
ol
lowed
by
a
s
uccess
fu
l
re
duct
ion
of
the
noise
without
di
storting
the
qual
it
y
of
the
r
ecei
ve
d
sign
al
.
2.1.
St
age
-
1:
N
oise ca
ncel
la
tio
n
using
st
ochast
ic
a
ppr
oach
This
is
the
first
sta
ge
of
pr
opose
d
m
od
el
design
that
em
ph
as
iz
es
the
gen
e
ra
ti
on
of
noise
f
ollow
e
d
by
a
reducti
on
of
the
no
ise
.
For
this
pur
po
se
,
a
Henkel
m
a
trix
is
con
str
ucted
wh
ic
h
act
s
as
a
reposit
or
y
f
or
GS
M
sign
al
s
an
d
se
gm
ented
dat
a.
The
segm
ented
data
is
then
cl
assifi
ed
into
no
isy
sig
nals
and
no
ise
-
f
ree
sign
al
s
.
The
no
ise
-
f
ree
sign
al
s
ar
e
th
en
tra
ns
f
or
m
ed
to
a
n
in
div
i
du
al
si
gn
al
ret
ai
nin
g
l
ocal
optim
a
m
at
rix
wh
ic
h
is
aggre
gated
f
or
const
ru
ct
in
g
gl
ob
al
opti
m
a
m
at
rix.
Finall
y,
the
no
ise
-
fr
ee
s
ign
al
is
app
li
e
d
with
the
stoc
hastic
base
d
a
ppro
ac
h wh
e
re t
he de
sign
e
d ne
w
lo
w pass
filt
er is
app
li
ed
for t
he
el
i
m
inati
on
of
the noise.
On
e
of
the
i
nteresti
ng
c
ontribu
ti
ons
of
this
op
e
rati
on
is
that
it
is
capab
le
of
ide
ntifyi
ng
the
resi
du
a
l
no
ise
e
ve
n
aft
er
noise
cance
ll
at
ion
op
e
rati
on
has
been
c
arr
ie
d
out.
Ac
cordin
g
to
this
con
ce
pt,
a
sto
chasti
c
process
is
i
ntr
oduce
d
in
t
he
pro
po
se
d
sys
tem
in
order
t
o
re
duce
the
art
ifact
s
of
th
e
no
isy
c
om
po
ne
nts.
The
pro
pose
d
syst
e
m
a
lso
app
li
es
a
dig
it
al
filt
er
in
or
de
r
to
carry
out
nor
m
al
iz
ation
al
ong
with
sm
oo
theni
ng
op
e
rati
on
ove
r
the
data
with
a
n
ai
d
of
the
sta
ti
sti
c
at
tribu
te
.
Fo
r
this
purpos
e,
the
pro
po
se
d
syst
em
m
akes
use
of
the
di
gital
fi
lt
ers
that
are
also
de
pende
nt
on
the
siz
e
of
m
ulti
ple
fr
am
es
al
ong
w
it
h
de
grees
of
po
ly
no
m
ial
functi
on. T
he
prop
os
ed
syst
em
, th
erefore, u
ses a
un
i
qu
e
no
ise
cancell
at
ion
tech
nique
as
sh
ow
n belo
w:
Algori
th
m
for
Noise C
ance
ll
at
i
on
Inpu
t
: A
(Aud
io File
)
Out
p
ut
:
A
den
(
denoise
d
a
ud
i
o fil
e)
St
ar
t
1.
i
nit
A
2.
A
f
1
(
x
, F
s
)
3.
N
oise
2*ra
nd(
x)
-
1
4.
Sta
te
Sele
ct
edFilt
er Case
5.
If
(c
onditi
on=t
ru
e)
6.
H
d
ge
n
lp
f
(F
pass
)|gen
hp
f
(F
pass
)|ge
n
bpf
(F
pass
))
7.
En
d
8.
Sta
te
Sele
ct
edFilt
er Case
9.
A
den
f
2
(H
d
)
End
Algorithm
Operati
on
:
T
his
al
gorithm
ta
kes
the
input
of
GS
M
file
w
hich
is
in
the
w
ave
f
or
m
at
A
(Line
-
1).
A
s
pe
ci
fic
functi
on
f
1
(x
)
is
c
onstr
uc
te
d.
T
his
f
ur
t
her
process
the
input
of
the
G
SM
sign
al
i
n
orde
r
to
ob
ta
in
a
s
am
pl
ed
data
x
a
nd
sam
ple
rate
f
s
(Line
-
2)
f
ollo
wed
by
co
ns
tr
uction
of
hypo
theti
cal
ran
do
m
no
ise
ov
e
r
t
he
sam
pled
data
(Line
-
3).
A
functi
on
ran
d
()
is
appl
ie
d
ov
er
sam
pl
ed
data
x
in
order
to
ge
ner
at
e
this
rand
om
no
ise
.
The
ne
xt
pa
rt
of
the
al
gorith
m
is
about
a
pply
ing
di
ff
e
ren
t
form
s
of
the
fil
te
r
in
t
he
form
of
use
case sel
ect
ed fil
te
r
(Line
-
4)
t
ha
t offers
the
use
r
sel
ect
a m
echan
ism
to
ap
pl
y
m
ulti
ple co
ndit
ion
s
of the
f
i
lt
er.
In
t
he
case
of
the
low
-
pass
fi
lt
er
,
the
syst
em
check
s
if
th
e
low
pass
fr
e
qu
e
ncy
F
pass
is
m
or
e
than
0
first.
I
f
the
F
pass
value
is
fou
nd
t
o
be
l
ess
than
(F
s
/2
-
30)
tha
n
sta
rt
an
d
sto
p
f
requen
cy
is
c
onfi
gure
d.
This
op
e
rati
on
is
f
ollow
e
d
by
ap
plyi
ng
a
dis
crete
f
un
ct
i
on
gen
lp
f
(x)
c
onsideri
ng
in
pu
t
a
r
gu
m
ents
as
sta
rt
an
d
stop
fr
e
qu
e
ncy
and
sam
ple
rate
F
s
(Line
-
6).
A
sim
il
ar
m
eth
od
is
al
so
a
ppli
ed
f
or
high
-
pa
s
s
fi
lt
er
wh
e
re
the
conditi
on
ap
pl
ic
able
f
or
it
is
if
the
value
of
F
pass
is
le
ss
than
(F
s
/2
-
30)
f
ollo
wed
by
ap
plyi
ng
a
s
pecific
functi
on
ge
n
hp
f
(x)
w
hile
the
sam
e
pr
ocess
is
al
so
car
ried
ou
t
t
o
co
ns
tr
uct
a
ba
nd
-
pa
ss
filt
er
us
i
ng
gen
bp
f
(Line
-
6).
Final
ly
, th
e pr
ocess of
in
ver
si
on is
init
ia
te
d
in
ord
er to ge
ner
at
e
a
noisy
sig
nal as
foll
ow
s:
A
noise
=
m
o
(
1/10po
w
(
SN
R/
10))
*(N
oise
-
m
ean) *st
d_de
v(x)
/
std
_d
e
v (Nois
e)
In
t
he
a
bove
e
xpressi
on,
the v
aria
ble
N
oise is
const
ru
ct
e
d
by
a
pply
ing
a
r
especti
ve
filt
er
over
H
d
a
nd
Noise
com
po
ne
nt
w
hile
the
sta
nd
a
rd
-
dev
ia
ti
on
of
sam
ple
d
data
an
d
no
i
se
is
al
so
us
e
d
for
co
ns
tr
ucting
t
he
no
isy
sig
nal.
The
final
ste
p
is
to
app
ly
ano
t
her
f
unct
ion
f
2
(x)
to
the
no
is
y
sign
al
in
or
de
r
to
ob
ta
in
de
no
is
e
d
sign
al
A
den
(Li
ne
-
9). T
he
esse
ntial
co
nt
rib
ution o
f
this al
gor
it
h
m
is that i
t assi
sts i
n
ide
ntif
yi
ng
the
prese
nc
e o
f
rand
om
no
ise
s
and
it
is
al
so
c
apab
le
of
m
inim
iz
ing
the
le
ve
l
of
no
ise
.
A
nothe
r
sig
nifica
nt
co
ntributi
on
of
t
his
al
gorithm
is
th
at
it
of
fer
s
a
on
e
-
wind
ow
f
ra
m
ewo
r
k
that
is
capab
le
of
nor
m
al
iz
ing
any
fo
rm
of
sp
eech/
aud
i
o
file
in
the
GSM
sign
al
with
custom
iz
ed
low
fr
e
quency
a
s
well
as
it
a
lso
offe
rs
the
ca
pab
il
it
y
to
cancel
no
ise
against
any
s
pe
ci
fic
sign
al
-
to
-
noise
rati
o
val
ue.
He
nce,
a
si
m
pl
i
fied
stoch
ast
ic
no
ise
can
cel
la
ti
on
approac
h
is
i
m
ple
m
ented
in this
stage
of
researc
h w
ork.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
5
,
Oct
ober
20
19
:
4
0
1
0
-
4
0
1
9
4014
2.2.
St
age
-
2:
C
om
prehensi
ve
de
tection/
cancell
at
i
on
of n
oise
This
sta
ge
of
w
ork
is
ba
sic
al
ly
an
exte
nsi
on
of
t
he
pri
or
sta
ge
w
he
re
a
ra
ndom
wav
e
f
or
m
i
s
gen
e
rated
in
orde
r
to
re
pr
e
s
ent
the
de
noise
d
si
gn
al
obta
ined
from
the
pri
or
ste
p.
It
i
m
pr
oves
t
he
pri
or
op
e
rati
on
by
con
si
der
i
ng
the
powe
r
fact
or
a
sso
ci
at
ed
with
the
G
SM
sig
na
l
that
is
ge
nerat
ed
by
t
he
m
ovi
ng
veh
ic
le
.
This
pa
rt
of
t
he
im
pl
e
m
entat
ion
c
o
ns
ide
rs
t
he
power
facto
r
a
ss
ociat
ed
with
good
GS
M
sig
na
l
al
ong
with
hi
gh
e
r
fe
asi
ble
powe
r
s
cor
e
dem
and
.
The
pr
opos
e
d
syst
e
m
m
akes
us
e
of
tim
e
as
well
as
fr
eq
ue
ncy
-
base
d
trans
for
m
at
ion
s
syst
em
in
or
der
to
i
m
ple
m
ent
the
power
fact
ors.
Finall
y,
a
rev
ise
d
al
go
rithm
is
i
m
ple
m
ented
in
order t
o pe
rfor
m
d
et
ect
ion
of noise
whose
sig
nificant ste
ps
a
re as
d
isc
usse
d belo
w:
Algori
th
m
for
o
b
ta
ini
ng a
t
r
an
sie
n
t
si
gn
al
Inp
ut:
s
(s
ourc
e nod
e
),
f
3
(funct
ion
for
a
dd
i
ng
no
ise
),
A
den
(
sp
eec
h
sig
nal)
Ou
t
pu
t:
A
t
(t
ra
ns
ie
nt si
gn
al
)
Start
1.
s
f
3
(A
den
)
2.
If α<
T
3.
flag
a
;
4.
Else
5.
flag
b
;
6.
A
t
f
4
(
σ
)
En
d
Algorithm
Op
e
rati
on
:
The
al
gorithm
sta
rts
with
the
baseli
ne
of
co
ns
tr
uctin
g
a
un
i
qu
e
at
tri
bu
te
cal
le
d
as
repre
sent
wh
ic
h
is
a
ty
pical
am
ou
nt
of
GS
M
sig
na
ls
colle
ct
ed
f
r
om
the
so
urce
node
.
T
he
e
m
pirica
l
com
pu
ta
ti
on
of
the
re
present
can
be
car
rie
d
ou
t
by
di
vidi
ng
the
powe
r
qua
ntit
y
of
good
sig
nal
by
powe
r
qu
a
ntit
y
of
noisy
sign
al
.
H
owever,
f
or
bette
r
pr
eci
si
on
i
n
i
den
ti
fyi
ng
t
he
GS
M
sig
nals,
the
propose
d
sy
stem
rev
ise
d
the
po
wer
fact
or
for
a
no
isy
signa
l
with
m
axi
m
u
m
po
we
r
at
a
reg
ula
r
tim
e
int
erv
al
.
Af
te
r
obta
inin
g
the
re
pr
ese
ntator
patte
rn,
the
seco
nd
-
or
der
at
tribu
te
s
are
extracte
d
t
oo.
The
pro
po
s
e
d
syst
em
con
sider
s
pr
im
ary
and
se
conda
ry
at
tribut
e
as
the
m
ini
m
al
value
of
th
e
diff
e
re
nt
va
riants
of
the
GSM
sign
al
a
nd
c
orrup
t
sign
al
.
T
he
pr
opos
e
d
syst
em
carries
out
the
m
ini
m
a
est
i
mati
on
with
a
n
a
id
of
a
th
res
holded
value
w
he
re
al
l
the num
erical
v
al
ues
lesser t
ha
n
th
res
ho
l
d
ed
value
is
obtai
ne
d.
The
e
valuati
on
of
the
pr
opos
e
d
al
go
rithm
is
carried
ou
t
c
on
siderin
g
a
f
re
quency
of
2.4
G
Hz
a
nd
this
config
ur
at
io
n
represe
nts
a
window
of
1m
s
slots
of
tim
e.
The
pr
opose
d
al
gorith
m
e
m
ph
asi
zes
ov
er
encapsulat
in
g
the
qua
ntit
y
of
the
tran
sie
nt
G
SM
sign
al
that
has
the
lo
wes
t
du
rati
on
due
to
the
f
reque
nc
y
of
gr
a
nula
r
sam
pling
facto
r.
Th
e
im
ple
m
entat
i
on
of
t
he
al
gorithm
is
carried
ou
t
co
ns
ide
r
20ns
of
f
requ
ency
.
The
im
plica
ti
o
n
of
t
he
al
gorit
hm
resu
lt
s
in
i
niti
at
ing
com
m
un
ic
at
ion
bet
ween
the
s
our
ce
de
vices
in
orde
r
to
pr
eci
sel
y
c
onfigure
t
he
GS
M
sign
al
.
T
he
pro
po
s
ed
stu
dy
m
akes
us
e
of
a
ddit
ive
wh
it
e
G
aussian
no
ise
f
or
th
e
pur
po
se
of
in
c
lud
in
g
the
tra
nsi
ent
no
ise
in
t
he
GS
M
si
gn
a
l
that
is
con
ti
nu
ed
by
com
puti
ng
al
l
so
rts
of
the
error
rates.
T
he
syst
em
al
so
carries
out
the
identific
at
ion
of
a
cl
ass
of
the
noise
in
order
to
perf
or
m
ind
exi
ng
of the si
gn
al
t
ha
t i
s ex
tract
e
d.
The
pro
posed
syst
e
m
per
f
orm
s
an
evalua
ti
on
of
t
he
no
ise
co
ntent
within
the
du
rati
on
of
the
1
m
illi
secon
d
th
at
is
f
ur
t
her
f
oll
ow
e
d
by
dr
a
fting
a
c
onditi
onal
sta
tem
ent
for
bo
t
h
th
e
cat
e
gory
of
noise
f
or
th
e
pur
po
se
of
pe
rfor
m
ing
the
s
up
erior
f
or
m
ind
e
xing
of
th
e
m
e
asur
em
ent
signa
ls
associat
ed
with
the
GS
M
no
is
e
connecte
d wit
h t
he
tra
ns
ie
nt
noise
. T
he
i
niti
al
co
ndit
ion
al
sta
tem
ent
of
fe
rs a su
gg
e
sti
on th
at
in
case t
he
ra
te
of
error
is
fou
nd
to
be
lowe
r
th
an
the
thres
holded
value
tha
n
it
is
assum
ed
to
be
a
su
pe
ri
or
f
or
m
of
the
GS
M
sign
al
.
O
n
t
he
oth
e
r
hand,
if
t
he
rate
of
e
rro
r
e
is
fou
nd
to
be
higher
tha
n
the
th
res
ho
l
de
d
value
tha
n
i
t
is
sti
ll
consi
der
e
d
as
an
erro
r
-
pro
ne
sign
al
.
T
he
rat
e
of
er
ror
as
a
sp
eci
fic
f
or
m
of
the
rate
that
ha
s
su
r
faced
ow
ing
to
the
oth
e
r
ty
pes
of
er
rors
tha
t
are
cor
recte
d.
On
ce
the
at
tr
ibu
te
of
the
m
ini
m
a,
as
well
as
a
su
m
m
a
tio
n
of
at
tribu
te
s
of
second
m
ini
m
a
,
are
ob
ta
i
ned
via
no
rm
al
un
it
evaluati
on,
the
pro
pos
ed
syst
e
m
ultim
at
ely
evaluates al
l t
he
se v
al
ues.
Finall
y,
the
al
gorithm
gen
erates
a
transient
GS
M
sig
nal
us
in
g
f
unct
ion
f
4
(x)
tha
t
us
es
input
argum
ents
as
t
he
durati
on
of
tim
e,
a
m
plit
ud
e,
un
it
ste
p
fun
ct
ion
,
rin
se
tim
e,
fr
e
qu
e
ncy.
On
e
of
the
inte
resti
ng
us
a
ges
of
t
his
al
gorithm
is
it
s
threshol
d
value
T
,
w
hich
ca
n
be
al
te
red
a
s
per
dif
fer
e
nt
spe
ech
-
base
d
app
li
cat
io
n.
T
he
refor
e
,
this
al
gorithm
can
be
su
it
ably
us
ed
for
ide
ntifyi
ng
dif
fer
e
nt
ty
pe
s
an
d
le
ve
l
s
of
noise
in
the
G
SM
si
gn
al
an
d
it
ca
n
al
s
o
diff
e
re
nt
ia
te
good
sig
nal
f
r
om
the
ba
d
si
gn
al
.
T
he
pro
posed
stu
dy
al
so
assum
es
that
t
her
e
is
two
dis
ti
ng
uis
he
d
for
m
of
GS
M
sig
nal
of
tra
ns
ie
nt
char
act
erist
ic
s
that
are
con
si
der
e
d
to
be
e
xtracted
from
the
fast
-
mo
ving
veh
ic
le
.
E
xactl
y,
a
sim
il
a
r
f
or
m
of
t
he
ti
m
e
interval
is
us
e
d
that
is
re
porte
d
to
be
us
e
d
in
GS
M
sign
al
as
the
or
ie
ntat
ion
ba
se
is
con
side
red
for
the
pro
posed
s
tud
y.
T
he
pro
po
s
ed
al
gorithm
al
so
carries
ou
t
a
n
equ
i
valent
co
m
pu
ta
ti
on
al
operati
on
t
hat
is
al
so
re
ported
t
o
be
a
ppli
ed
towa
r
ds
the
de
viati
on
of
the
sig
nals
to
noise
rati
o.
T
he
refor
e
,
the
pr
opos
e
d
syst
em
is
capa
ble
to
i
den
ti
fy
a
nd
el
im
inate
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A novel
and
i
ntegr
ated arc
hitec
ture for
ide
ntif
ic
ation
and ca
ncell
atio
n of noise f
rom
G
SM
sig
na
l (
Rek
ha
N
)
4015
the
no
ise
el
em
ents
f
r
om
the
ov
e
rall
co
nte
nt
s
of
the
nois
y
GS
M
si
gnal
i
n
a
m
uch
c
ost
-
eff
ect
iv
e
m
ann
e
r.
The
al
gorithm
is ap
plica
ble fo
r
ide
ntifyi
ng a
nd elim
inati
ng
no
ise
s
fro
m
all G
SM
sig
nals s
ta
ti
c o
r
tra
ns
ie
nt.
2.3.
St
age
-
3:
Wav
el
et
-
b
as
ed
den
oising
t
ech
niq
ue
T
he
pr
im
e
obje
ct
ive
of
this
sta
ge
of
im
ple
m
entat
ion
is
a
pp
li
ed
to
a
wa
velet
-
base
d
tra
ns
f
or
m
at
ion
te
chn
iq
ue
f
or
a
ssist
ing
i
n
the
el
i
m
inati
on
of
no
ise
f
ro
m
on
e
dim
ension
al
GS
M
si
gn
al
t
o
rec
on
st
ru
ct
e
d
sign
al
fr
ee
f
ro
m
no
is
e.
It
is
a
fact
t
hat
there
is
a
presen
ce
of
real
-
tim
e
no
ise
over
al
l
the
ran
ge
s
of
f
re
qu
e
nci
es
an
d
hen
ce
it
is
qu
i
te
chall
en
ging
to
el
im
inate
th
e
noise
from
t
he
GS
M
sig
na
l.
O
ne
of
the
adv
a
ntage
s
of
us
in
g
wav
el
et
-
base
d
trans
form
ation
schem
e
is
tha
t
it
is
capa
ble
of
a
ddressi
ng
de
no
isi
ng
pro
blem
s
of
al
m
os
t
al
l
ty
pes
of
noise
s
e.g.
el
ect
r
onic
noise
,
el
ect
r
om
agn
et
ic
noise
,
ac
ou
sti
c
noise
,
a
nd
el
ect
ro
st
at
ic
no
ise
.
I
n
orde
r
t
o
perform
den
oi
sing
on
GS
M
aud
i
o
sig
nal,
t
he
pr
opos
e
d
al
gorithm
is
su
bject
ed
t
o
serie
s
of
m
echan
is
m
viz.
deco
m
po
sit
io
n
of
GS
M
si
gn
al
,
a
pp
ly
in
g
dif
fer
e
nt
va
riants
of
thr
esh
old
in
g
m
e
chan
ism
,
and
finall
y
reconstr
uctio
n
of
t
he
or
i
gin
al
sign
al
by
an
el
i
m
inati
on
of
noise
.
T
he
flo
w
of
the
al
gorith
m
is
to
co
ns
id
er
the
or
i
gin
al
sig
nal
wh
ic
h
is
f
ur
t
he
r
co
rru
pted
by
sta
nd
a
rd
w
hite
no
ise
fo
ll
o
we
d
by
deco
m
posit
ion
te
ch
n
iq
ue
and
thres
ho
l
ding.
The
rec
onstruc
ti
on
is
ob
ta
ined
afte
r
pe
rfor
m
ing
an
inv
e
rting
op
e
rati
on
on
wa
velet
trans
form
ation
.
Th
e
sig
nifican
t st
eps of t
he p
rop
os
ed
alg
or
it
hm
are
as foll
ows:
Algori
th
m
for
wav
el
e
t
-
b
as
e
d d
en
oising
Inpu
t
:
A
t
(tra
nsi
ent audio
sig
nal)
Out
p
ut
:
A
densig
(d
e
noise
d si
gn
al
)
St
ar
t
1.
[t
sig
F
s
n
b
]
f
1
(A
t
)
2.
t
sigN
no
ise
(
a
m
p*
t
sig
, SNR)
3.
Sele
ct
f
4
(x)
wt
4.
[a
1
a
2
a
3
a
4
]
f
5
(w
t
)
5.
[b
1
b
2
]
f
6
(t
s
igN
,
le
v
, a
1
, a
2
)
6.
[
c
]
f
7
(b
1
, b
2
,w
t,
le
v
)
7.
[d
1
d
2
d
3
]
f
8
(b
1
,
b
2
)
8.
[E
1
,
E
2
, E
3
, E
4
]
f
9
(c, b
1
, b
2
, a
3
, a
4
,
n
)
9.
Sele
ct
t
selRule
10.
Th
1
th
Sel
(E
2
, E
3
, E
4
, [t
selRule
])
11.
th
n
f
10
(E,
th
op
, th
1
)
12.
A
densig
E
1
+t
h
n
13.
err
a
rg
m
ax
(|t
signN
-
A
densig
|)
End
Algorithm
Operati
on
:
T
his
above
-
sho
wn
ste
ps
of
al
gori
thm
per
fo
rm
t
he
ap
plica
ti
on
of
disc
rete
wav
el
et
tra
nsf
or
m
s
in
orde
r
t
o
perform
denoisin
g
ope
rati
on
on
t
he
GS
M
sp
eec
h
sig
nal.
Si
m
il
ar
fu
nctio
n
f
1
(
x)
is
app
li
ed
to
the
transie
nt
au
dio
sig
nal
A
t
(
Line
-
1)
in
or
de
r
to
ob
ta
i
n
true
sig
nal
t
sig
,
sa
m
ple
rate
F
s
,
and
a
nu
m
ber
of
bits
n
b
.
Th
e
nex
t
st
ep
of
this
al
gor
it
h
m
is
to
ad
d
up
an
a
dd
it
ive
wh
it
e
Gaussi
an
no
ise
ove
r
th
e
tru
e
sign
al
that
is
ob
ta
ine
d
f
ro
m
pro
du
ct
of
init
ia
li
zed
a
m
plit
u
de
amp
,
sign
al
-
to
-
noise
rati
o
SN
R,
an
d
t
sig
valu
e
(Lin
e
-
2).
He
nc
e,
a
noisy
sig
na
l
t
sigN
is
no
w
ob
ta
ine
d.
T
he
nex
t
process
is
to
ap
ply
wa
ve
le
t
deco
m
po
s
it
ion
us
in
g
a
discret
e
functi
on
f
4
(
x),
wh
ic
h
can
be
any
form
of
transfo
rm
ation
in
orde
r
to
obta
in
differe
nt
for
m
s
of
wav
el
et
s
wt
(Line
-
4).
This
process
is
f
ur
t
h
er
f
ollo
wed
up
by
app
ly
in
g
a
functi
on
f
5
(x)
w
hich
pe
rfor
m
s
a
filt
ering
proce
ss
for
the
gi
ve
n
wa
ve
an
d
le
ads
to
the
ge
ner
at
io
n
of
f
our
dif
fe
ren
t
f
or
m
s
of
coe
ff
i
ci
ents
deco
m
po
se
d
l
ow a
nd
high
pas
s f
il
te
r
(a
1
a
nd
a
2
)
w
hile rec
onstructed
lo
w
a
nd
high p
a
ss f
il
t
er (
a
3
a
nd a
4
) (
Line
-
4).
Deco
m
po
si
ti
on
of
the
wa
velet
is
carried
out
us
in
g
functi
on
f
6
(
x)
on
no
isy
sign
al
t
sigN
and
deco
m
po
s
ed
coeffic
ie
nts
of
filt
ers
a
1
an
d a
2
with
res
pect to
the
giv
e
n ran
ge
of level
of de
com
po
sit
ion
le
v
(Line
-
5)
.
This
operati
on
le
ads
to
the
ge
ner
at
io
n
of
dec
om
po
sit
ion
an
d
book
keep
i
ng
vector
b
1
and
b
2
(Line
-
5)
.
Appro
xim
ation
of
the
c
oeffic
ie
nt
is
then
ca
rr
ie
d
out
us
in
g
f
unct
ion
f
7
(x)
ove
r
b
1
an
d
b
2
as
well
as
with
wav
el
et
wt
with
res
pect
to
the
def
ine
d
le
vel o
f
deco
m
po
sit
ion
le
v
(Li
ne
-
6).
Extracti
on
of the
detai
l
coef
f
ic
ie
nt
is t
hen
carr
ie
d
ou
t usi
ng a f
un
ct
ion
w
it
h
res
pe
ct
to
b
1
and
b
2
in o
r
der
to
ge
ne
rate
m
ulti
ple
coeffic
ie
nts
c
(
Line
-
7).
Finall
y,
the
reco
nst
ru
ct
i
on
pr
oce
ss
is
carr
ie
d
ou
t
over
b
1
,
b
2
,
an
d
recon
stru
ct
ed
c
oeffici
ent
filt
er
a
3
a
nd
a
4
us
in
g
functi
on
f
9
(x
)
.
T
he
propose
d
syst
em
also
a
pp
li
es
a
di
screte
sel
ect
io
n
of
t
hr
es
hold
strat
egy
t
selRule
wh
ic
h
is
app
li
ed
over
E
2
,
E
3
,
a
nd
E
4
(all
reconstr
uc
te
d
coe
ff
ic
ie
nt
)
(Line
-
10
-
11
).
Fi
nally
,
the
pro
po
se
d
al
gor
it
h
m
app
li
es
tw
o
disti
nct
form
s
of
thre
sholdi
ng
i.e.
ha
rd
a
nd
so
ft
th
res
ho
l
din
g
us
i
ng
funct
ion
f
10
(
x)
(Lin
e
-
11)
.
Althou
gh,
ha
r
d
thres
ho
l
ding
is
one
of
the
ea
sie
st
m
et
ho
ds
of
im
ple
m
entation
for
co
ntr
ol
li
ng
th
reshold
factor
sti
ll
the
pr
op
ose
d
syst
e
m
adv
ocates
the
us
e
of
s
of
t
thres
ho
lding
m
echan
is
m
as
it
of
fer
s
bette
r
com
patib
il
it
ies
with
m
at
he
m
a
ti
cal
pr
ope
rtie
s.
T
he
ac
qu
isi
ti
on
of
de
noise
d
si
gn
al
A
den
sig
is
ob
ta
i
ned
by
s
umm
ing
up
a
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
5
,
Oct
ober
20
19
:
4
0
1
0
-
4
0
1
9
4016
reconstr
ucted
coeffic
ie
nt
filt
er
at
the
3
rd
l
evel
of
ap
pro
xim
a
ti
on
an
d
the
ne
wly
ob
t
ai
ned
th
res
ho
l
d
val
ue
(Line
-
12)
.
T
he
propose
d
syst
e
m
al
so
com
pu
te
s
the
error
va
lue
err
by
obta
ining
t
he
dif
fe
r
ence
of
noisy
s
ign
al
t
sigN
with d
e
no
i
sed
si
gn
al
A
densig
(Line
-
13
).
3.
RESU
LT
A
N
ALYSIS
The
im
ple
m
entat
ion
of
the
pro
posed
st
ud
y
has
bee
n
a
sses
sed
over
norm
al
windows
syst
e
m
wh
ere
MATLAB
is
use
d
f
or
scri
ptin
g
the
al
gorith
m
.
A
cal
l
reco
rd
in
g
ap
p
was
us
e
d
for
extrac
ti
ng
the
voic
e
sign
al
s
ov
e
r
the
4G
ne
twork
w
ho
se
siz
e
is
60
kilo
byte
s
an
d
bitra
te
is
256
kil
obyt
e
per
sec
onds.
The
overall
GSM
sign
al
co
ntent
of
au
dio
is
a
rou
nd
3
seco
nds,
wh
ic
h
is
ba
sic
al
ly
of
wav
e
file
form
at
wh
ic
h
ca
n
be
easi
ly
processe
d by
MATLAB
u
si
ng it
s in
buil
t fun
ct
ion
.
Fig
ure
2(a)
highli
gh
ts t
he
s
pe
ct
ru
m
o
f
the o
r
iginal GSM sig
nal, which
w
he
n
ap
plied to
al
gorithm
-
1
yi
el
ds
an
ou
tc
om
e
of
the
deno
ise
d
i
m
age.
Th
e
sp
ect
ro
m
et
er
of
the
G
SM
sign
al
co
rru
pted
with
noise
is
sh
ow
n
in
Fig
ur
e
2(b
)
,
w
hile
the
de
no
ise
d
si
gn
al
in
the
form
of
th
e
s
pectr
om
et
er
is
sh
ow
cased
in
Fig
ure
2(
c
).
The
sp
ect
ro
m
et
er
visu
al
iz
at
io
n
ex
hib
it
s
that
m
axi
m
u
m
i.e.
85%
of
no
ise
s
are
act
ually
rem
ov
ed
in
this
ph
a
se
it
sel
f.
This
ou
tpu
t
of
the
de
no
ise
d
sig
nal
wh
e
n
s
ubj
ect
e
d
to
the
al
gor
it
h
m
-
2
yi
el
ds
furthe
r
bette
r
no
ise
reducti
on
perf
or
m
ance.
Algo
rithm
-
1
was
focuse
d
m
or
e
on
ran
dom
no
ise
el
i
m
inati
on
w
hile
al
go
rithm
-
2
was
m
or
e
fo
c
us
e
d
on
the
el
i
m
inatio
n
of
wh
it
e
G
aussian
noise
.
This
ca
n
be
se
en
in
the
outc
om
e
of
Fig
ur
e
3
w
hic
h
is
basical
ly
a
scat
te
red
pl
ot
to
s
how
that
t
he
de
ns
it
y
of
t
he
no
ise
-
relat
e
d
com
po
ne
nts
i
s
sign
i
ficantl
y
getti
ng
reduce
d
with t
he
pr
ogress of t
he
su
m
o
f
m
ini
m
a. Con
side
ring
a t
hr
es
holde
d
val
ue
of
1m
s
d
urat
io
n,
the
s
econd
al
gorithm
of
fer
arou
nd
92%
of
el
i
m
inati
on
of
the
noisy
co
m
po
nen
ts
from
the
G
SM
sig
n
al
s.
A
par
t
fro
m
this,
the
processi
ng
o
f
the p
r
opos
e
d
al
gorithm
is
al
so
fou
nd
t
o
be
yi
el
din
g
a
res
ult
in
m
uch
fas
te
r
track
irres
pe
ct
ive
of any
form
s o
f
c
ut
-
off
v
al
ues
b
ei
ng
us
e
d
i
n
t
he
a
naly
sis.
(a)
(b)
(c)
Figure
2
.
Vis
ua
l
ou
tc
om
es o
f st
age
-
1
im
plem
entat
ion
Figure
3
.
Vis
ua
l
ou
tc
om
es
of
sta
ge
-
2
im
plem
entat
ion
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A novel
and
i
ntegr
ated arc
hitec
ture for
ide
ntif
ic
ation
and ca
ncell
atio
n of noise f
rom
G
SM
sig
na
l (
Rek
ha
N
)
4017
The
pri
m
e
rea
so
n
be
hind
thi
s
is
the
us
age
of
the
re
pres
ent
that
offer
s
the
bette
r
capab
il
it
y
to
disti
nguish
bet
ween
si
gn
al
s
.
Anothe
r
intere
sti
ng
fact
ab
out
this
ou
tc
ome
is
that
co
m
plete
ou
tc
om
e
c
an
be
analy
zed
just
f
ro
m
the
sign
al
s
receive
d
from
the
base
s
ta
ti
on
with
out
ha
vin
g
a
ny
de
pe
ndencies
to
ca
pture
an
y
GS
M
sig
nals
r
igh
t
f
ro
m
the
m
ov
ing
ve
hicle
.
This
ope
rati
on
m
akes
a
fas
te
r
analy
sis
of
the
noise
in
th
e
GS
M
sign
al
with m
or
e
pr
eci
se
ness
t
o detec
t t
he
s
puri
ou
s
sig
nal.
Fig
ure
4
highli
gh
ts
the
vis
ua
l
ou
tc
o
m
e
of
the
visu
al
ou
tc
om
e
of
the
al
gorithm
-
3whe
re
it
is
sh
ow
n
that
the
or
igi
na
l
GS
M
sig
nal
i
s
f
ur
the
r
s
ubj
e
ct
ed
to
T
DM
A
noise
.
T
his
pa
rt
of
the
a
naly
s
is
consi
ders
T
DMA
no
ise
as
it
include
s
var
i
ous
ot
her
f
or
m
s
of
no
ise
s
to
o
wit
hin
a
GS
M
si
gnal
.
The
visu
al
ou
tc
om
e
al
so
sh
ows
the
ou
tc
om
e
wav
ef
or
m
as
well
as
sp
ect
ro
gr
a
m
to
be
fr
ee
from
TDMA
no
ise
.
F
or
an
e
f
fecti
ve
analy
sis,
the
pro
po
se
d
syst
e
m
is al
so
s
ubj
e
ct
ed
to c
om
parat
ive an
al
ysi
s
with a
n
e
xisti
ng syst
em
.
Figure
4
.
Vis
ua
l ou
tc
om
es o
f St
age
-
3
i
m
plem
entat
ion
Table
1
highli
gh
t
t
hat
the
pro
po
s
ed
syst
em
offer
s
be
tt
er
noise
reducti
on
perform
ance
as
com
par
e
d
t
o
al
l
the
m
ajo
r
denoisin
g
ap
proac
hes
in
e
xisti
ng
a
ppr
oa
ches.
Ap
a
rt
from
per
centi
le
of
noise
re
duct
ion
par
am
et
ers,
th
e
pro
posed
sys
tem
is
al
so
wit
nesse
d
to
offe
r
faster
proc
essi
ng
tim
e
in
co
r
e
i7
process
.
E
ven
in
lowe
r
process
or
ve
rsion,
the
n
the
dif
fer
e
nc
e
in
no
ise
reducti
on
in
te
rm
s
of
SN
R
is
ve
ry
m
uch
neg
l
igible
.
Hen
ce
,
the
pro
po
s
ed
syst
e
m
can
be
cl
ai
m
ed
to
offe
r
h
ig
hly
extensi
ble,
c
ost
-
eff
ect
ive
,
a
nd
pr
act
ic
al
deno
isi
ng
appr
oach usi
ng one
window
operati
on i
n ord
er to ad
dr
e
ss th
e noise
prob
le
m
s in
GS
M si
gnal
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2088
-
8708
In
t J
Elec
&
C
om
p
En
g,
V
ol.
9
, N
o.
5
,
Oct
ober
20
19
:
4
0
1
0
-
4
0
1
9
4018
Table
1
.
C
om
par
at
ive
a
naly
sis
Sl.
No
.
Den
o
isin
g
App
roa
ch
es
% o
f
Nois
e Red
u
ctio
n
Proces
sin
g
T
i
m
e
(
s
)
1
Bu
tterworth
Filter
7
5
.21
%
1
.98
2
2
Ch
eb
y
sh
ev
f
ilter
6
7
.98
%
2
.56
1
1
3
Elliptical f
ilter
73%
1
.48
8
2
4
W
ein
er
f
ilter
71%
0
.96
7
2
5
Ad
ap
tiv
e Nois
e ca
n
cellatio
n
82%
2
.67
7
1
6
Prop
o
sed
Sy
ste
m
98%
0
.32
8
7
4.
CONCL
US
I
O
N
N
oi
s
e
c
a
nc
e
l
l
a
ti
on
f
r
om
t
he
a
co
us
t
i
c
G
S
M
s
i
gn
a
l
i
s
de
f
i
ni
t
e
l
y
no
t
a
n
e
a
s
y
t
a
s
k.
A
t
pr
e
s
e
n
t
,
ba
s
i
c
a
ll
y
,
t
he
ha
r
dw
a
r
e
-
b
a
s
e
d
a
pp
r
oa
c
h
e
s
a
r
e
us
e
d
f
o
r
e
ns
ur
i
n
g
t
ha
t
t
he
r
e
i
s
no
no
i
s
e
em
be
dd
e
d
w
i
t
h
t
he
t
r
a
nsm
it
t
e
d
G
S
M
s
i
gn
a
l
.
H
ow
e
ve
r
,
t
he
r
e
a
r
e
l
e
s
s
e
f
f
e
c
t
iv
e
m
o
de
l
i
ng
f
o
un
d
f
or
t
hi
s
r
e
a
s
on
.
T
he
r
e
f
or
e
,
t
hi
s
pa
pe
r
pr
e
s
e
nt
s
a
n
i
nt
e
gr
a
t
e
d m
od
e
li
ng
w
he
r
e
t
he
n
oi
s
e
c
a
n
c
e
l
l
at
i
on
i
s
do
n
e
i
n
a
pr
og
r
e
s
s
i
ve
m
a
nn
e
r
.
T
h
e
no
ve
l
t
y
of
t
hi
s
pa
pe
r
a
r
e
i
)
m
ul
t
i
pl
e
f
or
m
s
of
no
i
s
e
e
.
g.
r
a
n
do
m
no
i
s
e
,
w
hi
t
e
no
i
s
e
,
T
D
M
A
no
i
s
e
a
r
e
po
s
s
i
b
l
e
t
o
be
i
de
nt
i
f
i
ed
a
n
d
c
a
nc
e
l
e
d,
2)
t
he
im
pl
em
e
nt
at
i
on
of
pr
op
os
e
d
s
y
s
t
em
i
s
no
t
ba
s
e
d
o
n
a
ny
s
op
hi
s
t
i
c
a
te
d
f
i
l
t
e
r
w
i
t
h
hi
gh
e
r
r
e
s
ou
r
c
e
de
pe
n
de
nc
i
e
s
a
nd
he
nc
e
i
t
i
s
c
os
t
ef
f
e
c
t
i
ve
,
3)
a
pa
r
t
f
r
om
no
i
s
e
ca
nc
e
l
l
a
ti
on
,
pr
op
os
e
d
s
y
s
t
em
of
f
e
r
s
hi
gh
e
r
r
e
t
e
nt
i
on
o
f
s
i
g
na
l
qu
a
l
i
ty
t
oo
,
a
nd
4
)
pr
op
os
e
d
s
y
s
t
em
of
f
e
r
s
a
hi
g
hl
y
pr
a
c
t
i
c
a
l
ap
pr
oa
c
h
w
he
r
e
po
w
e
r
a
l
l
oc
a
ti
on
ov
e
r
t
r
a
nsm
it
t
i
ng
de
vi
c
e
i
s
c
on
s
i
de
r
e
d
a
s
o
ne
pr
om
i
ne
nt
po
i
nt
i
n
n
oi
s
e
c
a
nc
e
l
l
at
i
on.
REFERE
NCE
S
[1]
Q
.
S
.
M
a
h
d
i
,
e
t
a
l
.
,
“
A
v
a
i
l
a
b
i
l
i
t
y
a
n
a
l
y
s
i
s
o
f
G
S
M
n
e
t
w
o
r
k
s
ys
t
e
m
s
,”
A
n
t
e
n
n
as
P
r
op
a
g
a
t
i
o
n
a
n
d
E
M
T
h
e
o
r
y
(
I
S
A
P
E)
,
2
0
1
0
9
th
I
n
t
e
r
n
a
t
i
o
n
a
l
S
y
m
p
o
s
i
um
o
n
.
I
E
E
E
,
2
0
1
0
.
[2]
G
u
G
.
a
n
d
P
e
ng
G
.
,
“
T
h
e
s
urv
e
y
o
f
G
S
M
wi
r
e
l
e
s
s
c
o
m
m
u
n
ica
t
i
o
n
s
y
s
t
e
m
,”
In
C
o
mp
u
t
e
r
a
n
d
I
n
f
o
rm
a
t
i
o
n
A
p
p
l
i
c
a
t
i
o
n
(
I
CC
I
A
)
,
2
0
1
0
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
,
I
E
E
E
,
p
p
.
1
2
1
-
124
,
D
e
c
2010
.
[3]
Š
k
r
b
i
ć
M
.
,
e
t
a
l
.
,
“
W
e
b
-
b
a
s
e
d
se
r
v
i
c
e
i
m
p
l
e
m
e
n
t
a
t
i
o
n
v
i
a
G
S
M
n
e
t
w
o
r
k
,”
In
T
e
l
e
c
o
m
m
u
n
i
c
a
t
i
o
n
s
F
o
r
u
m
T
e
l
f
o
r
(
T
E
L
F
O
R)
,
2
0
1
4
2
2
nd
,
I
E
E
E
,
p
p
.
2
5
2
-
255
,
2014
.
[4]
V
.
P
.
V
e
n
k
a
t
e
s
a
n
,
“
A
r
c
h
i
t
e
c
t
u
r
a
l
P
a
t
t
e
r
n
o
f
H
e
a
l
t
h
C
a
r
e
S
y
s
t
e
m
U
s
i
n
g
GS
M
N
e
t
w
or
k
s
,”
a
r
X
i
v
p
r
e
p
r
i
n
t
a
r
X
i
v
:
1
3
1
2
.
2
3
2
3
,
2013
.
[5]
N
.
R
e
k
h
a
a
n
d
F
.
J
a
b
e
e
n
,
“
S
t
u
d
y
o
n
a
p
p
r
o
a
c
h
e
s
of
n
o
i
s
e
c
a
n
c
e
l
l
a
t
i
o
n
i
n
GS
M
c
om
m
u
n
i
c
a
t
i
o
n
c
ha
n
n
e
l
,”
C
o
m
m
u
n.
A
p
p
l
.
E
l
e
c
t
r
o
n
,
v
o
l
/
i
s
s
u
e
:
3
(
5
)
,
p
p
.
5
-
11
,
2015
.
[6]
S
.
C
.
M
o
ho
n
t
a
,
e
t
a
l
.
,
“
S
t
u
d
y
o
f
D
i
f
f
e
r
e
n
t
T
y
p
e
s
o
f
N
o
is
e
a
n
d
I
t
s
E
f
f
e
c
t
s
i
n
C
om
m
u
n
i
c
a
t
i
o
n
S
y
s
t
e
m
s
,”
I
n
t
e
rn
a
t
i
o
na
l
J
o
u
r
n
a
l
o
f
E
n
g
i
n
e
e
r
i
n
g
a
n
d
M
a
n
ag
e
m
e
n
t
R
e
s
e
a
r
c
h
(
I
J
E
M
R
)
,
v
o
l
/
i
ss
u
e
:
5
(
2
)
,
p
p
.
410
-
413
,
2015
.
[7]
I
.
C
l
a
e
s
s
o
n
a
n
d
A
.
N
i
l
s
so
n
,
“
C
a
nc
e
l
l
a
t
i
o
n
o
f
h
u
m
m
i
n
g
G
SM
m
o
b
ile
t
e
l
e
p
h
o
n
e
n
o
i
s
e
,”
I
n
f
o
r
m
a
t
i
o
n
,
C
o
m
mu
n
i
c
a
t
i
o
n
s
a
n
d
S
i
g
n
a
l
Pr
o
ces
s
i
n
g
,
2
0
03
a
nd
F
o
u
r
t
h
P
a
c
i
f
ic
R
i
m
C
o
n
f
e
r
e
nc
e
o
n
M
u
l
t
i
m
e
d
i
a
.
P
r
o
c
e
e
d
i
n
g
s
o
f
t
h
e
2
0
0
3
J
o
in
t
C
o
n
f
e
r
e
n
c
e
o
f
t
h
e
F
o
u
r
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
.
,
I
E
E
E
,
v
o
l
.
1
,
2
0
0
3
.
[8]
M
.
A
.
Ru
d
e
r
,
e
t
a
l
.
,
“
S
i
n
g
l
e
A
n
ten
n
a
I
n
t
e
r
f
e
r
e
n
c
e
C
a
n
c
e
l
l
a
t
i
o
n
f
o
r
GS
M
/V
AM
OS
/
E
D
G
E
U
s
i
ng
$
L
_
{
p
}
$
-
N
o
rm
D
e
t
e
c
t
i
o
n
a
n
d
D
e
c
o
d
i
n
g
,”
I
E
E
E
T
r
a
ns
a
c
t
i
o
n
s
o
n
W
i
r
e
l
e
s
s
C
o
mm
un
i
c
a
t
i
o
n
s
,
v
o
l
/
i
s
su
e
:
14
(
5
)
,
p
p
.
2
4
1
3
-
2425
,
2015
.
[9]
J
.
B
i
t
z
e
r
a
n
d
J
.
R
a
d
e
m
a
c
h
e
r
,
“
De
t
e
c
t
i
o
n
,
I
n
t
e
r
p
o
l
a
t
i
o
n
a
n
d
C
a
n
c
e
l
l
a
t
i
o
n
A
l
g
o
r
i
t
h
m
s
f
o
r
G
SM
b
ur
s
t
R
e
m
o
v
a
l
f
o
r
F
o
r
e
n
s
i
c
A
u
d
i
o
,
”
R
e
s
e
a
r
c
h
G
a
t
e
,
2
0
1
5
.
[
1
0]
H
.
M
e
h
t
a
,
e
t
a
l
.
,
“
0
G
t
o
5
G
m
ob
i
l
e
t
e
c
h
n
o
l
o
g
y
:
A
s
u
r
v
e
y
,”
J
.
o
f
B
a
s
i
c
a
n
d
A
p
p
l
i
e
d
E
n
g
i
n
e
e
r
i
n
g
R
e
se
a
r
c
h
,
v
o
l
/
i
s
su
e
:
1
(
6
)
,
p
p
.
56
-
60
,
2014
.
[
1
1]
R
e
k
h
a
N
.
a
n
d
F
.
J
a
b
e
e
n
,
“
S
t
u
d
y
o
n
A
p
p
r
o
a
c
h
e
s
o
f
N
o
i
s
e
C
a
n
c
e
l
l
a
t
i
o
n
i
n
G
SM
C
o
m
m
un
i
c
a
t
i
o
n
C
h
a
n
n
e
l
,
”
C
o
m
mu
n
i
c
a
t
i
o
n
s
o
n
A
p
p
l
i
e
d
E
l
e
c
t
r
o
n
i
c
s
,
v
ol
/
i
s
s
u
e
:
3
(
5
)
,
N
o
v
2
0
1
5
.
[
1
2]
S
.
M
.
Nø
r
h
o
lm
,
e
t
a
l
.
,
“
S
i
n
g
l
e
-
c
h
a
n
n
e
l
n
o
i
s
e
r
e
du
c
t
i
o
n
u
s
i
n
g
un
i
f
i
e
d
j
o
i
n
t
d
i
a
g
o
n
a
l
i
z
a
t
i
o
n
a
n
d
o
p
t
i
m
a
l
f
i
l
t
e
r
i
n
g
,”
E
U
R
A
S
I
P
J
o
u
r
n
a
l
o
n
A
d
v
a
n
c
e
s
i
n
S
i
g
n
a
l
P
r
o
c
e
s
si
n
g
,
v
o
l
/
i
s
s
u
e
:
2014
(
1
),
p
p
.
37
,
2014
.
[
1
3]
B
e
r
t
r
a
n
d
A
.
,
e
t
a
l
.
,
“
A
d
a
p
t
i
v
e
d
i
s
t
r
i
b
u
t
e
d
n
o
i
s
e
r
e
d
u
c
t
i
o
n
f
o
r
sp
e
e
c
h
e
n
h
a
n
c
e
m
e
n
t
i
n
w
i
r
e
l
e
s
s
a
c
o
u
s
t
i
c
s
e
n
s
o
r
n
e
t
w
o
r
k
s
,”
I
n
p
r
oc
.
O
f
t
h
e
I
n
t
e
r
n
a
t
i
o
n
a
l
W
o
r
k
s
h
o
p
o
n
A
c
o
u
s
t
i
c
E
c
h
o
a
n
d
N
o
i
s
e
C
o
nt
r
o
l
(
I
W
A
E
N
C)
,
J
a
n
2010.
[
1
4]
S
a
y
o
u
d
A
.
,
e
t
a
l
.
,
“
A
t
wo
-
c
h
a
n
ne
l
f
a
s
t
N
L
MS
alg
o
r
i
t
h
m
f
or
sp
e
e
c
h
e
n
h
a
n
c
e
m
e
n
t
a
n
d
a
c
o
u
s
t
i
c
no
i
s
e
r
e
d
u
c
t
i
o
n
,”
I
n
e
l
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
-
B
o
u
m
e
rd
e
s
(
IC
E
E
-
B
)
,
20
1
7
5
t
h
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
,
I
E
E
E
,
p
p
.
1
-
4
,
O
c
t
2017
.
[
1
5]
H
.
R
a
h
i
m
a
,
e
t
a
l
.
,
“
B
l
i
n
d
s
p
e
e
c
h
e
n
h
a
n
c
e
m
e
n
t
a
n
d
a
c
o
u
s
t
i
c
n
o
i
s
e
r
e
d
u
c
t
i
o
n
b
y
S
F
T
F
a
d
a
p
t
i
v
e
a
l
g
o
r
i
t
h
m
,”
E
l
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
-
B
o
um
e
r
d
e
s
(
I
C
E
E
-
B)
,
2
0
1
7
5
t
h
I
n
t
e
rn
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n
,
I
E
E
E
,
2
0
1
7
.
[
1
6]
L
e
h
m
a
n
n
A
.
M
.
,
e
t
a
l
.
,
“
S
i
n
g
l
e
-
a
n
t
e
n
n
a
i
n
t
e
r
f
e
r
e
n
c
e
c
a
n
c
e
l
l
a
t
i
o
n
f
o
r
c
o
m
p
l
e
x
-
va
l
u
e
d
s
i
g
n
a
l
c
o
n
s
t
e
l
l
a
t
i
o
n
s
w
i
t
h
a
p
p
l
i
c
a
t
i
o
n
s
t
o
G
SM
/
E
D
G
E
.
I
np
e
r
s
o
n
a
l
I
n
d
o
or
a
n
d
M
o
b
i
l
e
R
a
di
o
C
o
m
m
un
i
c
a
t
i
o
n
s
(P
I
MR
C
)
,
”
2
0
1
1
I
E
E
E
2
2
nd
I
n
t
e
r
n
a
t
i
o
n
a
l
S
y
m
p
o
s
i
um
o
n
,
I
E
E
E
,
p
p
.
1
4
1
7
-
1422
,
S
e
p
2011
.
[
1
7]
M.
H
am
i
d
i
a
a
n
d
A
.
Am
r
o
u
c
h
e
,
“
In
f
l
u
e
n
c
e
o
f
n
o
i
sy
c
h
a
n
n
e
l
o
n
a
c
o
u
s
t
i
c
e
c
h
o
c
a
n
c
e
l
l
a
t
i
o
n
i
n
m
o
b
i
l
e
c
o
m
m
u
n
i
c
a
t
i
o
n
,”
M
i
c
r
o
e
l
e
c
t
r
o
n
i
c
s
(
I
C
M)
,
2
0
1
2
2
4
th
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
e
r
e
n
c
e
o
n,
I
E
E
E
,
2
0
1
2
.
[
1
8]
S
a
d
o
k
M
.
,
e
t
a
l
.
,
“
E
n
h
a
n
c
e
d
s
i
ng
l
e
a
n
t
e
n
n
a
i
n
t
e
r
f
e
r
e
n
c
e
c
a
n
c
e
l
l
a
t
i
o
n
f
r
o
m
MM
S
E
t
h
i
r
d
-
o
r
d
e
r
c
o
m
p
l
e
x
V
o
l
t
e
r
r
a
f
i
l
t
e
r
s
,”
I
n
a
c
o
u
s
t
i
c
s
,
S
p
e
e
c
h
a
n
d
S
i
g
n
a
l
P
r
o
c
e
s
s
i
n
g
(
I
C
A
S
S
P)
,
2
0
1
7
I
E
E
E
I
n
t
e
r
n
a
t
i
o
n
a
l
C
o
n
f
er
e
n
c
e
o
n
,
I
E
E
E
,
p
p
.
4
1
9
7
-
4201
,
M
a
r
2017
.
[
1
9]
S
.
V
i
h
a
r
i
,
e
t
a
l
.
,
“
C
om
p
a
r
i
so
n
o
f
s
p
e
e
c
h
e
n
h
a
n
c
e
m
e
n
t
a
l
g
o
r
i
t
h
m
s
,”
P
r
o
c
e
d
i
a
c
o
m
p
u
t
e
r
s
c
i
e
n
c
e
,
v
o
l
.
89
,
p
p
.
666
-
676
,
2016
.
Evaluation Warning : The document was created with Spire.PDF for Python.
In
t J
Elec
&
C
om
p
En
g
IS
S
N: 20
88
-
8708
A novel
and
i
ntegr
ated arc
hitec
ture for
ide
ntif
ic
ation
and ca
ncell
atio
n of noise f
rom
G
SM
sig
na
l (
Rek
ha
N
)
4019
[
2
0]
M.
K
a
l
a
m
a
n
i
,
et
a
l
.
,
“
M
o
d
i
f
i
e
d
N
o
i
s
e
R
e
d
u
c
t
i
o
n
A
l
g
o
r
i
t
hm
fo
r
S
p
e
e
c
h
E
n
h
an
c
e
m
e
n
t
,”
A
p
p
l
i
e
d
M
a
t
h
e
m
a
t
i
c
al
S
c
i
e
n
c
e
s
,
v
o
l
/
i
s
s
u
e
:
8
(
89
)
,
p
p
.
4
4
4
7
-
4452
,
2014
.
[
2
1]
N
.
U
p
a
d
h
y
a
y
a
n
d
R
.
K
.
J
a
i
sw
a
l
,
“
S
i
ng
l
e
c
h
a
n
n
e
l
s
p
e
e
c
h
e
n
h
a
n
c
e
m
e
n
t
:
u
s
i
n
g
W
i
en
e
r
f
i
l
t
e
r
i
n
g
w
i
t
h
r
e
c
u
r
s
i
v
e
n
o
i
se
e
s
t
i
m
a
t
i
o
n
,”
P
r
o
c
e
d
i
a
C
o
m
p
u
t
e
r
S
c
i
e
n
c
e
,
v
o
l
.
84
,
p
p
.
22
-
30
,
2016
.
[
2
2]
B
.
S
.
P
r
e
m
a
n
a
nd
a
a
n
d
B
.
V
.
U
m
a
,
“
S
p
e
e
c
h
e
n
h
a
n
c
e
m
e
n
t
a
l
g
o
r
i
t
h
m
t
o
r
e
d
u
c
e
t
h
e
e
f
f
e
c
t
o
f
b
a
c
k
g
r
o
u
n
d
n
o
is
e
i
n
m
o
b
i
l
e
p
h
o
n
e
s
,”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
W
i
r
e
l
e
s
s
&
M
ob
i
l
e
N
e
t
w
o
r
k
s
,
v
o
l
/
i
s
s
u
e
:
5
(
1
)
,
p
p
.
177
,
2013
.
[
2
3]
A
.
S
h
u
k
a
l
a
,
e
t
a
l
.
,
“
N
o
i
s
e
R
e
d
u
c
tio
n
a
n
d
E
c
h
o
C
a
n
c
e
l
l
a
t
i
o
n
U
s
i
n
g
T
h
r
e
s
h
o
l
d
F
i
l
t
e
rs
i
n
H
a
n
ds
Fr
e
e
C
o
m
m
u
n
i
c
a
t
i
o
n
S
y
s
t
e
m
s
,
”
J
o
u
r
n
a
l
o
f
E
n
g
i
n
e
e
r
i
n
g
R
e
s
e
a
r
c
h
a
n
d
A
p
p
l
i
c
a
t
i
o
n
,
2018
.
[
2
4]
F
.
A
f
r
o
z
,
e
t
a
l
.
,
“
P
e
r
f
o
rm
a
n
c
e
a
n
a
l
y
s
i
s
o
f
a
d
a
p
t
i
v
e
n
o
i
s
e
c
a
n
c
e
l
l
e
r
e
m
p
l
o
y
i
n
g
N
L
M
S
A
l
g
or
i
t
h
m
,”
I
n
t
e
r
n
a
t
i
o
nal
J
o
u
r
n
a
l
o
f
W
i
r
e
l
e
s
s
&
M
o
b
i
l
e
N
e
t
w
o
r
k
s
,
2015
.
[
2
5]
P.
G
up
t
a
,
e
t
a
l
.
,
“
P
e
r
f
o
rm
a
n
c
e
a
n
a
l
y
s
i
s
o
f
s
p
e
ech
e
n
h
a
n
c
e
m
e
n
t
u
s
i
n
g
L
MS
,
NL
M
S
a
n
d
UN
AN
R
a
l
g
o
r
i
t
hm
s
,”
C
o
m
p
u
t
e
r
,
C
o
mm
u
n
i
c
a
t
i
o
n
,
a
n
d
C
o
n
t
r
o
l
(
I
C
4)
,
20
1
5
I
n
t
e
r
n
a
t
i
o
n
al
C
o
n
f
e
r
e
n
c
e
o
n
.
I
E
E
E
,
2
0
1
5
.
[
2
6]
G
b
a
d
a
m
os
i
S
.
A
.
,
e
t
a
l
.
,
“
N
o
n
-
In
t
r
u
s
i
v
e
N
o
is
e
R
e
d
u
c
t
i
o
n
I
n
Gs
m
Vo
i
c
e
S
i
g
n
a
l
U
s
i
n
g
N
o
n
-
P
a
ra
m
e
t
r
i
c
M
o
d
e
l
i
n
g
T
e
c
h
n
i
q
u
e
,”
2
0
1
5
.
[
2
7]
J.
B
i
t
z
e
r
a
n
d
J
.
R
a
d
e
m
a
c
h
e
r
,
“
De
t
e
c
t
i
o
n
,
I
n
t
e
r
p
o
l
a
t
i
o
n
a
n
d
C
a
n
c
e
l
l
a
t
i
o
n
A
l
g
o
r
i
t
h
m
s
f
o
r
G
SM
b
u
r
s
t
R
e
m
o
v
a
l
f
o
r
F
o
r
e
n
s
i
c
A
u
d
i
o
,”
C
o
n
f
e
r
e
n
c
e
o
f
t
h
e
I
n
t
'
l
.
A
s
s
o
c
.
F
o
r
F
o
r
e
n
s
i
c
P
h
o
ne
t
i
c
s
(
I
A
F
P
2
0
0
3
)
,
n
o
.
2
1
,
2
0
0
3
.
[
2
8]
U.
M
a
h
b
u
b
a
n
d
S
.
A
.
F
a
t
t
a
h
,
“
G
r
a
d
i
e
n
t
B
a
s
e
d
A
d
a
p
t
i
v
e
A
l
g
o
r
i
t
h
m
f
o
r
E
c
h
o
C
a
n
c
e
l
l
a
t
i
o
n
f
r
o
m
R
e
c
o
r
d
e
d
E
c
h
o
C
o
r
r
u
p
t
e
d
S
p
e
e
c
h
,”
A
d
v
a
n
c
e
s
i
n
E
l
e
c
t
r
i
c
a
l
E
n
g
i
n
e
e
r
i
n
g
,
2014
.
[
2
9]
D
.
X
.
Wa
n
g
,
e
t
a
l
.
,
“
S
p
e
e
c
h
E
n
h
a
n
c
e
m
e
n
t
C
o
n
t
r
o
l
D
e
s
i
g
n
A
l
g
o
r
i
t
hm
f
o
r
D
u
a
l
-
M
i
c
r
o
p
h
o
n
e
S
y
s
t
e
m
s
U
s
i
n
g
β
-
NM
F
in
a
C
o
m
p
l
e
x
E
n
v
i
r
o
n
m
e
n
t
,”
C
o
m
p
lex
i
t
y
,
2018
.
[
3
0]
Y
.
Y
.
C
h
e
n
,
“
Spe
e
c
h
E
n
h
a
n
c
e
m
e
n
t
o
f
Mo
b
i
l
e
D
ev
i
c
e
s
B
a
s
e
d
o
n
t
h
e
I
n
t
e
g
r
a
t
i
o
n
o
f
a
D
u
a
l
M
i
c
r
o
pho
n
e
A
r
r
a
y
a
n
d
a
B
a
c
k
g
r
o
u
n
d
N
o
is
e
E
l
i
m
i
n
a
t
i
o
n
A
l
g
o
r
i
t
h
m
,”
S
e
n
s
or
s
,
v
o
l
/
i
s
s
u
e
:
18
(
5
)
,
p
p
.
1467
,
2018
.
[31]
A
.
R
.
Shakee
b
and
K.
H.
Sa
y
i
dm
ari
e,
“
A
ce
llular
base
stat
io
n
ant
enna
conf
i
gura
ti
on
for
var
ia
bl
e
cove
rag
,”
Inte
rnational
Jo
urnal
of El
e
ct
ri
c
al
&
Co
mputer
Engi
ne
ering,
vol
/i
ss
ue:
9
(
3
)
,
201
9
.
[32]
S.
Kolle
m
,
et
a
l.,
“
Im
age
Denoising
b
y
using
Modifie
d
SG
HP
A
lgori
thm
,”
Int
ernati
onal
Journal
of
El
ectric
a
l
&
Computer
Engi
n
ee
ring
,
vol/is
sue:
8
(
2
),
2018
.
[33]
A.
Aw
ad,
“
Re
m
oval
of
Fixe
d
-
val
ued
Im
pulse
Noise
base
d
on
Probabi
lit
y
o
f
Exi
sten
c
e
of
the
Im
ag
e
Pixel
,”
Int
ernational
Journal
of
El
e
ct
rica
l
and
C
omputer
Engi
n
e
ering
,
vo
l/
issue:
8
(
4
),
pp
.
2106
,
2
018
.
[
3
4]
N
.
R
e
k
h
a
a
n
d
F
.
J
a
b
e
e
m
,
“
S
AN
C
:
S
t
o
c
h
a
s
t
i
c
a
p
p
r
o
a
c
h
f
o
r
n
o
i
s
e
c
a
n
c
e
l
l
a
t
i
o
n
i
n
G
SM
s
i
g
n
a
l
s
,
”
20
1
5
I
n
t
e
r
n
a
t
i
o
n
al
C
o
n
f
e
r
e
n
c
e
o
n
A
p
p
l
i
e
d
a
n
d
T
h
e
o
re
t
i
c
a
l
C
o
m
p
u
t
i
n
g
a
n
d
C
o
mm
u
n
i
ca
t
i
o
n
T
e
c
h
n
o
l
o
gy
(
i
C
A
T
c
c
T)
,
D
av
a
n
g
e
r
e
,
p
p
.
7
0
1
-
707
,
2015
.
[35]
R
e
k
h
a
N
.
a
n
d
F
.
J
a
b
e
e
n
,
“
N
o
ve
l
T
e
c
h
n
i
q
u
e
f
o
r
C
o
m
p
r
e
h
e
ns
i
ve
N
o
i
s
e
I
d
e
n
t
i
f
i
c
a
t
i
o
n
a
n
d
C
a
n
c
e
l
l
a
t
i
o
n
i
n
G
S
M
S
i
g
n
a
l
,
”
I
n
t
e
r
n
a
t
i
o
n
a
l
J
o
u
r
n
a
l
o
f
E
l
e
c
t
r
i
c
a
l
a
n
d
C
o
m
p
u
t
e
r
E
n
g
i
n
e
e
r
i
n
g
,
v
ol
/
i
s
s
u
e
:
8
(
2
)
,
p
p
.
1
2
2
2
-
1229
,
A
p
r
2018
.
BIOGR
AP
H
I
ES
OF
A
UTH
ORS
Rekh
a
N.
is
Ass
oci
a
te
Profess
or
in
Depa
rtment
o
f
Te
lecom
m
unic
at
ion
Engi
n
ee
r
in
g,
K.
S.
Instit
ute
of
Technol
og
y
,
Benga
luru
,
Ka
rna
ta
k
a
,
Ind
ia.
She
recei
ved
B
.
E
d
egr
e
e
in
El
e
ct
roni
cs
and
Com
m
unic
at
ion
Engi
ne
eri
ng
fro
m
Vi
svesvara
y
a
Te
chno
logi
c
al
Univer
sit
y
,
B
el
ag
a
vi,
Karna
t
aka
in
the
y
ea
r
2002
an
d
M.T
ec
h
degr
ee i
n
Digit
al
El
e
ct
r
onic
s,
from
the
sam
e
unive
rsit
y
i
n
the
y
ea
r
2009
.
She
is
cur
r
en
tly
pursuing
her
Ph.
D.
degr
e
e
in
the
Depa
rtment
of
El
e
ct
roni
cs
and
Com
m
unic
at
ion
Engi
ne
eri
ng,
Vi
svesvara
y
a
Tec
hnologi
c
al
Univer
sit
y
,
Be
la
gav
i.
Her
rese
arc
h
i
nte
rests
inc
lud
e
wire
le
ss
comm
unic
a
ti
on,
sec
ur
e
comm
unic
atio
n
net
works
,
D
at
a
Com
m
unic
a
ti
on,
MA
T
LAB
m
odel
ing
and
si
m
ula
ti
on
.
Fathima
Jabee
n,
Ph.
D.
,
is
Prin
ci
pa
l
of
Isl
amia
h
Instit
ut
e
of
Technol
og
y
,
Beng
al
uru,
Karna
ta
ka
,
India
.
She
is
a
Supervisor
for
t
he
Ph.D.
schol
a
rs
under
Visvesvara
y
a
Technol
ogic
a
l
Univer
si
t
y
,
Bel
ag
avi
,
Dr.
M
GR
Univer
sity
,
Jain
Univ
ersity
and
al
so
a
n
ext
ern
al
exam
ine
r
for
The
s
is
eva
lu
at
ion
/
Publi
c
Viva
-
vo
ce
o
f
Ph.D.
studen
ts.
Sh
e
has
b
ee
n
in
th
e
teac
hing
for
pro
fession
cour
ses
under
UG
/PG
le
vel
for
nea
rl
y
2
9
y
ea
rs
.
She
is
a
rev
ie
wer
of
se
ver
al
Nat
iona
l
a
nd
Int
ern
ationa
l
journa
ls.
Dr
.
Fat
hima
Jabe
en
is
a
Fell
ow
Mem
ber
of
the
Insti
tut
io
n
of
Engi
ne
ers
(
IND
IA).
She
is
al
so
a
li
fe
m
ember
of
seve
ral
profe
ss
iona
l
bodie
s,
inc
lud
in
g
India
n
Socie
t
y
for
Te
chn
ical
Educ
a
ti
on
(IST
E)
and
m
ember
IEE
E
.
She
has
nea
rl
y
35
publica
t
ions
under
her
name.
Her
area
of
int
er
est
includes
Com
m
unic
at
ion
,
Instrum
ent
a
ti
o
n,
Embedde
d
S
ystems
,
W
ire
le
ss
comm
unic
at
ion,
Autom
oti
ve
E
le
c
troni
cs
and
VLS
I.
Evaluation Warning : The document was created with Spire.PDF for Python.