Internati
o
nal
Journal of Ele
c
trical
and Computer
Engineering
(IJE
CE)
V
o
l.
6, N
o
. 2
,
A
p
r
il
201
6, p
p
.
67
4
~
68
1
I
S
SN
: 208
8-8
7
0
8
,
D
O
I
:
10.115
91
/ij
ece.v6
i
2.8
935
6
74
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
A Hybrid Approach of Fuzzy C-
Means Clustering and Neural
Network to Make Energy-Effi
cient Heterogeneous Wireless
Sensor Network
Amit Kr. Kau
s
hik
Manav Rachna
University
, Fa
ridabad, Har
y
an
a,
India
Article Info
A
B
STRAC
T
Article histo
r
y:
Received Aug 31, 2015
Rev
i
sed
No
v
20
, 20
15
Accepted Dec 12, 2015
The Wireless sensor network has been
highl
y
fo
cus
e
d res
ear
ch a
r
ea in re
cent
tim
es
due to its wide
app
lic
atio
n
s
and
ad
aptability
to
d
i
ffer
e
nt en
vironments.
The energ
y
-con
strained sensor
nodes are alw
a
y
s
under
consideration to
incre
a
se their l
i
f
etim
e. In this paper
we have used the advantages of two
approach
es
i
.
e
.
fuzz
y c-m
eans
clus
ter
i
ng
and
neural
network
to
m
a
ke an
energ
y
ef
ficient
network b
y
prolonging
the
lifetime of network.
The cluster
formation is don
e using FCM to
form e
qually
s
i
zed
clusters in
n
e
twork and
the decision of
choosing cluster
head
is done u
s
ing neural network having
input f
actors
as
distance from b
a
sestat
ion, h
e
ter
ogeneity
and
en
erg
y
of
the
node etc. Our
Approach has successful
ly
incr
eased th
e lifetime and data
capacity
of the n
e
twork and ou
tp
erformed
differ
e
nt approaches ap
plied
to the
network pres
ent
in li
ter
a
tur
e
.
Keyword:
FCM
Neu
r
al
net
w
or
k
Sens
or
n
o
d
e
W
i
rel
e
ss se
ns
o
r
net
w
or
k
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
:
Am
it Kr.
Ka
us
hik
Man
a
v Rachn
a
Un
iv
ersity
Fari
da
ba
d, Har
y
ana, In
di
a
Em
a
il: a
m
itk
au
sh
ik@m
ru
.edu.in
1.
INTRODUCTION
W
i
rel
e
ss Se
ns
or
Net
w
or
k (
W
S
N
) i
s
a s
e
l
f-co
n
f
i
g
uri
n
g
net
w
or
k w
h
i
c
h i
s
hi
ghl
y
adapt
a
bl
e t
o
diffe
re
nt envi
ronm
ent scenarios. Th
ey are
highly useful network in
human-inaccessi
ble envi
ronm
e
n
t for
m
oni
t
o
ri
ng
p
u
r
pos
es.
It
i
s
c
o
m
posed
of
a s
e
t
of
se
nso
r
nodes
(which are
also ca
lled
mo
tes). Th
e lifet
i
m
e o
f
W
S
N
is alw
a
y
s
a to
p
i
c
of
r
e
sear
ch
b
ecau
s
e
th
e sensor
nodes ar
e co
nstr
ain
e
d in
ter
m
o
f
b
a
tter
y
lif
e and
it is
not fea
s
ible to recha
r
ge t
h
e battery at regular basis b
eca
use
of their
depl
oym
e
nt in re
m
o
te and hostile places.
So
we nee
d
t
h
ose r
o
ut
i
ng
pr
o
t
ocol
s o
r
a
p
p
r
o
aches w
h
i
c
h ca
n m
a
xim
i
ze t
h
e dat
a
capaci
t
y
and al
s
o
p
r
ol
o
ng t
h
e
lifeti
m
e o
f
senso
r
nod
e.
So th
e ch
alleng
e is to
d
e
v
e
l
o
p
l
o
w-
po
we
r c
o
m
m
uni
cat
i
on ap
pr
oac
h
es
wi
t
h
l
o
w
-
cos
t
on
-
n
o
d
e
pr
oce
ssi
ng a
n
d sel
f
-
o
r
g
a
n
i
zed c
o
n
n
ect
i
v
i
t
y
/
p
rot
o
col
s
. Se
ve
ral
p
r
ot
ocol
s
were
devel
ope
d t
o
m
a
ke t
h
e
comm
unication e
n
ergy-e
ffec
tive to i
n
creas
e lifetim
e of t
h
e net
w
or
ks.
The wi
rel
e
ss
s
e
ns
or net
w
o
r
k
s
can
be
categorized i
n
tohom
ogenous
an
d hetere
oge
n
ous
networks
according t
o
th
e type
of
nodes network is
usi
ng.
Th
e
ho
m
o
g
e
nou
s
n
e
two
r
k is t
h
e
n
e
two
r
k
where all
n
o
d
e
s hav
e
equ
a
l en
erg
y
and all h
a
v
i
n
g
eq
u
a
l
prob
ab
ilit
y
t
o
bec
o
m
e
cl
ust
e
r hea
d
.
T
h
e
het
e
re
o
g
en
o
u
s
net
w
o
r
ks
ha
v
e
di
f
f
ere
n
t
ki
n
d
of
n
o
d
es
ha
vi
n
g
di
ffe
re
nt
ener
gy
an
d
prob
ab
ilit
ies o
f
b
e
co
m
i
n
g
cl
u
s
terh
eads. In
th
is
p
a
per we
will wo
rk
o
n
h
e
tereo
g
e
no
us n
e
t
w
o
r
k
s
.
Di
ffe
re
nt
pr
ot
ocol
s
ha
ve be
en f
o
r
h
o
m
ogeno
us a
nd
he
t
e
r
oge
n
o
u
s
net
w
or
ks.
T
hese p
r
o
t
ocol
s used di
f
f
ere
n
t
approaches to
im
prove the
com
m
uni
cat
i
on and t
r
an
sm
i
s
si
on o
f
t
h
e pac
k
ets in
th
e n
e
twork. Bu
t th
ey lack
di
ffe
re
nt
t
h
i
n
g
s
such as t
o
h
a
ve a o
p
t
i
m
a
l
di
st
ri
b
u
t
i
on
of
no
des i
n
net
w
o
r
k
,
eq
ual
l
y
si
zed cl
ust
e
rs
, rol
e
o
f
resi
d
u
al
ene
r
g
y
i
n
el
ect
i
on o
f
cl
ust
e
r
head
et
c. The n
o
n
-
o
pt
im
al
di
st
ri
bu
t
i
on o
f
t
e
nl
y
m
a
ke t
h
e
di
ssi
pa
t
i
on o
f
the energy m
o
re rapidly in the
cluste
r th
an
t
h
e o
p
tim
al
d
i
strib
u
tion
in
th
e
network. The
unequal
-
sized clusters
lead to the les
s
er data
-capaci
ty of
the net
w
ork as c
o
m
p
ared to the e
q
ua
l-sized clusters
. Differe
n
t protocols
have
be
en
m
a
de t
o
m
a
ke t
h
e
ro
ut
i
n
g
efficient s
u
ch as L
E
A
CH [1,
2],
[3],
SEP
[4], DE
EC [5] etc. T
h
ese all
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
67
4 – 6
8
1
67
5
pr
ot
oc
ol
s
were
base
d
on
co
nc
ept
o
f
cl
u
s
t
e
ri
n
g
i
n
t
h
e
net
w
o
r
k t
o
c
ons
er
ve
t
h
e ene
r
gy
o
f
t
h
e
no
des.
T
h
i
s
was
fi
rst
l
y
ap
pl
i
e
d
i
n
t
h
e
LEAC
H
pr
ot
ocol
.
LE
A
C
H p
r
ot
oc
ol
w
a
s de
vel
o
pe
d a
s
cl
ust
e
ri
ng
ba
sed
ro
ut
i
n
g
pr
o
t
ocol
in
wh
ich th
e
clu
s
ter-h
e
ad
is el
ected
using
p
r
ob
ab
ility-b
ased
th
resh
o
l
d
i
ng
mech
an
ism
.
LEAC
H
[1] el
ects the cluste
r hea
d
by us
i
n
g a
t
h
res
h
ol
di
n
g
exp
r
essi
on
as
T(
)=
/(
1
−
(
1/
))
(
1
)
Wh
ere i is
no
de, p is th
e
probab
ility o
f
ch
oosin
g
t
h
e cl
u
s
ter h
e
ad
and
r is roun
d nu
m
b
er. A ran
d
o
m
n
o
‘s’ is
gene
rat
e
d
bet
w
een
0 a
nd
1 a
n
d i
s
c
o
m
p
ared
wi
t
h
T
(
i
)
.
If
T(i) is greater t
h
a
n
s t
h
en node i
becom
e
cluster hea
d
o
t
h
e
rwise no
t. Bu
t
as
it
was based
o
n
prob
ab
ility
b
a
sed
cl
u
s
ter-h
e
ad
electio
n
wh
ich
can lead
to
no
n-op
ti
m
a
l
and
u
n
eq
ual
-
si
zed cl
ust
e
r a
n
d
al
so di
d
not
t
a
ke acco
u
n
t
t
h
e
resi
d
u
al
ener
g
y
fact
or i
n
det
e
rm
i
n
at
i
on of
cl
ust
e
r
heads.
The
[3]
was t
o
m
a
ke im
pro
v
e t
h
e
LEAC
H
p
r
ot
oc
ol
by
rem
ovi
ng
i
t
s
di
scre
panci
e
s
by
t
a
ki
n
g
t
h
e
fact
or
o
f
resi
d
u
a
l
ener
gy
i
n
de
t
e
rm
i
n
at
i
on of
cl
ust
e
r head
.
The
SEP
[
4
]
i
n
t
r
o
duce
d
t
h
e
h
e
t
e
ro
genei
t
y
i
n
t
h
e
net
w
o
r
k
b
y
m
a
ki
ng t
w
o
t
y
pes o
f
n
ode
s
i
.
e.
n
o
rm
al
an
d adv
a
n
ced h
a
v
i
ng
d
i
fferen
t
en
erg
i
es
an
d pro
b
a
b
ilities to
b
e
g
e
ttin
g
elected
as clu
s
ter
h
e
ad
. Th
is
h
e
tero
g
e
n
e
ity is ex
trem
ely h
e
lp
fu
l to m
a
k
e
th
e lifeti
m
e o
f
n
e
twork b
e
tter
[8].
The DEEC [5] used t
h
e ratio of resi
dual and av
e
r
age e
n
ergy in calculating t
h
e thres
hold for the
electio
n
of cl
uster h
e
ad
wh
ich
im
p
r
o
v
ed
t
h
e SEP
fu
rth
e
r. Bu
t th
is also
fram
e
d
no
n-o
p
ti
m
a
l clu
s
ter in
the
net
w
or
k.
The E
D
EEC
[
6
]
i
n
creased t
h
e
het
e
r
oge
nei
t
y
by
o
n
e m
o
re n
ode i
.
e
.
n
o
rm
al, ad
vance
d
a
n
d
supe
r n
o
d
e
in
th
e
n
e
twork. The norm
a
l
h
a
s lowest en
erg
y
an
d pro
b
a
b
ility to
b
ecome clu
s
ter h
e
ad
in th
e cl
u
s
ter. Th
i
s
al
go
ri
t
h
m
co
m
b
i
n
e
d
as
pect
s
of
het
e
r
oge
nei
t
y
and
rat
i
o
o
f
resi
d
u
al
an
d a
v
era
g
e e
n
er
gy
per
ro
u
nd t
o
i
m
prove
th
e stab
ility p
e
riod
an
d lifetime o
f
t
h
e
n
e
two
r
k
.
TADEEC [7] routing algorithm
used
the bes
t
of heteroge
ne
ity and
TEEN [8] to incr
ease the
chance
s
of
hi
g
h
ene
r
g
y
no
des t
o
be
com
e
cl
ust
e
r head
s m
o
re
than the low energy nod
es.
A
l
so i
t
rem
oved t
h
e
red
u
nda
ncy
o
f
t
h
e dat
a
at
cl
ust
e
r
heads a
n
d sens
o
r
n
o
d
e
s
by
im
pl
em
enti
ng t
h
e t
w
o t
h
r
e
sh
ol
ds i
.
e.
ha
rd a
nd
soft
t
h
re
sh
ol
d
so t
h
at
sam
e
dat
a
don
’t
have
t
o
be pro
p
a
g
a
t
ed t
o
t
h
e cl
ust
e
r heads a
nd
basest
at
i
on [
9
]
.
Thi
s
use
d
f
o
u
r
het
e
ro
g
o
n
o
u
s n
o
d
e
s
i
n
t
h
e
net
w
o
r
k i
.
e
.
n
o
r
m
a
l
,
adv
a
nce
d
,
su
per a
n
d s
upa
d
v
ance
d
ha
vi
n
g
t
h
ei
r
increasing e
n
e
r
gies and
probab
ilities o
f
cho
o
s
i
n
g
as clu
s
ter h
ead
s i
n
the n
e
twork
.
This o
u
t
p
e
rfo
r
m
e
d
th
e
LEAC
H
,
SEP,
DEEC and
EDEEC in
term
s of stab
ility p
e
ri
od
, lifetim
e an
d
d
a
ta capacity of th
e n
e
t
w
ork.
The
[
10]
use
d
t
h
e
wel
l
k
n
o
w
n cl
ust
e
ri
n
g
t
e
chni
que
based
on
f
u
zzi
ne
ss i
.
e. f
u
zzy
c
-
m
e
ans cl
ust
e
ri
n
g
to
fram
e th
e clu
s
ters in
t
h
e
network. Th
e
op
ti
m
a
l a
nd e
qualized clusters
are
fram
e
d by
the FCM al
gorith
m
.
The
Fuzzy c
-
means clusteri
ng basica
lly as
sociates to ea
c
h
node a
corre
sponding m
e
mbers
h
ip val
u
e t
o
eac
h
cluster num
b
er which
have
to be form
ed. The clus
te
rs
are form
ed according to the highest de
gree of
bel
o
ng
ness
(al
s
o
kn
o
w
n
as
deg
r
ee
of
rel
a
t
i
ons
hi
p)
t
o
a
part
i
c
ul
a
r
cl
us
t
e
r n
u
m
b
er.
A
f
t
e
r t
h
e
cl
ust
e
rs are
fo
rm
ed t
h
e cl
ust
e
r hea
d
s are
cho
s
en
base
d o
n
t
h
e m
a
xim
u
m
resi
dual
ene
r
gy
n
o
d
e am
ong t
h
e cl
ust
e
r m
e
m
b
er
s
and al
s
o
i
t
s
p
r
o
x
i
m
it
y
t
o
corres
p
on
di
n
g
cl
ust
e
r ce
nter.
The cluste
r he
ads furt
her c
o
mmunicate the data
collected through the cl
us
t
e
r
m
e
m
b
er sens
or
n
ode
s t
o
t
h
e
b
a
sest
at
i
on.
[11] Realized the fuzzy c-m
e
a
n
s
cl
ust
e
ri
ng al
go
ri
t
h
m
on wi
r
e
l
e
ss senso
r
ne
t
w
o
r
k
.
Thi
s
pap
e
r real
i
zed
t
h
e 50 se
ns
or
no
des i
n
t
h
e
har
d
wa
re ha
vi
n
g
Ti
ny
O
S
ope
rat
i
n
g sy
st
em
.Thi
s har
d
ware i
m
pl
em
ent
a
t
i
o
n
success
f
ully prove
d
the e
f
fectiveness
of FCM in wi
reless se
ns
or network
a
n
d this
im
ple
m
entation
out
per
f
o
r
m
e
d t
h
e LE
AC
H.
[1
2]
Pr
op
ose
d
a
f
u
zzy
-ba
s
ed
sim
u
l
a
t
i
on sy
st
em
for se
nso
r
net
w
or
ks a
nd
cal
cul
a
t
i
on o
f
t
h
e l
i
f
et
im
e
of
a se
ns
or
by
con
s
i
d
eri
n
g
t
h
e
rem
a
i
n
i
ng
bat
t
e
ry
power, sleep
tim
e rate an
d tran
sm
issio
n
time rate.
[13] Used the
fuzzy logic
mech
an
ism
in
h
e
terog
e
n
e
ous n
e
twork
where it ap
p
lied th
e en
erg
y
,
het
e
r
oge
nei
t
y
and
p
r
o
x
i
m
it
y t
o
base
st
at
i
o
n
fact
o
r
s i
n
fu
zzy in
feren
ce syste
m
in
d
e
termin
atio
n
of cl
u
s
ter
heads.
2.
R
E
SEARC
H M
ETHOD
In
t
h
i
s
pap
e
r,
we
have
a
ppl
i
e
d t
w
o m
e
t
h
o
d
s
fo
r
bet
t
e
rm
ent
o
f
t
h
e
wi
r
e
l
e
ss sens
o
r
n
e
t
w
o
r
ks
. T
h
e
fuzzy
c
-
m
eans cl
ust
e
ri
ng
w
h
i
c
h h
a
s
been
u
s
ed t
o
cl
u
s
t
e
r
t
h
e se
nso
r
n
o
d
e
s an
d
ne
ural
net
w
or
ks
w
h
i
c
h
ha
s
been
u
s
ed
t
o
t
a
ke t
h
e
deci
si
on
of
el
ect
i
o
n
o
f
cl
ust
e
r
head
a
m
ong t
h
e cl
ust
e
r m
e
m
b
ers i
n
t
h
e cl
ust
e
r
.
T
h
e FC
M
alg
o
r
ith
m
is ill
u
s
tr
ated in f
i
gu
r
e
1. Th
e
d
e
gr
ee
o
f
b
e
l
o
ngness of
no
d
e
i t
o
cluster
j
is giv
e
n
b
y
. For e
ach
no
de
i
a
n
d cl
u
s
t
e
r
j,
is calculated
an
d m
a
x
i
m
u
m
v
a
lu
e is
selected
to
wh
ich
t
h
e nod
e
b
e
lo
ng
s. In a cluster
t
h
e cl
ust
e
r
hea
d
i
s
t
h
e
dom
i
n
ant
n
ode a
s
t
h
e
fu
rt
he
r com
m
uni
cat
i
o
n t
o
t
h
e basest
at
i
o
n
wo
ul
d
be
do
ne
by
t
h
e
cl
ust
e
r
hea
d
.
Seve
ral
fact
or
s ha
ve t
o
be
con
s
i
d
ere
d
w
h
i
l
e
el
ect
i
ng t
h
e cl
ust
e
r
hea
d
s. I
n
a
het
e
r
o
gene
o
u
s
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
A Hy
bri
d
A
ppr
oac
h
of
F
u
zzy
C
-
Mea
ns C
l
ust
e
ri
ng
a
n
d
N
e
u
r
al
N
e
t
w
ork t
o
Make E
n
er
gy-
…
(Amit Kr Kau
s
h
i
k)
67
6
n
e
two
r
k
th
e sen
s
or nod
es h
a
v
e
d
i
fferen
t energ
i
es an
d
p
r
ob
ab
ilities o
f
gettin
g
elected
as clu
s
ter h
e
ads. Th
e
het
e
r
oge
nei
t
y
f
act
or
pl
ay
s a i
m
port
a
nt
r
o
l
e
i
n
el
ect
i
o
n
of t
h
e cl
ust
e
r
hea
d
.
Fi
gu
re 1.
The
FC
M
al
go
ri
t
h
m
In
th
is p
a
p
e
r t
h
e h
e
tero
gen
e
i
t
y facto
r
is cal
cu
lated
fro
m
t
h
e p
r
ob
ab
ility
o
f
g
e
tting
elected
as clu
s
ter
h
ead
an
d
en
erg
y
o
f
th
e
n
o
d
e
, So
h
i
g
h
e
r energ
y
nod
e wou
l
d
h
a
v
e
larg
er p
r
ob
ab
ab
ility o
f
elected
as clu
s
ter
head
an
d
ul
t
i
m
a
t
e
l
y
woul
d
ha
ve
hi
g
h
er
het
e
r
oge
nei
t
y
fact
o
r
. The
resi
dual
ener
gy
al
so
ha
s a si
g
n
i
f
i
cant
rol
e
i
n
election of cluster hea
d
. T
h
e
residual energy is the
ener
gy
l
e
ft
i
n
t
h
e
no
de aft
e
r co
m
m
uni
cat
i
on has be
e
n
do
ne. A
n
ode havi
ng hi
g
h
re
si
dual
ene
r
gy
wo
ul
d have
m
o
re c
h
ances to be chose
n
as
cluster hea
d
than low
ener
gy
no
de a
s
com
m
uni
cat
ion
nee
d
s
di
ssi
pat
i
o
n
o
f
e
n
e
r
gy
. T
h
e
Ene
r
g
y
req
u
i
r
e
d
t
o
sen
d
t
h
e
pac
k
et
s i
s
p
r
op
ortio
n
a
l to th
e d
i
stan
ce between
th
e senso
r
n
d
o
e
s. So
th
e n
o
d
e
wh
ich
is at s
m
a
lles
t
d
i
stan
ce from th
e
basest
at
i
o
n
am
on
g
al
l
cl
ust
e
r
n
o
d
e
s s
h
oul
d
be
el
ect
ed as
cl
ust
e
r
hea
d
i
n
t
h
e cl
ust
e
r.
So
i
n
t
h
i
s
pape
r
t
h
es
e
three fact
ors i.e. Heterogenei
t
y of
the node, residual ene
r
gy of the
node
and di
st
ance
t
o
t
h
e basest
at
i
on
a
r
e
considere
d
while electing the cluster hea
d
. The
neur
al
net
w
or
k w
o
rk
s as art
i
f
i
c
i
a
l
hum
an brai
n
whi
c
h
b
a
sically tak
e
th
e d
ecision
based
on
th
e g
i
v
e
n
inpu
ts and ap
p
lied
wei
ghts. Th
e weigh
t
ap
p
lied
to
an
in
pu
t
sig
n
i
fies th
e i
m
p
o
r
tan
ce of th
at in
pu
t in
tak
i
ng
th
e d
ecisi
o
n
o
r
calcu
lati
n
g
th
e
o
u
t
pu
t o
f
th
e
n
e
two
r
k. In
th
is
app
r
oach
, t
h
e
neu
r
al
net
w
o
r
k
has t
a
ke
n t
h
e
s
e fact
o
r
s as i
n
put
fo
r eac
h n
ode i
n
t
h
e c
o
r
r
e
sp
on
di
n
g
cl
us
t
e
r an
d
appl
i
e
d
t
h
e
wei
ght
s
o
n
t
h
ese
f
act
ors a
n
d s
u
m
of
p
r
o
d
u
ct
o
f
i
n
p
u
t
fact
ors
an
d
wei
g
ht
s t
a
ke
deci
si
o
n
o
f
el
e
c
t
i
n
g
as cl
ust
e
r hea
d
or n
o
t
.
T
h
e i
n
put
s
has di
f
f
er
ent
rol
e
s i
n
cal
cul
a
t
i
on
of t
h
e
out
put
as o
n
e i
n
p
u
t
has
posi
t
i
ve r
o
l
e
wh
ile o
t
h
e
r h
a
s
n
e
g
a
tiv
e ro
le.
Th
e weigh
t
s are d
ecid
e
d
b
a
sed
on
th
e
how p
o
s
itiv
ely or n
e
g
a
tiv
ely facto
r
is
affect
i
n
g t
h
e
d
eci
si
on
or
out
p
u
t
o
f
t
h
e
net
w
or
k. T
h
e
wei
g
ht
i
s
ne
gat
i
v
e
whe
n
i
t
s
val
u
e
negat
i
v
el
y
af
f
ect
s t
h
e
deci
si
o
n
of c
h
oosi
n
g as cl
ust
e
r hea
d
suc
h
i
n
case of di
st
an
ce of n
ode
fr
o
m
basest
at
i
on as we have t
o
s
u
p
p
r
ess
th
e ro
le of prod
u
c
t
o
f
i
n
pu
t an
d
weigh
t
in
su
mmatio
n
and
p
o
s
itiv
e i
n
case o
f
h
e
terog
e
neity an
d
en
ergy as th
eir
h
i
gh
v
a
lu
es lead
to b
e
tter clu
s
ter h
e
ad
s.
3.
RESULTS
A
N
D
DI
SC
US
S
I
ON
In t
h
i
s
pr
o
j
ect
,
we ha
ve si
m
u
l
a
t
e
d ou
r ap
p
r
oach i
n
m
a
tl
ab.Ar
ea i
s
t
a
ken
as 10
0
*
1
0
0
s
q
ure m
e
t
e
rs
whe
r
e the
base
station is locat
ed at (5
0,50)
.
Th
e sen
s
o
r
nod
es ar
e r
a
n
domly
distributed in the fiel
d around the
basest
at
i
o
n
.
T
h
e l
o
cal
i
zat
i
on t
echni
que
has
n
o
t
bee
n
us
e
d
i
n
ou
r net
w
o
r
k
.
The n
o
d
es are
di
st
ri
b
u
t
e
d m
a
nual
l
y
aro
u
nd t
h
e
bas
e
st
at
i
on an
d l
o
cat
i
ons
of t
h
e
sens
or
n
odes
h
a
ve
been
k
n
o
w
n i
n
a
dva
nce
d
u
ri
n
g
t
h
e ne
t
w
o
r
k
pr
ocessi
ng
. Th
e no
des
f
o
rm
the cl
us
te
r and the elected cl
uster
hea
d
am
ong cluster m
e
m
b
ers aggrega
t
e the
dat
a
fr
om
al
l
the m
e
m
b
ers and se
n
d
i
t
fu
rt
her t
o
t
h
e
base
st
at
i
on.
We ha
ve us
ed t
h
e he
t
e
ro
gene
o
u
s ne
t
w
o
r
k
whe
r
e we ha
ve
fou
r
t
y
pes of
no
des i
.
e n
o
rm
al
, adva
nced
, supe
r an
d su
pa
d
v
ance
d. T
h
ese
no
des ha
ve en
ergi
es
in
th
is
way [7
]:
no
rm
al
=Eo, ad
vance
d
=
E
o
(
1
+
a),s
upe
r=E
o
(
1
+b
)
supadva
n
ced=
E
o(1+c
)
where a<b<c
The
di
st
ri
b
u
t
i
o
n
of
n
o
d
es i
s
d
one
u
s
i
n
g m
and m
0
[
7
]
.
A
s
=
∗
(1
−
);
=
∗
m ;
=(
0
∗
∗
)/2;
=(
0
∗
∗
)/2;
We
have
used two m
obility
models i.e.
free
space a
n
d m
u
lti
path m
odel as
use
d
in [14].
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
67
4 – 6
8
1
67
7
∗
∗
∗
0
(2
)
∗
∗
∗
0
(3
)
Eq
.
(1) is app
lied
in
free sp
ace n
e
two
r
k
m
o
del an
d
Eq
(2
) i
s
ap
p
lied
in
mu
ltip
ath
n
e
t
w
ork
m
o
d
e
l wh
ere d
0
is
calculated as:
0
(4
)
And
l is th
e size o
f
t
h
e m
e
ssa
g
e
,
i
s
t
h
e ener
gy
re
qui
re
d t
o
t
r
ansm
i
t
t
h
e m
e
ssage a
nd
d c
a
n be t
h
e di
st
a
n
ce
bet
w
ee
n
t
h
e
n
ode
s or
n
ode
and
basest
at
i
o
n.
T
h
e
is th
e am
p
lificatio
n
en
erg
y
for the
free
space
m
odel
wh
ile
is the amplification e
n
e
r
gy in m
u
ltipath network m
odel. The
hete
roge
neity of each node
is
give
n as
no
rm
al
:
P*E(
i)/ter;
adva
nce
d
:
P*(1+a)*E(i)/ter;
super:
P*(1+
b)*E(i)/ter;
su
padv
an
ced
:
P*(1+c)*E(i)/ter;
w
h
er
e
ter=(1+a*m-m*m0*(a-((b+c)/2)));
In these c>
b>a
;
whe
r
e c=
1; al
so
Th
e sim
u
latio
n
p
a
ram
e
ters h
a
v
e
b
e
en
in
t
a
b
l
e 1
.
Th
e tab
l
e lists o
u
t
all th
e v
a
lu
es
for th
e in
itial
ener
gy
,
t
h
e no
des perce
n
t
a
g
e
i
n
ne
twork, the size of
packe
t
s etc.
The
re
sul
t
s
o
f
ou
r
a
p
p
r
oac
h
h
a
s
bee
n
sh
o
w
n i
n
fi
g
u
re 3, 4
a
nd 5. The
se
ns
or n
odes f
o
rm
cl
ust
e
rs
a
n
d
clu
s
ter
m
e
m
b
e
r
s send
th
eir
sen
s
ed
d
a
ta to
th
eir
cor
r
e
spond
ing
clu
s
ter
h
e
ad
s. Th
e clu
s
t
e
r
h
e
ad
s th
en
fu
r
t
h
e
r
gi
ve t
h
i
s
dat
a
t
o
t
h
e
basest
at
i
o
n
.
T
h
e dat
a
s
e
ndi
ng
has
bee
n
m
oni
t
o
re
d i
n
t
w
o
way
s
i
.
e.
t
h
e dat
a
se
nt
b
y
t
h
e
no
des
pe
r r
o
un
d t
o
t
h
e cl
ust
e
r hea
d
s a
n
d
da
t
a
sent
pe
r r
o
u
nd
by
t
h
e
cl
ust
e
r hea
d
s t
o
t
h
e
basest
at
i
o
n. T
h
ese
th
ro
ugh
pu
ts h
a
v
e
b
e
en
sh
own in
f
i
gu
r
e
6
and
7.In
th
is
pap
e
r we
have c
o
m
p
ared o
u
r
pr
op
ose
d
ap
p
r
oa
ch t
o
t
h
e
two
d
i
fferen
t
ap
pro
ach
es
p
r
esen
ted
in
literat
u
re su
rv
ey
i.e.
FCM o
n
l
y and LEACH al
g
o
rith
m
.
W
e
h
a
v
e
tak
e
n
di
ffe
re
nt
pa
ra
m
e
t
e
rs on
whi
c
h we
ha
ve c
o
m
p
ared o
u
r
al
g
o
rith
m
with
th
ese
p
r
es
e
n
t a
p
proaches
. T
h
ese are
alive nodes i.e. nodes which a
r
e alive in the networks
and it has been trac
ed in each
round, dea
d
nodes
whic
h
have
died i
n
the networks in
each ro
und,
da
ta capacity of the clusters a
nd throughput of th
e network i.e. the
packet
s se
nt
p
e
r r
o
u
n
d
t
o
t
h
e
basesst
at
i
on a
nd cl
ust
e
r
hea
d
s. T
h
e a
n
al
y
s
i
s
t
a
ki
ng
di
ffe
r
e
nt
par
a
m
e
t
e
rs base
d
on
fi
gu
res
gi
ve
n
has
bee
n
do
n
e
i
n
t
h
e
f
o
l
l
o
wi
ng
p
o
i
n
t
s
.
1.
In
fi
g
u
r
e
3,
we
ha
ve
pl
ot
t
e
d t
h
e
dat
a
t
h
at
ha
s bee
n
se
nt
by
t
h
e cl
u
s
t
e
r m
e
m
b
ers t
o
t
h
ei
r
cor
r
es
po
n
d
i
n
g
cluster heads
with the i
n
cre
a
sing
rounds.
We ha
ve
ob
se
rve
d
that
data
has
higly aggregated at cluster
head i
n
o
u
r a
p
pr
oac
h
. T
h
ese
dat
a
ag
gre
g
at
i
on i
d
de
pen
d
e
n
t
up
o
n
t
h
e l
i
v
el
i
n
ess of t
h
e cl
ust
e
r m
e
m
b
er
no
des a
n
d
O
u
r a
p
pr
oac
h
i
s
t
r
y
i
ng t
o
n
o
t
el
ect
t
h
e l
o
w
ene
r
gy
n
o
d
e as cl
ust
e
r
hea
d
as m
a
jor
co
mm
u
n
i
catio
n
is do
n
e
b
y
th
e th
is nod
e.
Th
e thr
oug
hput f
o
r
cluster
lev
e
l d
a
ta h
a
s al
so
b
e
en
o
b
s
erved
and
pl
ot
t
e
d i
n
Fi
gu
re
6. T
h
i
s
t
h
r
o
u
g
h
p
u
t
i
s
t
h
e h
o
w
m
a
ny
packet
s
bei
n
g
gene
rat
e
d
by
a
si
ngl
e
no
de i
n
a
net
w
or
k.T
h
i
s
p
a
ram
e
t
e
r show
t
h
e avera
g
e d
a
t
a
send by
a c
l
ust
e
r m
e
m
b
er
no
de as pa
rt
of
a net
w
o
r
k
s
.
I
n
bot
h
ob
ser
v
at
i
o
sn
we
h
a
ve
o
u
t
p
e
r
f
o
rm
ed t
h
e ot
he
r t
w
o
r
t
e
chni
que
s em
pl
oy
ed i
n
past
.
2.
The sec
o
n
d
pa
ram
e
t
e
r bei
ng
t
a
ken
fo
r c
h
ec
ki
n
g
t
h
e e
ffi
ci
ency
o
f
o
u
r
ap
pr
oac
h
i
s
t
o
c
o
u
n
t
h
o
w
m
a
ny
n
o
d
e
s are still
aliv
e in
th
e n
e
twork
as effiecien
t p
r
o
t
o
c
o
l
always
m
a
in
tain
h
i
g
h
aliv
es in n
e
twork. Th
i
s
param
e
t
e
r i
s
cal
c
ul
at
ed o
n
l
e
vel
of e
n
er
gy
no
de ar
e
havi
ng as
no
de ha
vi
n
g
n
o
bat
t
e
r
y
or sen
d
i
n
g
o
r
receiving ca
pa
bities are consi
d
ere
d
dead as t
h
ey are
a
b
le to do furt
her c
o
mmuni
cation.
We ca
n see that
perce
n
t
a
ge
o
f
a
l
i
v
e n
odes
i
n
o
u
r al
go
ri
t
h
m
out
pe
rf
orm
e
d t
h
e FC
M
o
nl
y
an
d LE
AC
H
al
g
o
r
i
t
h
m
s
i
n
fi
gu
r
e
4
.
In
o
u
r
algor
ith
m
ev
en
m
o
r
e
th
an 50
% nodes ar
e alive in
1
0
k
ro
und
s.
3
.
Th
e
f
i
gu
r
e
5
sh
ow
t
h
e sam
e
case of
d
e
ad
no
d
e
s so
on
ly 20
-30
% nod
es
h
a
v
e
b
een d
i
ed
i
n
1
0k ro
unds
whe
r
e i
n
LE
ACH all nodes
have
been de
ad
and FCMo
nl
y
app
r
oach
has
6
0
-
7
0%
no
des
h
a
ve
been
de
ad
.
4.
The
dat
a
se
nt
t
o
t
h
e
ba
sest
at
i
o
n
by
t
h
e cl
ust
e
r
heads
ha
s al
so
bee
n
m
oni
t
o
re
d a
n
d i
t
i
s
o
b
ser
v
e
d
t
h
at
t
h
e
t
h
r
o
u
g
h
p
u
t
of
net
w
o
r
k
u
s
i
n
g
ou
r a
p
pr
oac
h
i
s
hi
g
h
e
r
t
h
an
ot
he
r a
p
pr
oache
s
.
The
c
l
ust
e
r
hea
d
s a
r
e
ag
gr
eg
atin
g the d
a
ta co
llected
f
r
o
m
1
0
cluster
m
e
m
b
er
s o
f
th
eir
co
rr
espond
ing
cluster
.
5
.
Th
e op
tim
al
a
n
d
sim
ilar size
d
clu
s
ters cap
c
b
ility
o
f
o
u
r ap
pro
ach
can
been
ch
eck
e
d
fro
m
fig
u
r
e
8
.
Th
e
nodes in each
clusters
have been pl
otted as num
ber of nodes in each
clust
e
r in similar
in nature and also
no
n
-
o
p
t
i
m
al
i
t
y i
ssue has bee
n
res
o
l
v
e
d
at
hi
ghe
r ext
e
nt
. I
n
o
u
r net
w
o
r
k we ha
ve t
a
ken
10 cl
ust
e
rs
a
n
d
FC
M
i
s
used t
o
assi
g
n
t
h
e
no
des t
o
eac
h cl
u
s
t
e
r. Ve
ry
few
cl
ust
e
rs are s
h
owi
ng
n
o
n
-
si
m
i
l
a
ri
t
y
i
n
count
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
A Hy
bri
d
A
ppr
oac
h
of
F
u
zzy
C
-
Mea
ns C
l
ust
e
ri
ng
a
n
d
N
e
u
r
al
N
e
t
w
ork t
o
Make E
n
er
gy-
…
(Amit Kr Kau
s
h
i
k)
67
8
of
t
h
ei
r cl
ust
e
r
m
e
m
b
er no
de
s. I
n
o
u
r
net
w
or
k i
n
eac
h cl
ust
e
r
has
bee
n
assi
g
n
ed
ave
r
age
8-
1
0
cl
sut
e
r
no
des
Tabl
e
1.
Si
m
u
lat
i
on Pa
ram
e
t
e
rs f
o
r t
h
e
net
w
or
k
Para
m
e
ters
Value
Ar
ea 100*
100
sq
uar
e
m
e
ter
s
Basestation (
50,
50)
(
i
n
m
)
Initial Energ
y
0.5J
T
r
ans
m
ission E
n
ergy
50nJ/bit
Receiver Energy
50nJ/bit
No Of Nodes
100
Free space
A
m
p
E
n
ergy
10pJ/bit/
Multipath A
m
plification Energy
0.
0013
pJ/bit/
M
e
ssage Size(
B
)
4000 bits
Round
1000
0
Aggregation Energy
5nJ/bit/packet
P 0.
10
m 0.
5
m
0
0.
2
Fi
g
u
r
e
3. T
h
e
dat
a
capaci
t
y
of
t
h
e
net
w
or
k
usi
n
g
di
ffe
re
nt
ap
pr
o
aches at
cl
ust
e
r
l
e
vel
Fig
u
r
e
4
.
A
liv
e no
d
e
s
v
s
round
s i
n
d
i
f
f
e
r
e
n
t
ap
pro
ach
es
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
67
4 – 6
8
1
67
9
Fig
u
r
e
5
.
D
e
ad no
d
e
s
v
s
round
s i
n
d
i
f
f
e
r
e
n
t
ap
pro
ach
es
Fi
gu
re 6.
The
t
h
r
o
ug
h
put
of
t
h
e net
w
or
k o
b
s
e
rve
d
at
basest
at
i
o
n
Fi
gu
re
7.
The
t
h
r
o
ug
h
put
o
b
s
e
rve
d
per
n
o
d
e
i
n
t
h
e
net
w
o
r
k
0
1
2
3
Throughput(Pac
ket
send
per
node)
Proposed
FCMonly
LEACH
0
1
2
3
4
5
6
7
Throughput(Pac
kets
send
per
round)
Proposed
FCMonly
LEACH
Evaluation Warning : The document was created with Spire.PDF for Python.
I
J
ECE
I
S
SN
:
208
8-8
7
0
8
A Hy
bri
d
A
ppr
oac
h
of
F
u
zzy
C
-
Mea
ns C
l
ust
e
ri
ng
a
n
d
N
e
u
r
al
N
e
t
w
ork t
o
Make E
n
er
gy-
…
(Amit Kr Kau
s
h
i
k)
68
0
Fi
gu
re
8.
The
C
l
ust
e
r m
e
m
b
er
no
des i
n
eac
h
cl
ust
e
r
4.
CO
NCL
USI
O
N
The
propose
d
approach use
d
the
fuzzy C-m
eans clusteri
ng In eac
h of
th
e
clu
s
ter ou
r al
go
rith
m
tried
t
o
ha
ve si
m
i
l
a
r
si
zed cl
sut
e
rs
of se
ns
or
n
o
d
e
s
an
d al
so
ne
ur
l
net
w
or
ks,
usi
ng t
h
e
di
ffe
re
n
t
fact
ors
ha
ve e
l
ect
ed
t
h
e cl
ust
e
r hea
d
s. O
u
r app
r
o
ach w
o
r
k
ed
o
n
o
p
t
i
m
a
l
i
t
y
o
f
cl
ust
e
r an
d p
e
rsi
s
t
e
ncy
of
n
ode
s i
n
net
w
o
r
k an
d
success
f
ully outpe
r
form
ed th
e pre
s
ent a
p
proache
s
. T
h
e t
h
roughput at both cluster and
network le
vel is highly
app
r
eci
at
i
ng a
nd s
h
owi
ng ef
f
ect
i
v
eness o
f
u
s
i
ng t
h
i
s
hy
bri
d
ap
pr
oac
h
. Di
ffe
rent
fat
o
rs
p
l
ay
a do
m
i
nat
rol
e
i
n
sel
ect
i
ng cl
ust
e
r
head
as
o
n
l
y
on
e f
act
or
c
a
n
n
o
t
j
u
dge
a
node as
clust
e
r
head a
n
d also t
h
e
probabi
listic
pers
pect
i
v
e
n
o
t
assure
u
s
a
b
o
u
t
sel
ect
i
n
g
a
go
o
d
cl
ust
e
r
h
ead s
o
we ca
n
say
ou
r a
p
pr
oa
ch
has t
a
ke
n c
a
re al
l
these fact
ors
a
n
d worke
d
on
these a
n
d succ
essfully a
p
proached that e
ffi
cient r
out
i
n
g
t
echni
que
. Ou
r fut
u
r
e
work
will b
e
t
o
co
m
b
in
e th
e
o
t
h
e
r clu
s
teri
ng
and
d
ecision ap
pro
ach
es i
n
sen
s
o
r
n
e
two
r
k
s
an
d
see
how th
ey
affect
our a
p
proach.
REFERE
NC
ES
[1]
W
.
Heinze
lm
an,
A. Chandrakas
an, and
H
.
Balakrishnan, "En
e
r
g
y
-
efficient
communication pro
t
ocol for wireless
sensor networks", in the Proceeding of the Hawaii Inte
rn
ation
a
l Conferen
ce S
y
s
t
em
S
c
ien
ces
,
Hawaii, J
a
nu
a
r
y
2000.
[2]
W.
R.
He
i
n
ze
l
m
an,
A.
P.
Cha
ndraka
s
a
n
,
H.
B
a
lakr
ishnan, “An app
licationspecifi
c protocol architecture
for
wir
e
less
microsensor networks”, IEEE
Tr
ansactions on W
i
reless Comm
uni
cations, 1
(4) (2
002) 660-670.
[3]
Rupesh Mehta,
Abhishek Pandey
& Pr
atik
Kap
a
dia, “Reforming Clusters
Using C-LEACH in
Wireless Sensor
Networks”, In International Conference on Comp
uter
Communication and Informatics (ICCCI -2012), Jan. 10 – 12,
2012, Coimbator
e
, INDIA.
[4]
G.
Sma
r
a
gda
kis, I.
Ma
tta,
A.
Be
sta
v
ros,
“SEP: A Sta
b
le
Election Protocol for cluste
r
e
d hetero
gene
ous wireless
sensor networks”, in
: Second
In
tern
ational Workshop on Sensor and. A
c
tor
Network Protocols
and Applicatio
ns
(SANPA 2004),
2004.
[5]
L. Qing
, Q.
Zh
u, M. Wang
, "
D
esign of a d
i
stributed
en
erg
y
-
e
ffic
i
ent
clus
t
e
ri
ng al
gorithm for heterogeneous
wireless sensor
networks",
ELSEVIER, Computer Communicat
ions, 29, 2006, p
p
2230- 2237
.
[6]
Parul Saini, A
j
ay
.K.Sharma,
“E-DEEC- Enh
a
nced
Dist
ributed Energ
y
Eff
i
cient Clusterin
g
Scheme for
heterog
e
neous
WSN”, in: 2010 1st International Conference
on
Parallel, Distr
i
b
u
ted a
nd Grid C
o
mputing (PDGC -
2010).
[7]
Anam
ika Chauh
a
n and Am
it K
a
ushik, “
T
ADE
EC: Thr
e
shol
d
Sensitive Advan
ced Distribu
ted
Energ
y
Eff
i
c
i
en
t
Clustering Routing Protocol fo
r Wireless Sensor
Networks”, Inter
n
ation
a
l
Journal
of Computer Ap
plications (0975
– 8887), Volume 96 No. 23 June
2014.
[8]
Amit K.
Kaushik,
“Performance Eval
uation of
Proactiv
e and
Reactive Rou
tin
g Protocols in
Wireless Senso
r
Networks”, In
In
ternational Journ
a
l
of
Computer
Applications (09
75 – 8887)
, Volume 110 – No. 1
6
, Januar
y
2015
.
[9]
A.
Manjeshwar
and D.
P. Agarwa
l, "TEEN:
a rou
ting pro
t
ocol for
enhan
ced
eff
i
ciency
in
wir
e
less sensor
networks",
In 1st Internatio
nal Workshop o
n
Parallel and
Distribut
ed
Computing Issues in Wi
reless Networks and Mobile
Computing, Apr
il 2001
.
[10]
D.
C.
Hoang,
R. Kumar,
Memb
er a
nd S.K. Panda, “Fuzzy
C-
Means Cluste
ring Protocol for
Wireless Sensor
Networks”, In
I
EEE Intern
at
ion
a
l S
y
mposium o
n
Industria
l
Electronics (ISIE)
,
2
010.
1
2
3
4
5
6
7
8
9
10
0
10
20
30
40
50
60
70
80
90
10
0
Cl
u
s
t
e
r
Nu
m
b
e
r
C
l
u
s
t
e
r
m
e
m
ber
N
ode
Evaluation Warning : The document was created with Spire.PDF for Python.
I
S
SN
:
2
088
-87
08
IJEC
E V
o
l
.
6, No
. 2, A
p
ri
l
20
16
:
67
4 – 6
8
1
68
1
[11]
Duc Chinh Hoang, Rasjesh Kumar, San
jib
Ku
mar Panda, “Realisation of
a cl
uster-based pro
t
ocol using
fuzzy C-
means algorithm for wir
e
less sen
s
or networks”, I
n
IET Wi
reless.
Sens. S
y
st., 201
3,
Vol. 3
,
Iss. 3
,
pp. 163–171
.
[12]
Q
i
W
a
ng, K
u
lla,
E.,
M
i
no
, G
.
,
B
a
rolli
, L.
“Prediction of Sensor Lifetime in
Wireless
Sensor Netwo
r
ks Using
Fuzzy
Logic”, In IEEE 28th International Conferen
ce on Adva
nced Information Networking and Ap
plications (AINA),
2014.
[13]
Pa
l R.
,
Sha
r
ma,
A.
K.
,
“FSE
P-E
:
E
nha
nce
d
sta
b
le e
l
ec
tion
pro
t
oco
l
based on fuzzy Logic for
cluster head selection
in WSNs”, 2013
Sixth In
tern
atio
nal Conf
er
en
ce
of Contemporar
y
Computing
(I
C3).
[14]
T. R
a
ppapor
t, Wireless Communi
cation: Prin
ciples
and Practice, Englewood,
NJ: Prentice H
a
ll, 199
6.
[15]
MATLAB 7.4.0(
R2007a) www.mathworks.com
BI
O
G
R
A
P
HY
OF
A
U
T
HO
R
Amit Ku
mar Kaushik is working
as As
sistant prof
essor in department of computer scicen
ce and
techno
log
y
in M
a
nav Rachna University
, Farid
a
b
a
d,
Ha
ry
ana,
Indi
a
.
He
com
p
le
t
e
d his
M
.
T
ech
in Information s
y
stems from
Delhi Technolo
g
ic
al University, New Delhi and B.Tech from
UIET,
Kurukshetra University
.
Hi
s re
se
arc
h
inte
re
sts
inc
l
ude the wire
le
ss se
nsor ne
tworks,
Distributed
S
y
stems and LTE n
e
tworks.
Evaluation Warning : The document was created with Spire.PDF for Python.