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.
10
,
No.
5
,
Octo
be
r
2020
,
pp. 4
798~
4808
IS
S
N: 20
88
-
8708
,
DOI: 10
.11
591/
ijece
.
v10
i
5
.
pp4798
-
48
08
4798
Journ
al h
om
e
page
:
http:
//
ij
ece.i
aesc
or
e.c
om/i
nd
ex
.ph
p/I
J
ECE
Energy e
ffici
ent clu
ster
in
g and r
ou
ting opti
mizati
on
model f
or
maximi
zin
g life
ti
me of wi
re
l
ess s
en
sor netwo
rk
Savith
a
S
.
1
,
S.
C.
Li
ngared
d
y
2
, San
j
ay Ch
itnis
3
1
Depta
r
m
en
t
of
Com
pute
r
Scie
n
ce
and
Engi
ne
ering
,
CMR
Insti
tute
of
T
e
chnol
og
y
,
India
2
Depta
r
m
en
t
of C
om
pute
r
Scie
n
ce
and Engi
ne
ering
,
Sri V
enka
t
eshwara
Col
le
g
e
o
f
E
ngin
ee
ring
,
In
dia
3
Depta
r
m
en
t
of C
om
pute
r
Scie
n
ce
and Engi
ne
ering
,
Da
y
a
nand
a
Sagar
Univ
ersity
,
India
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
A
pr
12
, 201
9
Re
vised
A
pr 16
,
2019
Accepte
d
Apr 22
, 202
0
Rec
en
tly
,
th
e
wi
de
adop
ti
on
of
W
SNs
(W
ire
le
ss
-
Sensor
-
Networks)
is
bee
n
see
n
for
provisi
on
non
-
rea
l
t
ime
and
real
-
ti
m
e
appl
i
ca
t
i
on
servi
ce
s
such
as
int
ellige
n
t
tra
n
sportat
ion
and
hea
lt
h
ca
r
e
m
onit
oring,
int
el
l
ige
n
t
tra
nsporta
ti
on
e
tc
.
Provisioning
the
se
services
req
uire
s
energ
y
-
e
fficie
n
t
W
SN
.
The
cl
usteri
ng
techniqu
e
is
an
eff
icient
m
ec
han
ism
tha
t
pl
a
y
s
a
m
ain
r
o
l
e
i
n
r
e
d
u
c
i
n
g
t
h
e
e
n
e
r
g
y
c
o
n
s
u
m
p
t
i
o
n
o
f
W
S
N
.
H
o
w
e
v
e
r
,
t
h
e
e
x
i
s
t
i
n
g
m
o
d
e
l
is
designe
d
consid
eri
ng
r
educ
ing
e
ner
g
y
-
consum
pt
i
on
of
the
sensor
-
devi
c
e
for
the
hom
ogenous
net
work.
How
eve
r,
it
inc
urs
energ
y
-
ov
erh
ea
d
(E
O)
bet
wee
n
cl
uster
-
h
ea
d
(C
H).
Further,
m
axi
m
iz
ing
cove
r
a
ge
ti
m
e
is
not
conside
red
b
y
the
exi
sting
clus
te
ring
appr
oa
ch
consid
eri
ng
heteroge
neous
net
works
aff
ecting
l
ife
tim
e
per
form
ance.
In
ord
er
to
over
come
th
e
se
rese
ar
ch
cha
l
le
nges,
thi
s
work
pre
sents
an
ene
rg
y
eff
i
ci
en
t
cl
uste
ring
and
routi
ng
opti
m
iz
ation
(E
ECRO)
m
odel
adopt
ing
cro
ss
-
lay
er
d
esign
for
h
et
ero
g
ene
ous
net
works
.
Th
e
E
ECRO
uses
cha
n
nel
g
ai
n
informa
ti
on
from
th
e
ph
y
sic
al
lay
e
r
and
TDMA
base
d
comm
unic
at
io
n
is
adopt
ed
for
comm
unic
at
ion
among
both
int
ra
-
cl
uster
and
int
er
-
cl
ust
er
co
m
m
unic
at
ion.
F
urthe
r,
cl
ust
eri
n
g
and
rout
ing
opti
m
iz
ation
are
pre
sente
d
to
b
ring
a
good
trade
-
off
among
m
ini
m
iz
ing
the
en
erg
y
of
CH,
enha
n
ci
ng
co
ver
age
ti
m
e
and
m
axi
m
iz
ing
th
e
li
fe
ti
m
e
of
sensor
-
net
work
(SN
).
The
expe
riments
ar
e
condu
ct
ed
t
o
esti
m
ate
the
p
erf
orm
ance
of
EE
CRO
over
th
e
ex
isti
ng
m
odel
.
The
signifi
c
ant
-
per
form
anc
e
is
at
t
ai
ned
b
y
EECRO
over
the
exi
sting
m
odel
in
te
rm
s
of
mi
nimizi
ng
routi
ng
and
co
m
m
unic
at
ion
o
ver
hea
d
and
m
axi
m
iz
ing
the
li
fe
ti
m
e
of
W
SNs.
Ke
yw
or
d
s
:
Cl
us
te
rin
g
Cros
s
lay
er
Hop
Lifet
i
m
e
WSNs
Copyright
©
202
0
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
:
Savith
a
S
.
,
Dep
ta
r
m
en
t
of
Com
pu
te
r
Scie
nce a
nd E
ng
i
ne
erin
g
,
CM
R In
sti
tute
of T
ech
nolo
gy,
Ban
galor
e
, In
dia
.
Em
a
il
:
s.sav
it
ha4
47
@r
e
diff
m
ai
l.com
1.
INTROD
U
CTION
The
i
ncr
ease
d
gro
wth
of
se
ns
or
te
c
hnol
ogie
s
has
le
d
t
o
in
crease
d
a
doptio
n
of
W
SN
ac
r
os
s
diff
e
re
nt
for
prov
isi
on
i
ng
f
ut
ur
e
c
omm
un
icati
on
syst
e
m
s
and
wireless
-
ba
sed
ap
plica
ti
on
s.
F
or
e
xam
pl
e,
it
is
util
iz
ed
acro
s
s
diff
e
re
nt
areas
li
ke
in
dustria
l
m
anu
f
act
ur
e,
healt
h
-
care
-
m
on
it
or
i
ng,
an
d
intel
li
gen
t
trans
portat
ion,
[1
]
et
c.
Furthe
r
m
or
e,
the
WSN
has
bee
n
uti
li
zed
in
differe
nt
intel
li
gen
t
a
pp
li
cat
io
ns
,
no
n
-
real
and
real
-
ti
m
e
s
m
art
li
ke
ta
ct
ical
i
ntern
et
[
2
]
,
a
wear
a
ble
c
om
pu
ti
ng
de
vic
e
[
3
]
,
a
nd
sm
art
ci
ty
[
4
]
.
The
pr
im
e
respo
ns
ibil
it
y
of
WSNs
is
ac
cur
at
el
y
gathe
r
i
ng
us
ef
ul
data
and
s
ensi
ng
s
uch
a
s
yi
el
ding
sen
sed
big
da
ta
,
ai
r
qu
al
it
y,
bio
m
edical
,
hu
m
idit
y
m
easur
em
ents
and
chem
ic
al
inf
or
m
at
ion
for
fu
t
ur
e
a
naly
sis
[5
]
.
At
the
s
i
m
i
la
r
tim
e,
the
CC
(Cloud
-
Com
puti
ng
)
al
lo
we
d
te
chnolo
gies
li
ke
Cl
oud
-
RA
N
[
6
]
an
d
F
og
-
R
AN
[
7
]
a
nd
that
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
Ener
gy
ef
fi
ci
en
t cl
us
te
ring a
nd r
ou
ti
ng opti
miza
ti
on
m
od
el
for
…
(
Savi
th
a S
.
)
4799
pro
vid
e
WSN
s
with
c
omm
un
ic
at
io
n,
c
om
pu
ta
ti
on
ad
va
ntages
an
d
t
he
st
or
a
ge
res
ources
[
8],
as
well
as
prom
isi
ng
the t
echn
i
qu
e
to p
r
ocess
a
nd m
anag
e the
hu
ge
a
m
ou
nt of a
ggre
gated dat
a
[9
]
.
The
se
nsor
-
de
vice
(
SD
)
is
placed
in
a
haza
rdo
us
area
w
he
re
batte
ry
re
pl
aci
ng
a
nd
rec
hargin
g
a
re
no
t
possible
a
nd
al
so
hum
a
n
m
on
it
or
i
ng
con
ta
in
s
huge
-
risk.
T
he
S
D
can
be
ei
the
r
tim
e
or
eve
nt
-
dr
i
ve
n,
in
both
cases,
the
batte
ry
en
erg
y
is
ex
haus
te
d
ex
pone
ntial
ly
.
The
sense
d
data
is
ei
the
r
transm
it
te
d
t
o
base
sta
ti
on
(
BS
)
or
neig
hbor
i
ng
de
vices.
I
n
a
few
scen
arios,
sim
il
ar
data
can
be
transm
i
tt
ed
t
o
BS.
Th
us
,
a
ff
ect
in
g
ene
r
gy
-
ef
fici
ency
(
EE
)
of
the
W
S
N.
I
n
order
t
o
ov
e
rc
om
e
the
redundan
cy
pro
blem
s
and
pro
du
ce
a
m
or
e
ene
rg
y
-
e
ff
ic
i
ent
Data
A
ggr
egati
on
(
DA
)
m
et
ho
d
is
uti
li
zed
in
[
10
]
.
To
pro
visio
n,
t
he
acce
ss
of
real
-
ti
m
e
[1
1]
to
se
nsor
-
data
for
the
a
pp
li
cat
io
n
of
r
el
ia
ble
ind
ust
r
ie
s,
data
proc
essing,
an
d
ac
cur
at
e
gathe
rin
g
is
need
e
d
[
12
]
.
An
y
ways,
perform
ing
the
DA
posses
s
t
he
c
halle
ngin
g
s
olu
ti
on,
w
hich
i
s
represe
nted
i
n
[13].
I
n
[14],
represe
nting
the
sc
hem
e
of
energy
prese
rvat
ion
to
giv
e
m
or
e
eff
ic
ie
nt
DA.
The
desi
gn
of
energy
pres
erv
at
io
n
r
outi
ng
a
doptin
g
the
cl
us
t
eri
ng
protoc
ol
li
ke
low
-
ene
rg
y
-
ad
aptive
-
cl
us
te
rin
g
-
hier
arch
y
(L
EAC
H
)
a
nd
hybri
d
-
energy
-
e
ff
ic
ie
nt
-
distrib
uted
(
HEE
D
)
has
be
en
intr
oduce
d
in
[15].
An
y
ways,
t
hey
induce
the
e
ne
rg
y o
verhea
d
be
tween
CH
t
ha
t
is
ins
uffici
ent
f
or
the
la
rg
e
r
netw
ork
[16], d
ue
t
o
direct
-
DT
(
Data
-
Tra
ns
m
issi
on
)
via
CH
to
w
ard
s
the
sin
k
is
an
unfeasibl
e
m
et
ho
d
for
s
uch
ty
pe
of
ne
twork
.
I
n
orde
r
t
o
im
pro
ve
the
li
feti
m
e
of
WSNs
f
or
over
com
ing
the
e
ne
rg
y
ov
erh
ea
d
betwee
n
C
H
[17]
des
ign
e
d
a
r
ou
ti
ng
m
eth
od
f
or
t
he
se
le
ct
ion
of
hop
-
de
vice.
H
ow
e
ver,
it
acq
uire
s
hi
gh
e
r
c
omm
un
ic
at
ion
ov
erhe
a
d
b
e
c
a
u
s
e
o
f
c
h
a
n
n
e
l
c
o
n
t
e
n
t
i
o
n
b
e
t
w
e
e
n
c
l
u
s
t
e
r
d
e
v
i
c
e
a
n
d
h
o
p
-
d
e
v
i
c
e
a
n
d
i
t
i
s
i
m
p
r
o
v
i
n
g
t
h
e
N
P
-
d
e
t
e
r
m
i
n
i
s
t
i
c
.
In
[
18
]
,
it
re
presenti
ng
the
desig
n
of
e
ne
rg
y
-
e
ff
ic
ie
nt
f
or
la
r
ger
S
N
adoptin
g
a
f
uz
zy
-
base
d
cl
us
te
rin
g
m
eth
od.
H
ow
e
ve
r,
the
pe
rfo
rm
ance
of
a
li
fetim
e
in
ins
uffici
ent
,
CH
de
vices
ne
arer
t
o
BS
tha
t
dies
rap
i
dly. To
o
ve
rco
m
e these p
roblem
s,
[1
9] r
epr
ese
nted
cl
us
te
ring
d
e
sig
n
ut
il
iz
ing
T2
FL (Ty
pe
-
2
fu
zzy
log
ic
).
The
distrib
ute
d
loa
d
m
od
el
betwee
n
S
D
s
that
ai
de
d
in
de
velo
ping
the
li
fetim
e
of
SN.
Howev
e
r,
the
T
2F
L
-
cl
us
t
erin
g
m
et
ho
d
can
be
desig
ne
d
a
nd
co
ns
i
de
rin
g
the
hom
og
e
nous
ne
tw
ork.
The
refo
re
,
f
uture
routin
g
m
od
el
sh
ould
ass
ume
the
heteroge
neity
need
s
of
WSN
a
nd
it
s
app
li
cat
ion
[
20
-
23
]
.
I
n
ord
er
to
com
m
un
ic
at
e
and
accum
ulate
these
data
in
the
real
-
ti
m
e
of
ef
fici
ent
de
sig
ns
a
re
ne
eded.
In
[
24
,
25
]
,
repre
se
nting
the
m
od
el
of
da
ta
gathe
rin
g
and
the
e
ff
ic
ie
nt
m
od
el
of
da
ta
routin
g
a
doptin
g
t
he
cl
ust
ering
m
et
ho
d,
[26]
represe
nted
da
ta
fo
recasti
ng
m
et
ho
d
f
o
r
cl
us
te
r
-
base
d
WSNs
an
d
[27
,
28
]
re
presented
the
appr
oach
of
cr
os
s
-
la
ye
r
fo
r
the
cl
ust
er
-
base
d
W
S
Ns.
I
n
[
24
-
28]
,
this
m
od
el
reduced
the
energy
consum
ption
of
t
he
S
D.
H
ow
e
ve
r,
the
s
ta
te
-
of
-
the
-
art
m
od
el
didn’t
assum
e
and
fail
ed
to
im
pro
vise
the
c
ov
e
rage
t
i
m
e
of
WSN.
To
a
ddres
s
the
se
issues
of
c
ov
e
ra
ge
ti
m
e,
the
op
ti
m
iz
a
tio
n
f
unct
ion
a
doptin
g
the
evo
l
utio
na
r
y
co
m
pu
ti
ng
f
or
cl
us
te
r
f
or
m
at
ion
,
w
hic
h
is
rep
re
sente
d
in
[29].
Howe
ve
r,
in
[
30
]
exte
ns
ive
analy
sis
carri
ed
out
an
d
r
epr
ese
nts
the
evo
l
ution
a
ry
com
pu
ti
ng
f
or
the
heter
ogene
ou
s
WSN
inc
ur
s
Com
pu
ta
ti
on
al
-
O
verhea
d
(C
O
)
be
twee
n
S
Ns.
T
hu
s
,
it
aff
ect
in
g
the
pe
r
f
or
m
ance
of
WSN.
I
n
pa
pe
r
[31],
represe
nting
the
cl
us
te
rin
g
m
et
ho
d
f
or
he
te
rogen
e
ous
-
WSN
util
iz
ing
the
t
ree
str
uc
ture.
T
his
m
od
el
i
s
assum
ed
packet
loss
rate
and
li
nk
qual
it
y
in
or
de
r
to
r
edu
ce
the
e
ne
rg
y
co
nsum
pti
on
of
S
N.
H
oweve
r,
this m
od
el
d
id
n’
t as
su
m
e the
cov
e
ra
ge
ti
m
e.
As
a
n ou
tc
om
e
, affect
ing t
he
li
fetim
e p
erform
ance of SNs.
To
ov
e
rc
om
e
t
he
re
searc
h
c
ha
ll
eng
e,
t
his
w
ork
represe
nts
the
m
od
el
of
e
nergy
ef
fici
ent
cl
us
te
rin
g
and
r
outi
ng
op
tim
iz
at
ion
(
EE
CR
O
)
ad
op
ti
ng
the
desi
gn
of
a
cro
ss
-
la
ye
r
fo
r
im
pr
ov
i
ng
the
li
fetim
e
of
W
S
N
and
c
overa
ge
tim
e.
The
EECR
O
util
iz
es
ph
ysi
cal
la
ye
r
data
to
get
t
he
cha
nnel
ga
in
data
f
or
ga
inin
g
the
par
am
et
er
of
li
nk
qual
it
y.
Fu
rt
her
m
or
e,
i
n
the
MAC
la
ye
r
or
data
li
nk,
the
TDMA
ba
sed
com
m
un
ic
at
ion
can
be
ada
pte
d
for
com
m
un
i
cat
ion
betwee
n
inter
an
d
in
tracl
us
te
r
com
m
un
ic
at
ion
.
In
the
networ
k
la
ye
r,
the
r
ou
ti
ng
a
nd
cl
us
te
ri
ng
-
ba
sed
tra
ns
m
iss
ion
a
re
ass
ume
d.
T
his
w
ork
intende
d
at
c
onveyi
ng
the
good
trade
-
off
am
ong
im
pr
ov
in
g
c
ov
e
ra
ge
tim
e,
reducin
g
the
e
nergy
of
CH
a
nd
im
pr
ovin
g
the
li
fetim
e
of
SN
.
In
o
r
d
e
r
t
o
g
e
t
,
r
o
u
t
i
n
g
a
n
d
c
l
u
s
t
e
r
i
n
g
o
p
t
i
m
i
z
a
t
i
o
n
a
r
e
c
a
r
r
i
e
d
o
u
t
a
n
d
t
h
e
s
m
a
l
l
e
s
t
p
a
t
h
b
a
s
e
d
r
i
o
t
i
n
g
i
s
a
s
s
u
m
e
d
.
The
C
on
t
rib
ution o
f
t
his r
esea
rch w
ork
is
as
fo
ll
ows:
This
pap
e
r rep
r
esented
the
des
ign
of e
nergy
-
e
ff
ic
ie
nt
routin
g f
or
t
he hete
r
ogeneous
WSN.
Our
m
od
el
a
dopts the
sho
rtest
p
at
h t
o
cl
us
te
r
-
base
d
t
ran
sm
i
ssion.
The
previ
ous
work
has
not
assum
ed
cl
us
te
r
-
base
d
r
ou
ti
ng
op
ti
m
iz
ation
adoptin
g
the
de
sign
of
the
cr
oss
-
la
ye
r
an
d
c
on
si
der
i
ng the e
nviro
nm
ent o
f het
eroge
neous sen
so
rs
.
The
previ
ou
s
w
ork
h
as not
c
on
si
der
e
d
the
e
valuati
on o
f
li
f
et
i
m
e
per
f
or
m
ance
c
onside
rin
g
first n
ode
de
at
h,
loss
of
co
nnect
ivit
y and
t
otal
node deat
h co
nsi
der
in
g
t
he he
te
rogen
e
ous
W
SN
s.
The
perform
ance
of
li
fetim
e
analy
sis
is
carri
ed
ou
t
c
onside
r
ing
t
he
fi
r
st
se
ns
or d
e
vice,
lo
ss
of
c
onnecti
vi
t
y
and total
devic
e d
eat
h.
This
m
od
el
de
velo
ps
the
li
fetim
e
of
S
N,
c
om
m
un
ic
at
ion
over
hea
d
an
d
r
outi
ng
over
hea
d
of
the
real
-
ti
m
e
data acce
ss
an
d
al
s
o
im
pr
ov
e
s the lifet
im
e o
f WSN.
This
pap
e
r
is
orga
nized
i
n
s
uch
a
way
th
a
t
sect
ion
-
I
I
int
rod
uced
the
e
ne
rg
y
-
e
ff
ic
ie
nt
of
r
ou
ti
ng
and
the
cl
us
te
r
ing
opti
m
iz
at
i
on
m
od
el
is
re
pr
ese
nted
.
T
he
ne
xt
sect
io
n
r
epr
ese
nts
a
n
e
xp
e
rim
ental
st
ud
y
of
the
EECR
O
over
the
e
xisti
ng
te
chn
iq
ue.
T
he
f
uture
work
a
nd
c
on
cl
us
io
n
are
discusse
d
in
t
he
l
ast
sect
i
on
of
this pa
per.
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.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
47
98
-
48
08
4800
2.
ENERG
Y
-
E
F
FICIE
NT
CLUS
TE
RI
N
G
A
ND
ROUTIN
G
OPTIMIZ
AT
ION
MO
D
EL
ADOPTIN
G
CROSS
L
AY
E
R
DESI
GN
F
OR
ENH
ANCI
NG
COVER
AGE
TIME
A
N
D
MA
X
I
MIZ
IN
G LIFET
IME
OF WS
N
This
sect
ion
pr
esent
an
ene
rgy
eff
ic
ie
nt
cl
us
te
ring
an
d
r
outi
ng
opti
m
iz
at
io
n
m
od
el
ado
pti
ng
cr
oss
la
ye
r
desig
n
f
or
m
ini
m
izing
energy
co
nsu
m
pt
ion
of
cl
ust
er
hea
d,
e
nh
a
ncin
g
co
ve
rage
tim
e
and
m
a
xim
iz
e
li
fetim
e
of
sen
so
r
net
wor
k.
F
irstl
y,
we
desc
ribe
t
he
syst
em
m
od
el
of
hete
roge
neous
wir
el
ess
sens
or
ne
twork
,
then
we
de
scr
ibe
the
c
hann
el
m
od
el
us
e
d.
T
he
n,
we
descr
i
be
the
c
lusterin
g
a
nd
transm
issi
on
/r
ou
ti
ng
op
ti
m
iz
ation
to
enhance
cove
r
age tim
e and
th
e li
fetim
e o
f W
SN
.
Fig
ure
1
.
The
Ar
c
hitec
ture
of
EECR
O
m
od
e
l
2.1.
S
ys
te
m
ar
chitecture
The
arc
hitec
ture
of
EECR
O
m
od
el
is
sh
ow
n
in
Fig
ure
1.
From
Fig
ure
1
it
can
be
seen
tha
t
the
cl
us
te
r
cl
ose
r
to
base
sta
ti
on
is
com
po
se
d
of
le
ss
nu
m
ber
of
cl
ust
er
m
e
m
ber
s
we
cal
l
this
as
le
vel
1
an
d
cl
us
te
r
li
tt
le
far
away
from
cl
us
te
r
head
has
m
or
e
nu
m
ber
of
cl
us
te
r
m
e
m
ber
s
we
cal
l
this
has
l
evel
2.
This
way
the
far
cl
ust
er
w
il
l
hav
e
la
rg
e
densi
ty
of
cl
us
te
r
m
e
m
ber
s.
This
dep
l
oym
ent
m
et
ho
d
ai
d
i
n
m
ini
m
iz
ing
e
ne
rg
y
c
onsu
m
ption
of
cl
us
te
r
head.
Es
pecial
ly
,
the
cl
ust
er
he
ad
cl
ose
r
t
o
s
ink
.
T
hu
s
en
ha
ncin
g
cov
e
ra
ge
ti
m
e
and li
fetim
e o
f sens
or
net
work.
2.2.
Sy
ste
m, c
ha
n
nel,
trans
mi
ssion op
tim
iz
at
ion m
od
el
:
This
sect
io
n
de
scribes
the
syst
e
m
of
resear
ch
w
ork.
This
w
ork
co
ns
ide
rs
he
te
rogen
e
ous
W
SN
i.e
.,
le
t’s
co
ns
ide
r
cl
asses
of
se
nsor
de
vice
s
uch
cl
ass
A
,
cl
ass
B.
Cl
ass
A
ar
e
re
pr
ese
nted
as
sens
or
dev
i
ce
that
per
form
s
op
era
ti
on
su
c
h
as
sensing.
T
hese
de
vice
are
lowe
r
cost
and
the
ti
ny
dev
ic
es
that
are
dep
loye
d
acro
s
s
sensing
re
gion.
The
sens
or
ar
e
gr
ou
ped
to
ge
ther
to
f
or
m
a
cl
us
te
rs.
Cl
ass
B
sensor
dev
ic
e
is
m
or
e
po
w
erful
and
has
hi
gh
e
r
com
pu
ti
ng
ca
pab
il
it
y
than
C
la
ss
A
dev
ic
e
wh
ic
h
de
picte
d
as
cl
us
te
r
hea
d.
T
he
cl
ass
B
dev
ic
e
colle
ct
s and a
ggre
gates se
nsor
y data f
ro
m
it
s m
e
m
ber
and tr
ansm
it
s i
t t
ow
a
rd
s
sin
k/b
ase
stat
ion
th
r
ough s
et
o
f
hop/interm
ediat
e cluste
r hea
d dev
ic
e.
Let
us
c
on
si
de
r
t
her
e
are
and
no
des
t
ha
t
are
ra
ndoml
y
dep
l
oyed
in
a
netw
ork
an
d
t
hei
r
po
sit
io
ns
are
know
n.
Eac
h
se
ns
or
dev
ic
e
are
connecte
d/ass
ociat
ed
with
one
cl
us
te
r
hea
d
dev
ic
e
a
nd
ge
ner
at
es
m
ean
pac
ket
l
oad
of
bits/
sec
an
d
t
ran
sm
its
it
to
t
he
cl
ust
er
head,
wh
i
ch
furthe
r
rout
es
to
the
sin
k/b
ase
sta
ti
on
s
(
w
hich
in
t
his
w
ork
we
c
on
si
der
it
as
the
(
+
1
)
ℎ
cl
us
te
r
head
directl
y
or
th
rou
gh
i
nter
m
ediat
ed
cl
us
te
r
hea
d
de
vices.
Furthe
r,
this
w
ork
c
ons
iders
t
hat
the
c
luster
head
c
onsu
m
es
m
uch
hi
gh
e
r
e
nergy
t
ha
n
it
s
sens
or
de
vices
.
Since,
CH
is
act
ive
al
l
the
t
i
m
e
and
at
the
sa
m
e
t
i
m
e
the
m
e
m
ber
dev
ic
e
are
in
sle
ep
sta
te
.
As
a
res
ult,
th
is
wo
r
k
ai
m
s
t
o
re
du
ce
th
e
con
s
um
ption
of
energy
of
C
H
de
vice.
As
it
ai
ds
in
enh
a
ncin
g
netw
ork
c
over
age
resu
lt
in
g
i
n bett
er lifet
im
e of
WSNs.
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
Ener
gy
ef
fi
ci
en
t cl
us
te
ring a
nd r
ou
ti
ng opti
miza
ti
on
m
od
el
for
…
(
Savi
th
a S
.
)
4801
This
wor
k
c
onsi
der
s
Ra
yl
ei
gh
fad
i
ng
m
od
el
to
c
har
act
eriz
e
the
c
ha
nn
el
a
m
on
g
cl
us
te
r
he
ads,
a
nd
al
so
am
on
g
cl
us
te
r
hea
d
an
d
the
ba
se
sta
ti
on.
T
her
e
fore
the
cha
nnel
ga
in
am
on
g
se
nd
er
an
d
receive
r
f
or
com
m
un
ic
at
ion
is
obta
ti
ned
a
s foll
ow
s
:
(
)
=
(
0
)
(
0
)
−
(1)
w
he
re
(
0
)
is t
he p
at
h
loss
com
ponen
t e
xpone
nt
of
0
wh
ic
h
ca
n be c
om
pu
te
d
a
s foll
ow
s:
(
0
)
=
2
16
2
0
2
(2)
w
he
re
is
the
anten
na
gain
of
the
se
nd
e
r,
i
s
the
a
nten
na
gain
of
receive
r,
is
a
norm
alized
ar
bitrary
par
am
et
er
that
dep
ic
ts
the
va
riat
ion
in
t
he
fad
in
g
proces
s,
is
def
i
ned
as
wav
el
e
ngth
of
the
fr
e
que
ncy
carrier
,
is
de
f
ined
as
the
ex
pone
nt
of
path
loss
.
The
is
arb
it
ra
ry
a
nd
i
s
c
on
si
der
e
d
to
be
ex
po
nen
ti
al
ly
distrib
uted,
a
nd
the
recei
ve
d
sig
nal
is
al
s
o
ar
bitrary.
Th
eref
or
e
,
perfec
t
receptio
n
of
a
sig
nal
is
a
ssu
re
d
thr
ough
pro
ba
bili
sti
c
m
e
thod.
H
ence
it
is
de
sire
d
t
hat
{
≥
}
≥
f
or
i
deal
rec
eption,
w
her
e
is
t
h
e
e
n
e
r
g
y
o
f
o
b
t
a
i
n
e
d
s
i
g
n
a
l
,
i
s
p
r
e
d
e
t
e
r
m
i
n
e
d
e
n
e
r
g
y
t
h
r
e
s
h
o
l
d
,
a
n
d
i
s
t
h
e
e
x
p
e
c
t
e
d
l
i
n
k
i
d
e
a
l
p
a
r
a
m
e
t
e
r
.
Let
co
ns
ide
r
as
t
he
cum
ulate
d
intra
cl
us
te
r
loa
d
at
t
ai
ned
by
t
he
ℎ
CH
(
bit/
seconds
)
for
=
1
,
…
,
. Th
e
cluster
ing
optim
iz
at
io
n vecto
r
is e
xpresse
d
as
f
ollo
ws
:
=
(
1
,
…
,
)
.
(3)
An
im
po
rtant
t
hing
t
o
be
see
n
he
re
is
that,
the
nu
m
ber
of
sens
or
dev
ic
es
ass
ociat
ed
wit
h
cl
us
te
r
head
i.e.,
the size
o
f
cl
ust
er
, is e
xpresse
d
as
foll
ows
:
.
(4)
Fo
r
∈
{
1
,
2
,
3
,
…
,
}
and
∈
{
1
,
2
,
3
,
…
,
+
1
}
,
with
≠
,
le
t
be
the
inter
cl
us
te
r
loa
d
t
hat
is
transm
itted
f
rom
CH
to
the
CH
.
T
he
tra
nsm
issi
on
op
ti
m
iz
at
ion
m
a
trix
is
the
∗
(
+
1
)
m
at
ri
x
of
el
e
m
ent
,
=
1
,
…
,
and
=
1
,
…
,
+
1
.
This
w
ork
c
onsiders
=
0
.
The
obje
ct
ive
of
this
work
is
t
o
i
m
pr
ovise
t
he
co
ve
rag
e
ti
m
e
by
est
ablis
hi
ng
an
op
ti
m
i
zed
t
ran
sm
issi
on
m
at
rix
′
an
d
cl
us
te
r
vector
′
.
Let
co
ns
ide
r
as
the
m
ean
of
e
nergy
c
on
s
um
ption
of
th
e
ℎ
cl
us
te
r
hea
d.
The
n,
the
is
ex
pr
es
sed
as
fo
ll
ows
:
=
(
+
∑
1
≤
≤
0
,
≠
)
+
(
∑
1
≤
≤
0
+
1
,
≠
)
+
∑
1
≤
≤
+
1
,
≠
,
=
1
,
…
,
(5)
wh
e
re
are
the
ci
rcu
it
ene
rg
y
per
bit
dissi
pa
te
d
in
tra
ns
m
itti
ng
data,
are
the
ci
rcu
it
e
ne
rg
y
per
bit
dissipated
in
r
ecei
vin
g
data,
and
is
the
ene
rg
y
dissipate
d
from
cl
us
te
r
he
ad
to
cl
us
te
r
hea
d
.
Le
t
us
assum
e
that
as
the
dista
nce
a
m
on
g
cl
ust
er head
a
nd
,
t
her
e
fore u
sin
g
(
1)
t
he
receive
d
en
erg
y pe
r
bit
c
a
n
be
e
xpresse
d
a
s foll
ow
s
:
=
(
0
)
(
0
)
−
.
(6)
By
u
sin
g
Ra
yl
ei
gh
c
ha
nn
el
m
od
el
,
the li
nk ideal pa
ram
et
er can
be desc
rib
ed
as
foll
ows
:
=
{
≥
}
=
{
≥
(
0
)
(
0
)
}
=
−
(
0
)
0
(7)
Fr
om
(
7),
w
e
c
an desc
ri
be
as
fo
ll
ows
:
=
,
≠
(8)
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.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
47
98
-
48
08
4
802
w
he
re
is a c
onsta
nt that ca
n b
e ex
pr
es
sed
as
def
i
niti
on
a
s fo
ll
ow
s:
=
−
(
0
)
0
log
(9)
The
(5)
ca
n be
wr
it
te
n
c
onside
rin
g
f
or
=
1
,
…
,
, as
f
ollow
s
:
=
(
+
∑
1
≤
≤
,
≠
)
+
∑
(
+
)
1
≤
≤
+
1
,
≠
(10)
Let
dep
ic
ts
th
e
init
ia
l
ener
gy
of
the
ℎ
cl
us
te
r
hea
d,
=
1
,
…
,
.
Thi
s
wor
k
co
ns
id
ers
an
optim
izati
on
pro
blem
to
m
a
xim
iz
e cov
e
rage t
i
m
e as f
ollo
ws
:
{
,
}
{
1
1
,
2
2
,
…
,
}
.
(11)
Wh
e
n
cl
us
te
r h
eads a
re
dep
l
oy
ed
with
equal
energy, t
hat is,
=
∀
,
(12)
T
he o
pti
m
iz
ati
on pr
ob
le
m
o
f
(11) is sim
i
la
r
to as
foll
ows
:
{
,
}
{
1
,
…
,
}
(13)
This
w
ork
ai
m
s
a
t
add
res
s
ing
opti
m
iz
at
io
n
pr
ob
le
m
of
(13).
T
he
pr
oblem
in
op
ti
m
i
zi
ng
a
re
descri
bed
a
s
fo
ll
ows.
F
or cluster
hea
d
,
=
1
,
…
,
, th
e
inter cl
us
te
r o
pti
m
iz
at
ion
cond
it
io
n
m
us
t b
e ad
dr
es
sed
+
∑
1
≤
≤
,
≠
=
∑
+
,
+
1
1
≤
≤
,
≠
(14)
w
he
re
0
<
≤
1
is
the
perform
ance
pa
ram
et
er
of
a
ggre
gatin
g
data
f
un
ct
io
n
in
i
nt
ra
cl
us
te
r
.
Al
ong
with
,
the
pac
ket
load
com
po
sed
by
al
l
the
cl
us
t
er
head
c
onside
ring
ce
rtai
n
insta
nce
pe
rio
d
of
ti
m
e
m
us
t
be
identic
al
t
o
l
o
a
d
p
r
o
d
u
c
e
d
b
y
a
l
l
t
h
e
s
e
n
s
o
r
d
e
v
i
c
e
s
i
n
t
h
e
s
a
m
e
i
n
s
t
a
n
c
e
p
e
r
i
o
d
o
f
t
i
m
e
,
w
h
i
c
h
c
a
n
b
e
e
x
p
r
e
s
s
e
d
a
s
f
o
l
l
o
w
s
:
∑
=
1
=
.
(15)
The
ob
j
ect
ive
op
ti
m
iz
at
ion
funct
ion
(1
3)
and
c
on
st
raint
(1
4)
an
d
(
15)
can
be
tra
nsfo
rm
ed
into
li
near
pro
gr
am
m
ing
prob
le
m
of
,
and
by
introd
ucin
g
a
n
sup
plem
ent
ary
par
am
et
er
,
w
her
e
≥
ma
x
{
1
,
…
,
}
as foll
ows
:
{
{
,
,
}
ℎ
ℎ
∑
+
1
≤
≤
,
≠
−
∑
+
,
+
1
1
≤
≤
,
≠
=
0
,
=
1
,
…
∑
=
1
=
∑
1
≤
≤
,
≠
+
+
∑
(
+
)
1
≤
≤
,
≠
+
+
1
(
+
,
+
1
)
−
≤
0
,
=
1
,
…
,
≥
0
≥
0
,
=
1
,
…
,
;
=
1
,
…
,
+
1
(16)
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
Ener
gy
ef
fi
ci
en
t cl
us
te
ring a
nd r
ou
ti
ng opti
miza
ti
on
m
od
el
for
…
(
Savi
th
a S
.
)
4803
2.3.
Clus
ter
optimi
z
at
ion
This
sect
ion
pr
ese
nts
cl
us
te
r
opti
m
izati
on
te
chn
i
que
for
wire
le
ss
sensor
ne
t
work.
Let
′
=
(
1
′
,
…
,
′
)
be
t
he
optim
al
cl
us
te
ring
vecto
r
outc
om
e.
For
=
1
,
…
,
,
cl
ust
er
head
is
give
n
′
=
′
⁄
sens
or
de
vices.
The
sens
or
dev
ic
e
al
locat
i
on
is
ca
rr
ie
d
ou
t
in
seq
ue
ntial
m
ann
er,
i.e
.,
one
at
a
tim
e.
A
co
rr
esp
onding
se
nsor
de
vice
is
al
locat
ed
to
t
he
near
e
st
cl
us
t
er
hea
d
,
pr
ov
i
ded
that
nu
m
ber
of
sens
or
de
vices
to
cl
us
te
r
he
ad
is
not
great
er
tha
n
′
.
I
f
it
exceeds
the
n
nex
t
nea
rest
cl
us
te
r
hea
d
is
consi
der
e
d
a
nd
so
on. T
he
alg
or
it
hm
f
or
ob
ta
ining o
ptim
al
c
lusterin
g
is
represente
d
in
alg
or
it
hm
1
.
Algorithm 1:
Optimal clustering algorithm
Input:
;
=
′
=
(
1
′
,
…
,
′
)
Expected outcome:
1
,
…
,
Initialize:
1
=
⋯
=
=
∅
(cluster sets)
Start: For
=
1
to
For
=
1
to
Set
to distance among sensor device
and cluster head
End for
Iteration:
=
{
}
min
{
,
=
1
,
…
,
}
If
′
>
0
′
=
′
−
=
+
{
}
Else
=
∞
go to iteration
End if
End for
End:
2.4.
Tr
ansmi
s
s
io
n/R
outing
op
timi
z
at
i
on
This
sect
io
n
descr
i
bes
t
he
routin
g
op
ti
m
i
zat
ion
of
pro
po
s
e
d
ap
proac
h.
This
w
ork
co
ns
ide
rs
a
routing
base
on
s
hortest
pat
h
r
oo
t
f
r
om
c
l
us
te
r
head
devi
ce
to
base
sta
ti
on
th
rou
gh
nu
m
ber
of
hop
de
vices.
Fo
r
m
ini
m
iz
ing
hop
co
unt
w
hich
va
r
ie
s
for
diff
e
re
nt
trans
m
issi
on
.
As
a
r
esult,
this
wor
k
co
ns
ide
rs
qu
al
it
y
of
com
m
un
ic
at
ion
us
i
ng
pa
ram
et
er
for
com
pu
ti
ng
pro
ba
bili
ty
of
posit
ive
e
nd
-
to
-
en
d
r
ece
ption.
F
or
different
roots
of
paths
ex
per
ie
nce
di
ff
e
ren
t
fa
di
ng,
t
he
r
oo
t
r
el
ia
bili
ty
m
us
t
be
at
le
ast
1
.Co
ns
id
eri
ng
the
shortest
hop
case,
t
he
pac
kets
are
routed
thr
ough
nea
re
st
cl
us
te
r
hea
d
cl
os
er
to
t
he
nex
t
le
vel
toward
s
the
ba
se
sta
ti
on.
I
n
this
way
the
da
ta
is
tra
nsm
it
te
d
to
dif
f
eren
t
le
vel
(
=
1
)
ti
ll
it
reach
es
the
base
sta
ti
on.
This
w
ork
co
nsi
der
s
e
nergy
balance
d
cl
us
t
er
base
d
r
ou
ti
ng
desig
n
that
balance
ene
rg
y
of
dif
fer
e
nt
cl
us
te
r
heads. T
he
c
om
m
un
ic
at
ion
ra
diu
s
of clu
ste
r
can be
obta
ine
d
as
foll
ows
1
2
(
1
−
0
)
,
…
,
1
2
(
−
−
1
)
,
(17)
In
(
17)
is
t
he
i
m
po
rtant
fact
or
of
e
nergy
dissipati
on
at
diff
e
re
nt
cl
us
t
er
head
s
.
F
or
exam
ple,
m
inim
iz
ing
1
2
(
−
−
1
)
res
ults
in
sm
al
le
r
cl
us
te
r
siz
e
in
the
ℎ
le
vel,
w
hich
ai
ds
in
reducin
g
local
traff
ic
am
on
g
these
cl
us
te
r,
le
ss
er
r
ou
ti
ng
distance
am
on
g
c
orres
pondin
g
cl
ust
er
head
s
in
the
(
−
1
)
le
vel
an
d
a
gre
at
er
num
ber
s
of
cl
us
te
r
hea
ds
i
n
the
ℎ
le
ve
l.
Sinc
e
t
his
w
ork
co
ns
id
ers
sy
m
m
e
tric
al
top
ol
og
y
a
nd
pa
cket
distrib
ution,
the
load
from
the
cl
us
te
r
hea
d
in
the
ℎ
le
vel
will
be
unif
orm
ly
balanced
a
m
on
g
highe
r
nu
m
ber
of
cl
ust
er
heads
i
n
t
he
ℎ
l
evel,
so
the
quantum
of
tra
nsm
itted
loa
d
po
ssess
by
eac
h
cl
us
te
r
hea
ds
in
t
he
ℎ
le
vel
will
reduce.
This
ai
d
in
re
du
ci
ng
e
nergy
co
nsum
ption
at
the
cl
ust
er
hea
d
i
n
th
e
ℎ
rin
g.
Sim
i
lar
ly
,
re
du
ci
ng
a
rea
of
the
ℎ
le
vel
m
us
t
reim
bu
rse
d
f
or
oth
e
r
cl
us
t
ers
i.e.
,
cl
us
te
r
in
the
ℎ
le
vel,
be
cause
of
the
fi
xed
num
ber
of
le
vel
con
si
der
e
d
in
the
netw
ork
.
In
a
sim
il
ar
way,
energy
di
ssipati
on
at
cl
us
te
r
hea
ds
in
le
vel
will
increase.
Th
us
,
by
re
gu
l
at
ing
t
he
siz
e
of
cl
us
te
r
in
di
ff
ere
nt
le
vels,
a
m
or
e
balan
ced
energy
dis
sip
at
ion
at
different
cl
us
te
r
hea
ds
i
s
at
ta
ined,
w
hi
ch
ai
de
d
in
en
han
ci
ng
c
over
age
tim
e
of
WSN
.
T
hus
im
pr
ov
i
ng
t
he
li
feti
m
e
of
WSNs
wh
ic
h
i
s experim
ental
l
y pro
ven in
ne
xt secti
on
belo
w.
3.
SIMULATI
O
N RESULT
A
ND AN
NA
L
Y
SIS
This
sect
io
n
r
epr
ese
nts
t
he
perform
ance
evaluati
on
of
i
ntr
oduce
d
the
EECR
O
m
od
el
ove
r
the
existi
ng
te
chn
i
qu
e
co
ns
i
der
i
ng
li
feti
m
e
,
r
ou
ti
ng
over
head
an
d
c
omm
un
ic
at
ion
ov
erh
ea
d.
For
li
f
et
i
m
e
analy
sis,
this
work
co
ns
ide
rs
first
sen
sor
de
vice
deat
h
(
F
SDD
),
l
os
s
of
connecti
vity
(
LoC
)
a
nd
total
sen
s
or
-
dev
ic
e
death
(
TSDD
)
.
T
he
previ
ou
s
w
ork
ha
s
not
co
ns
ide
r
ed
s
uch
a
n
e
va
luati
on
t
o
est
im
at
e
the
per
f
orm
ance
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.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
47
98
-
48
08
4804
of
WSN.
The
env
i
ronm
ent
of
the
syst
e
m
is
util
iz
ed
fo
r
ex
per
im
ent
analysis
su
ch
as
wi
ndows
10
ente
rprises
editi
on
oper
at
ing
-
syst
em
(
OS
),
I
ntel
Pe
nt
ium
I
-
5
cl
ass,
a
64
-
bit
proce
sso
r
of
Q
ua
d
-
cor
e
,
4G
B
N
VIDIA
-
CUD
A
ena
ble
d
de
dicat
ed
G
PU
,
16GB
RAM.
The
S
E
NSORI
A
si
m
ulator
[
32]
is
util
iz
ed
to
carry
out
the
pe
rfor
m
ance
evaluati
on
of
the
EECR
O
m
od
el
over
e
xisti
ng
m
et
ho
ds
li
ke
LE
ACH
[
11]
.
The
EECR
O
an
d
LEAC
H
are
m
od
el
e
d
by
util
i
zi
ng
t
he
D
ot
Net
f
ram
ewo
r
k
4.5
a
nd
C#
pro
gr
am
m
ing
l
angua
ge.
The
LEAC
H
has
been
wide
ly
util
iz
ed
the
com
par
iso
n
prot
oco
l
ac
ro
s
s
diff
e
re
nt
existi
ng
m
et
hods
[
11]
.
As
an
outc
om
e,
this
w
ork
ass
um
es
the
LEAC
H
pr
oto
c
ol
as
a
case
stu
dy
f
or
com
p
arison.
T
his
sim
ulati
on
par
am
et
er
is
util
iz
ed
for
the
expe
rim
ental
an
al
ysi
s,
w
hic
h
is
des
cribe
d
in
Ta
ble
1
.
Table
1.
Sim
ul
at
ion
par
am
et
e
r
c
on
si
der
e
d
Netwo
rk Pa
ra
m
e
te
r
Valu
e
The size of
Wirele
ss
Network
1
0
0
m
×
1
0
0
m
Nu
m
b
e
r
o
f
the Sen
so
r
Dev
ices
5
0
0
,
1
0
0
0
,
1
5
0
0
&
2
0
0
0
Nu
m
b
e
r
o
f
BS
1
The Initial
en
ergy
o
f
Sens
o
r
Dev
ices
0
.1to
0.2
Jo
u
les (j)
TR (
Tr
an
s
m
iss
io
n
rang
e)
5
m
The ra
n
g
e of
Sens
in
g
3
m
Rad
io
-
Energy
-
Dis
sip
atio
n
5
0
nj/b
it
The len
g
th
of
Data
Pack
ets
5
0
0
0
bits
The sp
eed o
f
T
ran
s
m
iss
io
n
1
0
0
bit/s
Ban
d
wid
th
1
0
0
0
0
bit/s
Data proces
sin
g
-
d
elay
0
.1 s
Idle Energ
y
-
Co
n
su
m
p
tio
n
(E
ele
c)
5
0
nj/b
it
A
m
p
lif
icatio
n
-
Ene
rgy
(
E
m
p
)
1
00
/bit
/
m2
3.1.
The
e
valuat
i
on
of
Li
f
etime
perf
or
man
ce
e
valua
tion
c
on
sider
i
ng
tota
l
se
ns
or
de
vice
de
ath,
first
senso
r
de
vice
deat
h
an
d
loss
of c
on
nec
tivit
y
This
sect
ion
descr
i
bes
pe
rfor
m
ance
at
ta
i
ned
by
pro
posed
EECR
O
ov
e
r
LE
ACH
con
si
deri
ng
the
total
sen
sor
de
vice
death
,
LoC
,
fi
rst
se
ns
or
de
vice
de
at
h.
Fir
stl
y,
co
ns
ide
rin
g
the
case
of
total
s
ens
or
dev
ic
e
death
.
Her
e
t
he
se
nsor
de
vi
ce
is
va
ried
f
r
om
50
0,
1000,
1500
,
an
d
20
00
an
d
the
e
xperim
ent
is
cond
ucted
to
e
valuate
the
li
fe
tim
e
per
fo
rm
ance
an
d
the
res
ult
is
gr
ap
hical
ly
sh
own
in
Fi
g
ure
2.
The
outc
om
e
represe
nted
E
ECR
O
that
e
nhance
s
the
li
fe
tim
e
per
form
a
nce
by
69.
09
%
,
76.
22%,
82.96%
,
a
nd
83.
83
%
ove
r
LEAC
H
proto
col
consi
der
i
ng
500,
1000,
1500,
an
d
2000,
se
ns
or
dev
ic
e
res
pe
ct
ively
.
An
aver
a
ge
i
m
pr
ovem
ent
of
li
fetim
e
per
f
or
m
ance
is
at
t
ai
ned
by
78.
02
%
,
wh
ic
h
is
intr
oduce
d
EE
CR
O
ov
e
r
the
LEAC
H
consi
der
i
ng
t
he
total
sensor
dev
ic
e
death.
S
i
m
il
arly
,
an
exp
e
rim
ent
is
cond
ucted
to
evaluate
the
li
fetim
e
perform
ance
con
si
der
i
ng
1s
t
sens
or
dev
ic
e
death.
He
re,
t
he
se
ns
or
de
vi
ce
can
be
va
ried
f
ro
m
500,
1000,
1500,
a
nd
20
00
an
d
t
he
ex
pe
rim
ent
is
con
du
ct
e
d
to
e
sti
m
at
e
the
li
fetim
e
per
f
or
m
ance
an
d
the
outc
o
m
e
is
gr
a
phic
al
ly
sh
own
in
Fig
ur
e
3.
The
re
su
lt
r
epr
ese
nts
the
EECR
O
that
enh
a
nces
the
li
f
et
i
m
e
per
f
or
m
ance
by
82.44%
,
77.
67%,
88.
41%,
a
nd
92.
57
%
ov
er
the
LE
AC
H
prot
oco
l
tha
t
con
side
rin
g
500,
1000,
1500,
an
d
2000,
se
ns
or
de
vice
res
pecti
ve
ly
.
The
aver
a
ge
im
pr
ov
em
ent
of
li
feti
m
e
perform
ance
is
at
ta
ined
85.27
%
by
the
hel
p
of
i
nt
rodu
ce
d
E
ECR
O
ove
r
the
LEAC
H
that
consi
der
i
ng
1st
sens
or
de
vice
death
.
F
ur
t
he
rm
or
e,
the
e
xp
e
rim
ent
is
co
nducte
d
to
est
im
at
e
lif
et
i
m
e
per
f
or
m
ance
co
ns
ide
rin
g
L
oC.
He
r
e,
a
se
nsor
d
e
vice
is
var
ie
d
from
500,
1000,
1500,
and
2000
a
nd
t
he
e
xperim
ent
is
co
nducted
to
est
i
m
at
e
the
li
fetim
e
per
f
orm
ance
and
the
ou
tc
om
e
is
gr
ap
hica
ll
y
sh
own
in
F
ig
ure
4.
T
he
outc
om
e
rep
res
ents
EECR
O
dev
el
ops
the
li
fetim
e
perform
ance
by
81.46%
,
78.
52%,
85.
27%,
an
d
86.
74%
ov
e
r
t
he
L
EA
CH
protoc
ol
c
on
si
der
i
ng
500,
10
00,
1500,
an
d
2000,
sens
or
de
vice
resp
ect
ively
.
An
ave
ra
ge
li
fetim
e
per
fo
r
m
ance
i
m
pr
ove
m
ent
of
86.73%
i
s
at
ta
ined
by
in
tro
du
ce
d
EEC
RO
ov
e
r
L
E
ACH
co
ns
ide
r
ing
LoC.
T
he
overall
im
pr
ov
em
ent
of
a
ver
a
ge
perfo
rm
ance is att
ai
ned
b
y 8
3.3
5%,
wh
ic
h
is
introd
uced
by
the EECR
O
m
od
el
ov
e
r
LE
A
CH conside
rin
g
total
sens
or
de
vice
death,
LoC
,
fi
rst
sens
or
de
vi
ce
death
.
T
he
ov
e
rall
at
ta
ined
outc
om
e
represents
t
he
sca
la
ble
li
fetim
e p
erf
or
m
ance conside
rin
g
the
v
a
ried
netw
ork
de
ns
it
y.
3.2.
C
omm
u
nicati
on
ove
rhead
an
d
Routin
g/tr
an
s
mi
ssion
over
head
per
f
or
man
ce
e
va
lu
at
i
on
c
on
sider
in
g v
aried se
nso
r
d
evice
This
sect
io
n
de
fines
the
c
om
m
un
ic
at
ion
an
d
r
outi
ng
over
head
pe
rfor
m
ance
at
ta
ined
by
the
help
of
EECR
O
ov
e
r
LEAC
H.
F
or
the
e
xp
e
ri
m
ent
analy
sis,
the
sens
or
de
vice
is
var
ie
d
f
ro
m
500,
10
0,
1500,
a
nd
2000
a
nd
the
exp
e
rim
ent
is
cond
ucted
a
nd
the
ou
tc
om
e
is
gr
a
phic
al
ly
sh
ow
n
in
Fig
ure
5.
T
he
outc
om
e
sh
ows
,
EECR
O
m
ini
m
iz
es
t
he
CO
(Com
pu
ta
ti
on
ove
rh
e
ad)
by
32.
74%
,
26.25%,
48.644%,
an
d
41.
8
8%
ove
r
LEAC
H
co
ns
i
der
i
ng
500,
10
00,
15
00,
an
d
2000
the
se
nso
r
de
vice,
re
sp
e
ct
ively
.
An
a
ve
rag
e
c
omm
un
ic
at
ion
ov
e
r
head
r
ed
uc
ti
on
of 3
7.3
7%
is
at
ta
ined
by
EECR
O
over
LEAC
H.
Sim
i
la
rly
,
the
ex
perim
ent
is
con
du
ct
ed
t
o
est
i
m
at
e
the
pe
rfor
m
ance
of
r
ou
ti
ng
over
he
a
d
by
var
yi
ng
s
ens
or
de
vices
f
ro
m
500,
10
0,
1500,
a
nd
2000
a
nd
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
Ener
gy
ef
fi
ci
en
t cl
us
te
ring a
nd r
ou
ti
ng opti
miza
ti
on
m
od
el
for
…
(
Savi
th
a S
.
)
4805
the
outc
om
e
is
gr
a
phic
al
ly
rep
resen
te
d
in
Fi
g
ure
6.
T
he
ou
tc
om
e
sh
ow
s
,
E
ECR
O
m
ini
m
i
zes
routin
g
ove
rh
ea
d
by
51.
62
%
,
44.
06,
45.
08%,
a
nd
51.
93%
over
LEACH
c
onsideri
ng
500,
10
00,
15
00,
an
d
2000
sen
sor
de
vices,
resp
ect
ively
.
T
he
a
ver
a
ge ro
ut
ing
of over
hea
d red
uctio
n of
48.17%
is achi
eved by
EECR
O ov
e
r
t
he
LE
ACH.
Fig
ure
2
.
Net
w
ork
li
feti
m
e p
erfor
m
ance ev
al
uation co
ns
i
deri
ng
t
otal sens
or
d
e
vice d
e
at
h
Fig
ure
3.
Net
w
ork
li
feti
m
e p
erfor
m
ance ev
a
l
uation co
ns
i
deri
ng
first se
nsor
d
e
vice d
e
at
h
Fig
ure
4
.
Net
w
ork
li
feti
m
e p
erfor
m
ance ev
al
uation co
ns
i
deri
ng
l
os
s
of c
onnecti
vity
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.
10
, No
.
5
,
Oct
ob
e
r 2
020
:
47
98
-
48
08
4806
Fig
ure
5.
Com
m
un
ic
at
ion
ov
erh
ea
d per
f
or
m
ance e
valuati
on c
onsideri
ng
var
ie
d
se
nsor d
evice
Fig
ure
6
.
Ro
uting o
ve
rh
ea
d pe
rfor
m
ance ev
al
uation co
ns
i
deri
ng
va
ried
se
nsor
d
e
vice
3.3.
Res
ult
and discussi
on
over s
tate
-
of
-
ar
t
tec
hnique
This
work
car
ried
out
the
e
xp
e
rim
ent
evaluati
on
co
ns
i
de
rin
g
diff
e
ren
t
perform
ance
pa
ram
et
ers
li
ke
CO,
net
w
ork
li
feti
m
e
and
routin
g
ov
e
r
head
co
ns
i
der
i
ng
fir
st
sen
sor
de
vice
death,
LoC
a
nd
total
sens
or
dev
ic
e
death
.
This
sect
io
n
pa
rtic
ularly
est
im
at
es
the
evaluati
on
of
li
feti
m
e
per
f
or
m
ance
ov
e
r
the
sta
te
-
of
-
ar
t
m
et
ho
d.
Seve
r
al
existi
ng
m
eth
ods
ass
um
e
t
he
eval
uatio
n
of
li
feti
m
e
per
f
or
m
ance
co
ns
id
eri
ng
t
he
total
sens
or
dev
ic
e
deat
h.
Howe
ver,
the
evaluati
on
c
on
si
der
i
ng
1st
sens
or
de
vic
e
death
is
al
s
o
the
m
os
t
em
inent
perform
ance
par
am
et
er.
As
t
his
ou
tc
om
e
in
LoC
aff
ect
in
g
the
li
fetim
e
pe
rfor
m
ance
of
WSN.
As
an
outc
om
e,
this
pa
per
co
nsi
der
s
t
he
e
valu
at
ion
of
li
feti
m
e
perf
or
m
ance
consi
der
i
ng
fir
st
sens
or
de
vic
e
death
,
total
s
ens
or
dev
ic
e
d
eat
h,
and
L
oC.
Be
l
ow
in
Ta
ble
2
,
the
pe
r
form
ance
com
par
iso
n
of
int
rod
uced
EECR
O
a
nd
e
xisti
ng
protoc
ols
of
li
fetim
e
achievem
ent
ov
e
r
the
LEAC
H
prot
oco
l
is
ta
bula
te
d.
T
he
ov
e
ral
l
ou
tc
om
e
repr
esents
the
EECR
O
m
od
el
that
at
ta
ined
bette
r
pe
rfor
m
ance
im
pro
vem
ent
of
netw
ork
li
fetim
e
ov
er
sta
te
-
of
-
a
rt
m
od
el
[18
,
19
,
29
,
31
,
33]
co
ns
ide
rin
g
t
otal
sens
or
de
vice
death,
1st
se
nsor
de
vice
deat
h,
a
nd
L
oC.
T
he
m
os
t
i
m
po
rtant
ou
tc
om
e
is
achieved
in
this
wor
k
due
to
routing
a
nd
cl
us
te
r
op
ti
m
iz
a
ti
on
adoptin
g
the
de
sign
of
cro
ss
-
la
ye
r.
O
ur
m
od
el
re
duces
the
e
ne
rg
y
co
nsum
ption
of
CH,
im
pr
oving
c
overa
ge
ti
m
e
ai
din
g
in
li
fetim
e
perform
ance
im
pr
ov
em
ent
of
WSNs.
Th
us,
it
will
ai
d
i
n
pro
visio
n
in
g
t
he
ap
plica
ti
on
of
real
-
ti
m
e
serv
ic
e
that
needs the
d
esi
gn
of ene
rg
y
-
ef
f
ic
ie
nt.
Table
2
.
Per
for
m
ance co
m
par
ison o
f netw
ork
li
fetim
e achievem
ent o
ve
r
L
EACH
Alg
o
rith
m
Lif
eti
m
e i
m
p
rov
e
m
e
n
t achiev
ed
o
v
er
LE
AC
H
co
n
sid
ering
total
sen
so
r
d
ev
ice death
Lif
eti
m
e i
m
p
rov
e
m
e
n
t achiev
ed
o
v
er
LE
AC
H
co
n
sid
ering
f
irst
sen
so
r
d
ev
ice death
Lif
eti
m
e i
m
p
rov
e
m
e
n
t achiev
ed
ov
e
r
LE
ACH
co
n
sid
ering
los
s o
f
co
n
n
ectiv
ity
[
1
8
]
2
5
.0%
5
6
.7%
-
[
1
9
]
5
0
.0%
-
-
[
2
9
]
5
5
.0%
-
-
[
3
1
]
4
4
.0%
-
-
[
3
3
]
1
5
.0%
-
-
LL
E
ER
7
8
.02
%
8
5
.27
%
8
6
.7
3%
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
Ener
gy
ef
fi
ci
en
t cl
us
te
ring a
nd r
ou
ti
ng opti
miza
ti
on
m
od
el
for
…
(
Savi
th
a S
.
)
4807
4.
CONCL
US
I
O
N
Buil
ding
ene
r
gy
-
eff
ic
ie
nt
des
ign
for
prov
isi
on
i
ng
non
-
re
al
and
real
-
ti
m
e
app
li
cat
io
n
se
rv
ic
es
in
the
W
S
N,
w
hi
ch
is
ver
y
chal
le
ng
in
g.
An
e
xt
ensive
sur
vey
carried
out
shows
a
num
ber
of
ap
proac
hes
hav
e
been
re
prese
nted
la
te
ly
to
i
m
pro
ve
the
ene
r
gy
eff
ic
ie
ncy
of
SN
.
Am
on
g
them
,
cl
us
te
ring
ad
op
ti
ng
cr
oss
-
la
ye
r
play
an
i
m
po
r
ta
nt
ro
le
in
enh
a
ncin
g
the
perform
ance
of
the
sens
or
netw
ork.
H
ow
ever,
the
desi
gn
of
cro
ss
-
la
ye
r
a
rc
hitec
ture
with
m
ini
m
al
co
m
m
un
ic
at
ion
an
d
r
outi
ng
over
head
is
chall
e
ng
i
ng.
T
he
e
xi
sti
ng
m
od
el
did
no
t
con
si
der
c
overa
ge
ti
m
e
f
or
cl
us
te
r
opt
i
m
iz
ation
co
ns
iderin
g
heter
og
e
ne
ous
net
works.
As
a
res
ult,
incu
rs
ene
r
gy
ov
e
r
head
am
on
g
cl
ust
er
hea
d.
A
ff
ect
in
g
li
fetim
e
per
for
m
ance.
To
ov
erco
m
e
researc
h
chall
eng
e
s,
this
m
a
nu
s
cript
prese
nted
a
n
E
nerg
y
Eff
ic
ie
nt
Clu
ste
rin
g
an
d
Rou
ti
ng
O
ptim
iz
at
ion
m
od
el
adoptin
g
c
ro
s
s
-
la
ye
r
desig
n.
The
E
ECR
O
us
e
physi
cal
la
ye
r
inform
ation
t
o
ob
ta
in
c
hannel
gai
n
inf
or
m
at
ion
f
or
obta
inin
g
li
nk
qu
al
it
y
pa
ram
et
er.
The
n,
i
n
data
li
nk
or
MAC
la
ye
r
TDMA
base
d
com
m
un
ic
a
ti
on
is
adopted
f
or
c
omm
un
ic
ation
am
on
g
int
er
an
d
intracl
ust
er
com
m
un
ic
at
ion
.
I
n
the
ne
twork
la
ye
r,
cl
us
te
ri
ng
-
base
d
tra
ns
m
issi
on
or
r
ou
ti
ng
is
c
on
si
der
e
d.
F
ur
the
r,
cl
us
t
erin
g
a
nd
r
ou
ti
ng
o
ptim
iz
at
io
n
are
carried
out
a
nd
the
s
hortest
pa
th
base
d
r
ou
ti
ng
is
c
on
si
der
e
d
for
at
ta
inin
g
good
tra
de
-
off
be
tween
m
ini
m
i
zi
ng
the
energy
of
cl
us
te
r
hea
d,
enh
a
ncin
g
co
ver
a
ge
tim
e
a
nd
m
axi
m
iz
in
g
th
e
li
fetim
e
of
a
sens
or
netw
ork
.
The
ex
pe
rim
en
t
is
con
duct
ed
to
est
i
m
at
e
the
per
f
orm
ance
of
EE
CR
O
ov
er
the
existi
ng
m
od
el
.
The
outc
om
e
represe
nts
EECR
O
i
m
pr
oves
li
fetim
e
per
form
ance
of
78.
02%,
85.
27%,
and
86.73%
co
ns
ide
rin
g
total
sens
or
dev
ic
e
death
,
first
sens
or
dev
ic
e
death
,
and
l
os
s
of
connecti
vity
resp
ect
ively
.
The
ove
rall
aver
a
ge
perform
ance d
evelo
pm
ent o
f 83.3
5%
is at
ta
ined by the
pro
po
s
ed
EECR
O m
od
el
o
ver
L
E
ACH
c
onsideri
ng
all
the
cases
.
F
urt
her,
t
he
E
E
CR
O
m
od
el
r
edu
ce
s
c
ommun
ic
at
io
n
ove
r
head
a
nd
r
ou
ti
ng
over
head
ove
r
the
existi
ng
m
od
el
by
37.
37
%,
an
d
48.
17%
resp
ect
ively
.
The
overall
r
esult
a
tt
ai
ned
sh
ows
scal
able
li
fetim
e,
com
m
un
ic
at
ion
ov
e
rh
ea
d
a
nd
r
outi
ng
ove
r
head
pe
rfor
m
ance
c
on
si
der
i
ng
var
ie
d
net
work
de
ns
it
y.
Th
e
fu
t
ur
e
work
we
w
ou
l
d
c
on
si
der
e
ne
rg
y
c
on
s
um
ption
eval
uation
of
cl
us
te
r
hea
d
at
a
dif
fer
e
nt
le
vel
an
d
de
sign
an
op
ti
m
al
cluster
an
d hop se
le
ct
ion
desi
gn for f
ur
t
her
e
nhanci
ng the lifet
im
e
perform
ance o
f
W
S
N.
REFERE
NCE
S
[1]
N.
S.
Ali
,
Z.
A
.
A.
Al
y
asseri
,
an
d
A.
Abdulm
ohson,
“
Rea
l
-
ti
m
e
Hea
rt
Puls
e
Monitori
ng
Techni
q
ue
Us
ing
W
ire
l
e
ss
Sensor
Network
and
Mobil
e
Ap
pli
c
at
ion
,
”
In
te
r
nat
ion
al
Journa
l
of
Elec
tri
c
al
an
d
Com
pute
r
En
gine
er
ing,
vo
l.
8
,
no.
6
,
pp
.
5118
-
5126,
2018
.
[2]
K.
Pus
hpa,
S.
R
amac
handr
an,
a
nd
K.
R.
Kashw
an,
S.
,
“
A
Nove
l
Skeleton
Ext
ra
ct
ion
Algo
rit
hm
for
3d
W
ir
el
ess
Sensor Net
works
,
”
Globa
l
Journ
al
of
Com
pute
r S
ci
ence and
T
echnol
og
y
,
2014
.
[3]
S.
Cira
n
i and
M.
Picone,
“
W
ea
ra
ble
computing
f
or
the Int
ern
et of
Thi
ngs,
''
IT
Prof., vol
.
17,
no.
5,
pp.
35
-
41
,
2015
.
[4]
M.
Vijay
a
la
ksh
m
i
and
L
.
Kul
kami,
"Q
ual
i
t
y
awa
re
pro
toc
ol
to
support
m
ultim
edi
a
d
at
a
deliver
y
in
wire
le
s
s
net
work,"
3rd
In
te
rna
ti
ona
l
Conf
ere
nc
e
on
Devi
c
es,
Cir
cui
ts
and
S
y
stems
,
pp.
32
2
-
326,
2016
.
[5]
J.
Fan,
F.
Han
,
a
nd
H. Liu, ``
Cha
ll
eng
es
of
b
ig
da
ta
ana
l
y
sis
,
''
Nat
.
Sci. Re
v
.
,
vol
.
1
,
no
.
2
,
pp
.
293
-
314,
2014
.
[6]
Y.
Shi,
J.
Zha
n
g,
and
K.
B.
Letaief,
“
Group
sparse
bea
m
form
ing
for
gre
en
c
lo
ud
-
RAN
,
''
IE
EE
Tra
ns.
W
ire
l
ess
Com
m
un.
,
vol.
1
3,
no
.
5
,
pp
.
280
9
-
2823,
2014
.
[7]
M.
Peng,
S.
Y
a
n,
K.
Zha
ng,
an
d
C.
W
ang,
“
Fo
g
-
computing
-
bas
ed
rad
io
a
cc
ess
net
works
:
Iss
ues
and
ch
al
l
enge
s,
''
IEE
E
Netw.,
vo
l
.
30
,
no
.
4
,
pp
.
4
6
-
53,
2016
.
[8]
Y.
Shi,
J. Zhang
,
K
.
B
.
Le
t
ai
ef
,
a
nd
W
.
Chen
,
“
L
a
rge
-
sca
l
e conve
x
opti
m
izati
on
for
ult
r
a
-
dense
cl
ou
d
-
RAN
,
''
[9]
IEE
E
W
ire
l
ess Com
m
un.
,
vol.
2
2,
no
.
3
,
pp
.
84
-
91,
2015
.
[10]
M.
Chia
ng
and
T.
Zha
ng
,
“
Fog
and
IoT:
An
over
vie
w
of
rese
arch
opportuni
ti
es,
''
IEE
E
Inte
rne
t
Thi
ngs
J.,
vol.
3
,
no.
6
,
pp
.
854
-
8
64,
2016
.
[11]
B.
Prathi
ba
,
K.
J.
Sankar
,
and
V.
Sum
al
at
ha,
“
Novel
fra
m
ework
of
ret
ai
ning
m
axi
m
um
dat
a
qual
ity
and
ene
rg
y
eff
iciency
in
r
ec
onfigur
abl
e
wire
le
ss
sensor
net
work”
,
In
t
ern
ational
Journal
of
Elec
tri
c
al
and
Com
put
er
Engi
ne
eri
ng,
vo
l
.
9,
no.
4
,
pp.
2
983
-
2901,
2019.
I.
F.
Ak
y
i
ldi
z
,
W
.
Su,
Y.
Sankar
asubra
m
aniam
,
and
E.
C
a
y
ir
ci,
“
W
ire
le
ss
sensor ne
tworks:
A
surve
y
,
”
Com
put.
Netw.
,
vol. 38, n
o.
4
,
pp
.
393
-
42
2,
2002
.
[12]
A.
Na
y
ak
and
I.
Stojm
enovi
c,
“
W
ire
le
ss
Sensor
and
Actua
tor
Networks:
Algori
thms
and
Protocol
s
for
Scal
ab
le
Coor
dina
ti
on
an
d
Data Com
m
un
ic
a
ti
on,
”
Hobok
e
n,
NJ
,
USA:
Wi
l
ey
,
2010.
[13]
W
.
Dargi
e,
“
D
y
namic
power
m
ana
gement
in
wir
el
ess
sensor
netw
orks:
Stat
e
-
of
-
t
he
-
art,”
IE
EE
Se
nsor
s
J.
,
vol.
12,
no.
5
,
pp
.
1518
–
1528,
2012
.
[14]
M.
Sale
em,
G
.
A.
Di
Caro,
an
d
M.
Farooq,
“
Sw
arm
int
el
li
g
e
nce
base
d
routing
protoc
ol
for
wire
le
ss
sensor
net
works
:
Surve
y
and
fu
ture di
r
e
ct
ions,
”
In
f. Sc
i
.
,
vol. 181, no. 20, pp. 4597
–
4624,
2011.
[15]
A
.
M
a
h
b
o
u
b
,
M
.
A
r
i
o
u
a
,
E
l
M
.
En
-
N
a
i
m
i
,
“
E
n
e
r
g
y
-
E
f
f
i
c
i
e
n
t
H
y
b
r
i
d
K
-
M
e
a
n
s
A
l
g
o
r
i
t
h
m
f
o
r
C
l
u
s
t
e
r
e
d
W
i
r
e
l
e
s
s
S
e
n
s
o
r
N
e
t
w
o
r
k
s
,
”
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
o
l
.
7
,
n
o
.
4
,
p
p
.
2
0
5
4
-
2
0
6
0
,
2
0
1
7
.
[16]
D.
V.
Pus
pal
at
a
and
P.
Na
y
ak,
“
A
cl
usteri
ng
algorithm
for
W
S
N
to
opti
m
iz
e
t
he
net
work
l
ife
t
ime
usin
g
t
y
p
e
-
2
fuz
z
y
logi
c
m
odel
,
”
Proc. 3rd
In
t.
Con
f. A
rti
f. Intell., Mode
l
ing
S
i
mulati
ons (
AIM
S
)
,
Malay
sia,
pp.
53
–
58,
2015
.
[17]
S.
Band
y
opadh
ya
y
and
E
.
J.
C
o
y
l
e,
“
An
ene
rg
y
eff
i
cient
h
ie
r
a
rch
ical
cl
ust
eri
n
g
al
gor
it
hm
fo
r
wire
l
ess
sensor
net
works
,
”
IEEE
INFOCOM,
vo
l
.
3
,
2003
.
[18]
A
.
M
a
h
b
o
u
b
,
e
t
a
l
.
,
“
A
n
e
n
e
r
g
y
-
e
f
f
i
c
i
e
n
t
c
l
u
s
t
e
r
i
n
g
p
r
o
t
o
c
o
l
u
s
i
n
g
f
u
z
z
y
l
o
g
i
c
a
n
d
n
e
t
w
o
r
k
s
e
g
m
e
n
t
a
t
i
o
n
f
o
r
h
e
t
e
r
o
g
e
n
e
o
u
s
W
S
N
”
,
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
o
l
.
9
,
n
o
.
5
,
p
p
.
4
1
9
2
-
4
2
0
3
,
2
0
1
9
.
[19]
P.
Na
y
ak
and
B.
Vath
asa
vai,
"
Ene
rg
y
Eff
ic
i
e
nt
Cluste
r
ing
Al
gorit
hm
for
Multi
-
Hop
W
ireles
s
Sensor
Networ
k
Us
ing
T
y
pe
-
2
Fuzz
y
Logi
c
,
"
I
E
EE
S
ensors
Journal,
vo
l. 17, no.
14,
pp
.
4492
-
44
99,
2017
.
Evaluation Warning : The document was created with Spire.PDF for Python.