TELKOM
NIKA
, Vol.14, No
.2, June 20
16
, pp. 531~5
3
7
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v14i1.3375
531
Re
cei
v
ed
Jan
uary 13, 201
6
;
Revi
sed Ma
rch 2
2
, 2015;
Acce
pted April 8, 2016
Improved Energy Aware Cluster based Data Routing
Scheme for WSN
Khalid Ha
se
eb*
1
, Kamalr
ulnizam Abu
Bakar
2
, Abdul Hanan
Abdullah
3
,
Tasne
e
m Da
r
w
i
s
h
4
, Fase
e Ullah
5
, Adn
a
n Ahm
e
d
6
1,2,
3,4,
5
Facult
y
o
f
Computin
g, Universiti T
e
knol
ogi Ma
la
ysi
a
, UT
M,
Skuda
i,81
310 Johor,
Mal
a
ysi
a
6
Computer S
y
s
t
ems Departm
ent, Quaid-e-A
w
a
m
Univ
ersi
t
y
of Engine
eri
n
g
,
Science an
d
T
e
chnolog
y,
Na
w
a
bsh
ah, P
a
kistan
*Corres
p
p
o
n
d
i
ng auth
o
r, ema
il: khali
dutm.pc
rg@gma
il.com
1
, knizam@utm.my
2
, hana
n@
utm.m
y
3
,
eng
_simsim
8
3
@
hotmai
l
.com
4
, faseekha
n@g
m
ail.com
5
, adn
an.ahm
ed
03@
ya
ho
o.com
6
A
b
st
r
a
ct
W
i
reless se
ns
or netw
o
rk (W
SN) consists o
f
several tiny
d
e
vices th
at are
dispers
ed ra
n
d
o
m
ly for
gather
ing
netw
o
rk fiel
d. Clust
erin
g mech
ani
sm d
i
vid
e
s
the
W
S
N into d
i
fferent su
b-reg
i
o
n
s call
ed c
l
ust
e
rs.
Indivi
dua
l cl
uster is c
onsisti
n
g
of cl
uster
he
ad (C
H)
a
nd
me
mber
no
des
. T
he
mai
n
res
earch c
h
a
lle
ng
e
s
beh
ind c
l
uster
i
ng
mec
h
a
n
is
m are to o
p
ti
mi
ze netw
o
rk ov
erhea
ds w
i
th effi
cient d
a
ta d
e
liv
ery. Sensor
no
des
are o
per
ated
b
y
batteries
an
d
practica
lly it
is
not f
eas
ibl
e
to
repl
ace th
e
m
duri
ng se
nsi
n
g
the e
n
viro
n
m
e
n
t
so e
ner
gy sh
o
u
ld
be
effectiv
ely
utili
z
e
d
a
m
ong
se
ns
ors fo
r i
m
prov
in
g
ov
erall
n
e
tw
ork p
e
rformanc
e. T
h
is
researc
h
pap
e
r
presents an
improve
d
en
er
gy aw
are clus
ter based d
a
ta routin
g (i-ECBR) sche
m
e,
by
divid
i
n
g
the ne
tw
ork regions i
n
to unifor
m
si
zed squ
a
re parti
tions an
d loca
li
z
e
d
CH el
ectio
n
mec
h
a
n
is
m. In
add
ition, co
nsi
s
tent end-to-
e
nd d
a
ta
routi
n
g is perfor
m
ed
for improv
i
n
g
data d
i
sse
mi
natio
n. Si
mul
a
tio
n
results i
llustra
te that o
u
r p
r
opos
ed sc
he
me
out
per
for
m
s th
an
exist
i
ng w
o
rk
in t
e
rms
of d
i
fferen
t
perfor
m
ance metrics.
Ke
y
w
ords
: clu
s
tering, netw
o
r
k
lifetime, ener
gy consu
m
ptio
n, route discov
e
ry, clusters mana
ge
me
nt
Copy
right
©
2016 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
WSN co
nsi
s
t
s
of several
sensor node
s with
on
e o
r
many si
nk
no
des th
at are
scattered
in a physi
cal
environm
ent
to sen
s
e th
e events a
n
d
gatheri
ng d
a
ta. Senso
r
node
s sen
s
e
the
environ
ment
gatheri
ng info
rmation, integ
r
ate t
he data
and send the
achi
eved info
rmation to si
n
k
node
or ba
se
station
(BS).
The
sin
k
i
n
turn
que
rie
s
the sen
s
or n
ode
s for information [1]. T
he
sen
s
o
r
n
ode
s a
r
e
sm
all
device
s
with
sm
all batte
ries
and
re
ch
argin
g
of th
ese
batteri
es i
s
impra
c
tical in
deploye
d
scenari
o
s. T
h
e
r
efore to
re
du
ce e
n
e
r
gy u
s
age a
nd
prol
ong lifetime i
s
main de
sign
crite
r
ia for
se
nso
r
net
works. Struct
u
r
e a
nd un
stru
ctured are two m
a
in cate
go
rie
s
of
WSNs. Un
structured WS
N con
s
ist of de
nse
colle
ct
ion
of wirele
ss
sensor
no
de
s. After deploye
d
,
the nod
es l
e
ft unattend
ed to d
o
da
ta gathe
ring
and
rep
o
rti
ng. Man
age
con
n
e
c
tivity an
d
detectin
g
no
d
e
s o
r
lin
k fail
ure
s
i
s
difficul
t
in
unst
r
u
c
tured WS
N b
e
cause t
here
i
s
large num
ber
of
node
s.
On other side
sen
s
or
no
des are depl
oyed i
n
pre defin
e
d
man
ner i
n
stru
ctured
WSN.
Structu
r
ed n
e
twork redu
ces net
work
maintena
nce
and expen
ses [2]. Adhoc routin
g pro
t
ocol
s
are
not fe
asi
b
le in
WS
N
due to
limite
d
en
ergy
and
co
mputing
capabilitie
s. S
ensor nod
es
are
limited in
me
mory, ba
nd
wi
dth, tran
smi
s
sion
po
we
r a
nd e
nergy. Such
types of
con
s
trai
nts
h
a
ve
posed many rese
arch issu
es to
prolong
netwo
rk lifeti
m
e.
WSN
con
s
i
s
t numbe
r of
node
s with li
mited memo
ry, proce
s
sing
, powe
r
and
energy
resou
r
ces.
Due to
such li
mited resou
r
ce
s m
onitori
n
g
the
interested phy
sical
environ
ment
by
nume
r
ou
s
se
nsin
g devi
c
e
s
for a l
ong
pe
riod
of time
make
s it a
ch
allengin
g
ta
sk. Saving e
n
e
rgy
to enh
an
ce t
he n
e
two
r
k’
s
lifetime is a
critical
proble
m
in
WSN a
n
d
the
r
efore
“h
ow to
en
han
ced
and imp
r
ove
the WSN lif
etime” i
s
a
cruci
a
l qu
estio
n
. The p
r
ima
r
y obje
c
tive of WSN i
s
d
a
ta
colle
ction a
n
d
tran
smissio
n
. Free
spa
c
e and mult
ip
ath fading a
r
e two types
of model
s in
for
WSN. The
r
e i
s
a nonlin
ea
r relation
shi
p
b
e
twee
n each node in term
s of the energ
y
consumptio
n
of wirele
ss communi
catio
n
. The
ene
rg
y con
s
um
ptio
n of fre
e
spa
c
e m
odel
is
much
smalle
r than
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 2, June 20
16 : 531 – 53
7
532
that of multip
ath fadin
g
m
odel.
Whe
n
sensor nod
e i
s
fa
r a
w
ay f
r
om BS, it ma
y be
run
out
of
energy sho
r
tl
y that would lead to quality
redu
ction of netwo
rk [3].
Dynami
c
hi
erarchical routi
ng p
r
oto
c
ol L
EA
CH[4]
con
s
ist
s
of
seve
ral ro
und
s a
n
d
ea
ch
singl
e ha
s se
tup and
stead
y phases. All
sen
s
o
r
no
de
s con
s
i
s
t of ho
mogen
eou
s
structu
r
e. CH i
s
sele
cted in
setup pha
se
a
nd then dyn
a
m
ically differ
ent clu
s
ters a
r
e form
ed. Ea
ch
clu
s
ter h
a
s
its
local
CH an
d sen
s
o
r
nod
es. Selected CH se
nd
s an a
d
vertise
d
me
ssage to all n
ode
s and no
d
e
s
respons
e
on that rec
e
ived mess
age. Aft
e
r
rec
e
iving all the data
from nodes
CH aggregate
and
comp
re
ss the
m
and tra
n
sf
er to sin
k
no
d
e
. LEACH
sel
e
cts the
CH perio
dically. Ran
dom nu
m
ber
is ge
nerated
by each no
de
and the
nod
e
will be
sel
e
ct
ed a
s
CH if th
is rando
m nu
mber i
s
small
e
r
than predefin
e thre
shol
d value. The
pre
define value
set to zero if
this same n
o
de ha
s ele
c
t
e
d
again
a
s
a
CH
fo
r sam
e
round.
In ste
ady
ph
as
e T
D
MA i
s
used
so
that
every membe
r
no
de
sen
d
data to
CH
wh
en
re
ceive its o
w
n t
i
me sl
ot. The
thre
shold
formula for
LEACH i
s
sho
w
n
in
Equation (1).
∗
∊
0
(1)
P sh
ows th
e
desi
r
ed
pe
rce
n
tage
of
CHs,
r
den
ot
es the
cu
rrent
rou
n
d, and
G
is th
e set of
node
s that ha
ve not been e
l
ected a
s
CHs in the last 1
/
p round.
LEACH-B propo
sed th
at the proto
c
ol
need
s to e
n
sure that th
e
partition of
cl
uster i
s
balan
ce
and
uniform to
sa
ve the ene
rg
y con
s
umptio
n amon
g no
d
e
s a
nd to p
r
o
l
ong the lifeti
m
e
of the netwo
rk. To a
c
comp
lish this
obje
c
tive, the number of CHs n
eed
s to be do
minated a
nd the
netwo
rk
need
s an o
p
timal CHs. Authors in [5] hav
e proved that WSN will be e
n
e
rgy efficie
n
t if
the net
work
has bet
wee
n
3 a
nd
5
clu
s
ters from th
e total 1
00
sensor
no
de
s. It mea
n
s the
optimal pe
rcentage of CHs
rang
e fro
m
3% to
5%. LEACH-B i
m
prove
d
the
original LEA
C
H
algorith
m
by takin
g
the nod
e’s resid
ual e
nergy
into
co
nsid
eratio
n a
nd ke
epin
g
the con
s
tant an
d
near o
p
timal
numbe
r of CHs at ea
ch
ro
und.
Authors proposed Energy E
fficient Extended LEACH (EEE
LEACH) [6] prot
ocol t
o
enha
nce the
netwo
rk lifetime an
d e
nergy efficien
cy
by introd
uci
n
g an
app
ro
ach to mini
mize
th
e
comm
uni
cati
on di
stan
ce a
m
ong n
ode
s
based on
mu
lti le
vel cluste
ring. Ene
r
gy
efficien
cy of the
netwo
rk may
be in
cre
a
sed
by kee
p
ing
ra
dio commu
ni
cation dista
n
ce
minimum. Network
lifetime
and en
ergy efficien
cy ca
n be en
han
cing by in
cr
ease the nu
mber of
clu
s
ters b
e
cau
s
e
it
decrea
s
e
s
ra
tion co
mmun
i
cation
dista
n
c
e. It divide
s the net
work into two l
a
yers for
clu
s
t
e
r
formulatio
n. CHs sele
ctio
n and mem
b
er nod
es
sen
d
their data i
n
first layer. The procedu
re of
CH sele
ction
is
same
a
s
i
n
LEACH. In
seco
nd l
a
yer
CHs fin
d
the
nearest
ma
ster
CH in
orde
r to
forwa
r
d its d
a
ta .When m
a
ster
CH
re
ceived dat
a from CHs the
n
it aggreg
ate and comp
ress
them to forward towards
BS. EEE
LEACH improved the network lifet
ime by introduc
i
ng the role
of master
CHs in se
co
nd l
a
yer but it ca
use
s
ro
uting
overhe
ad
s that gives co
ng
estion
s an
d d
e
lay
probl
em
s in the network.
Authors pro
p
o
se
d [7] clust
e
r ba
sed
k-m
ean
s prot
o
c
ol
to form the
k-clu
s
ters of obje
c
ts
based
on th
e
eu
clidea
n di
stan
ce. Initiall
y the CHs
a
r
e sele
cted
on
the b
a
si
s
of
rand
om
num
ber
as in LEACH to form the initial cluste
rs. Then afte
r the formation
of initial clust
e
rs, centroid is
determi
ned
to find th
e
ce
nter lo
catio
n
of ea
ch
clu
s
t
e
r a
nd th
e n
ode i
s
sel
e
ct
ed a
s
a n
e
w CH
whi
c
h i
s
clo
s
er to
ce
ntroi
d
. The
role
o
f
CH i
s
rotat
ed when
en
e
r
gy is drain
out from
ce
rt
ain
energy thre
shold. It prolo
ngs n
e
two
r
k lifetim
e and
energy effici
ency of WS
N than p
r
op
ose
d
s
c
hemes
[4, 8].
A New
Routin
g Proto
c
ol for Efficient and
Secu
re
WSN
[9] improve
s
netwo
rk lifeti
m
e an
d
energy efficie
n
cy . It is round ba
se
d schem
e and
netwo
rk a
r
e
a
is divided in
to a numbe
r of
squ
a
re
s, ea
ch
ha
s sa
me numbe
r nod
e
s
.
Squ
a
re
s
remain
same and do not chang
e
in
eve
r
y
roun
d. A squ
a
re i
s
a
clu
s
t
e
r a
nd all
clu
s
ters d
o
not
cha
nge i
n
all
rou
n
d
s
. A cl
uster con
s
ist
s
of
four cell
s. CHs are el
ecte
d in each
cell at
the st
art of new ro
und. Da
ta is send to
ward
s sin
k
no
de
in multi h
op
manne
r. T
o
p
r
ovide
se
cu
re
co
mmuni
cati
on am
ong
no
des,
setu
p
se
rver
dist
ribut
e
s
different m
a
n
ageme
n
t keys in
ea
ch
ro
u
nd. It bala
n
ces th
e e
nerg
y
con
s
um
ptio
n amo
ng
se
n
s
or
node
s an
d gi
ves se
cu
re communi
catio
n
mech
ani
sm
for WSN.
In ene
rgy effi
cient
clu
s
teri
ng sch
e
me
(EEC
S) [10]
candid
a
tes
no
des co
mpete
for the
ability to elevate to CH for
a current round. Each
round has fixed time inte
rval and re-clusteri
ng
occurs at the
start of next roun
d. In this app
ro
ach candid
a
te nod
es b
r
oa
dcast
their re
sidu
a
l
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Im
proved En
erg
y
Awa
r
e Cluster b
a
sed
Data Routing
Schem
e for
WSN (K
halid
Ha
see
b
)
533
energy to neighbo
ring n
o
d
e
s. If a given
node d
oes
n
o
t find a node with more re
sidual ene
rgy, it
become
s
a
CH. EE
CS u
s
e
different
way fo
r
cl
u
s
ter fo
rmation
than LEA
CH.
LEACH p
e
rf
orm
s
clu
s
ter form
ul
ation ba
sed o
n
the minimu
m distan
ce of
sen
s
or n
ode
s to their corresp
ondi
ng CH.
EECS extends LEACH p
r
ot
ocol by dyna
mic si
zing of
clu
s
ters ba
se
d on clu
s
ter d
i
stan
ce from the
BS.
Tree b
a
sed
clu
s
terin
g
(T
BC) [11] is
anothe
r ene
rgy efficient appro
a
ch that aims to
prolo
ng net
work lifetime. At the start of each
rou
nd,
nodes a
r
e el
ected a
s
CHs on the basi
s
of
stocha
stic
proce
s
s. Ea
ch
clu
s
ter maint
a
ins a t
r
ee
s
t
r
u
c
t
ur
e us
in
g d
i
s
t
an
ce
fac
t
o
r
w
h
e
r
e e
l
ecte
d
CH i
s
th
e ro
ot of it. Nod
e
s
sen
d
thei
r data p
a
cket
s to p
a
re
nt
node i
n
si
de t
he tre
e
an
d
CH
forwa
r
d
s
the aggregate
d
d
a
ta towards
BS directly
. Although, it significa
ntly improved n
e
two
r
k
lifetime by decre
ases th
e long di
stan
ce
commu
nication du
ring int
r
a-cl
uste
r com
m
unication b
u
t
at the start of each
roun
d random
CHs
sele
ction
cau
s
e
s
unb
alan
ced clu
s
ters fo
rmation.Ta
ble
1
pre
s
ent
s the summ
ary of energy efficie
n
t
cluste
r
base
d
routing a
p
p
r
oa
che
s
that have discu
ssed
in this section.
LEACH-DT [12] take
s into acco
unt the dist
an
ce
s of sen
s
o
r
nod
es toward
s BS in orde
r
to app
oint th
e
set
of
CHs a
nd b
a
lan
c
e
s
t
he e
n
e
r
gy d
e
p
letion i
n
n
e
twork field. Ba
sically, LEACH-
DT is a custo
m
ized ve
rsio
n of LEACH proto
c
ol in
which p
r
ob
abili
ty of CH elect
i
on is modifie
d
by
inco
rpo
r
ating
distan
ce fact
or. Du
ring da
ta fo
rwa
r
ding,
source
CH b
r
oad
ca
st ADV message a
n
d
based
on l
e
a
s
t di
stan
ce, t
he n
e
xt-hop
relay no
de i
s
el
ecte
d. In thi
s
way, shorte
st multi-hop
route
is a
c
compli
shed to
wa
rd
s
BS. Although
, LEACH-DT
improved
ne
twork lifetime
as compa
r
e
to
origin
al LEA
CH, h
o
wever,
the num
ber
of clu
s
te
rs a
r
e not unifo
rm
ly distrib
u
ted
and
CH
ele
c
tion
mech
ani
sm is non-o
p
timize
d.
Table1. Meth
odolo
g
y and
Problem
s of Different
Cl
ustering Ba
sed
Energy Efficient Schem
es
Energ
y
Efficient
Protocols in WS
N
Methodolog
y
Clusters
Formati
o
n
Issues
LEACH
Probabilistic
based
Predefined C
H
s and residual
energ
y
is not considered for C
H
selection
LEACH-B
Random
number an
d
residual energ
y
Single hop and e
nerg
y
distribution is im
balancce.
EEE LEACH
Randoml
y
Sub-optimal clusters formation
K-means CH
selection
Euclidean clusteri
ng overheads and uneven
load distribution among clusters
Routing
Protocol for
Efficient and
Secure WSN
Grid based
Communication overheads
EECS Node
residual
energ
y
Unbalance ener
g
y
consumption
in network field
TBC
Randoml
y
Sub-optimal
clusters generations
and excessive energ
y
consumption
during inter-clust
er communications
LEACH-D
T
Distance
based
Addi
tional energ
y
consumption
and net
w
o
rk over
heads
Energy effici
e
n
cy with
ro
bu
st ro
uting i
s
a
vital task
am
ong
re
sea
r
ch
comm
unity for
WSN
appli
c
ation
s
.
Du
e to
limi
t
ed con
s
trai
n
t
s on
the
p
a
rt of
nod
es, relia
ble
an
d efficie
n
t d
a
ta
forwa
r
di
ng
sh
ould
be
con
s
i
dere
d
while
p
r
opo
sin
g
e
n
e
r
gy efficie
n
t schem
e. In a
d
d
ition, the
CHs
being a focal
point within clu
s
ter bo
und
ary, its
appro
p
riate sele
ction incr
ea
se
s netwo
rk
lifeti
m
e
and imp
r
ove
s
net
work
conne
ctivity.
Thus,
de
sign
of energy e
ffiicient clu
s
t
e
r ba
se
d ro
uting
scheme i
s
a
n
key re
se
arch p
r
obl
em for en
han
ci
ng
overall net
work
perfo
rma
n
ce [13, 1
4
]. Th
e
rest
of the
re
sea
r
ch
pape
r is organi
ze
d
as foll
ows. O
u
r
i
-
ECB
R
al
gorithm
is pr
es
e
n
t
ed
in se
ctio
n
2. Re
sea
r
ch
method i
s
sh
own i
n
sectio
n 3. Simulati
on re
sult
s an
d discu
s
sion
s are
cove
red
in
se
ction 4. Section 5 con
c
lu
des the p
ape
r and su
gge
st
s for future
work.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 2, June 20
16 : 531 – 53
7
534
2. Proposed
Metho
d
In this sectio
n we presen
t detailed de
scription of
i-ECBR
sch
e
m
e, includi
ng
how to
partition th
e
sen
s
in
g field
into unifo
rm
sized
squ
a
re
partition
s a
n
d
initiates CH sele
ction
within
each pa
rtition. Furthe
rmo
r
e, the r
outin
g me
chani
sm
for data
dissemin
ation i
s
also
discu
ssed.
i-ECBR
i
s
divi
ded i
n
to two
main
pha
se
s.
Firstly,
i-ECB
R
pe
r
f
or
ms
no
d
e
s
c
l
us
te
r
i
n
g
an
d
s
e
c
ond
ly
optimal path
s
are co
nstructed for robu
st
routing. Before goi
ng to descri
be
i-E
C
BR
schem
e, we
highlight the
variou
s ne
wo
rk a
s
sump
tio
n
s that are listed as follo
ws:
1.
All nodes a
r
e
deploye
d
ran
domly and re
main static.
2.
All nodes h
a
ve homog
eno
u
s
structu
r
e an
d sen
s
e d
a
ta perio
dically.
3.
BS has location inform
atio
n of all sen
s
o
r
node
s.
4.
All nodes
will use the
same
radio chan
ne
l for comm
uni
cation
with ea
ch othe
r.
5.
All the nodes
comm
uni
cate
via a share
d
bidire
ction
a
l wirel
e
ss chan
nel.
6.
No ne
w no
de
s ca
n be ad
d
ed after network d
eploym
e
nt.
The ste
p
s un
derta
ken in
i-ECBR
sch
e
m
e
are di
scussed as follo
ws:
Step 1:
Su
ppo
s
e
tha
t
n
n
u
m
ber
of sen
s
or n
ode
s i
s
randomly
depl
oyed in n
e
twork field.
At the beginn
ing, each se
n
s
or
node
sen
d
s its lo
ca
tion
information t
o
wa
rd
s BS using a
pprop
ri
ate
adja
c
ent n
o
d
e
. Base
d o
n
optimal n
u
me
ral of
clu
s
ters (p
) a
nd n
ode
s di
strib
u
tion
(n), BS i
n
itiate
s
the pro
c
e
d
u
r
e of virtual n
e
twork p
o
rtio
ning by di
vidi
ng entire sen
s
or field into
identical si
zed
squ
a
re
pa
rtitions via
eq. 2,
whe
r
e
strong
pos
itive and negative
lin
e
a
r correl
ation
exists betwe
en
input
n
and out
put
z
. Next, by
usin
g nod
es
positio
n, a ce
ntro
id for e
a
ch partition is
determi
ned.
∗
(2)
Step 2:
The
nodes
that
are relatively c
l
os
er to
wards res
p
ec
tive centroid are grouped
into singl
e clu
s
ter bo
und
ary
with a uniqu
e ID.
Step 3:
After the formatio
n of different
cluste
rs
, a locali
ze
d CH electio
n
is an
noun
ce
d
among
mem
ber
nod
es th
ereby
only li
mited num
be
r of no
de
s i
s
pa
rtiticip
ate
d
and
re
du
ci
ng
comm
uni
cati
on ove
r
he
ad
s. Duri
ng
ele
c
tion me
ch
ani
sm, node’
s l
o
cal information
is
sum
m
ed
up
in weig
hted
method ba
se
d on eq.3, where
1
.
Finally, nodes tho
s
e are rel
a
tively
nearer to ce
ntroid with h
i
gh ene
rgy condition an
d
neighb
or’
s
den
sity are electe
d as
CHs.
Afterwa
r
ds;
e
a
ch ele
c
ted CH
i
n
form
s
i
t
s
stat
u
s
to
all memb
er
node
s by b
r
oad
ca
sting A
D
V
messag
e.
∗
∗
∗
(
3
)
Step 4:
To
receive
and
aggregate th
e se
nsory d
a
ta from n
o
d
e
s, all
sele
ct
ed CHs
anno
un
ced t
heir lo
cal T
D
MA sche
du
les. Accordin
gly, each no
de se
nd
s its data towa
rds
asso
ciated
CH wh
en it receives time slo
t.
Step 5:
Su
bseque
ntly, to achieve
data t
r
ansmi
ssion, i
n
terme
d
iate
node
s a
r
e
el
ected
to
con
s
tru
c
t
rou
t
ing path
s
. E
a
ch
source
n
ode
deter
min
e
s
a comp
osi
t
e routin
g fun
c
tion
ba
sed
on
resi
dual
en
ergy, distan
ce
and lin
k
qual
ity factors.
T
he routing
fu
nction
give
s
energy efficie
n
t,
relatively clo
s
er
with lea
s
t estimated
transmi
ssi
o
n
time neighbor n
ode a
s
next-ho
p. This
pro
c
ed
ure
continue
d u
n
til an
optimi
z
ed
multi-
ho
p ro
uting
p
a
th is con
s
tructe
d to
wa
rds
destin
a
tion.
Step 6:
In a
ddition, to balance the en
ergy
co
nsum
ption amon
g data forward
e
r nod
es,
their role
i
s
shifted
du
ring
data
di
ssemi
nation wh
e
n
energy level i
s
rea
c
he
d to
certai
n th
re
sh
old.
In such ca
se,
current data
forwa
r
de
r no
de set its ne
xt-hop flag to false and se
nds route error
messag
e toward
s source
node. Acco
rd
ingly, so
u
r
ce node di
scove
r
s alte
rnative
data forward
e
r
node
ba
se
d
on
comp
osite
ro
uting fun
c
t
i
on a
s
menti
oned
in
step
5, whi
c
h
resu
lts in i
m
proving
data delivery
perfo
rman
ce
and re
du
cing
netwo
rk late
n
c
y.
Step 7:
As
in
d
a
t
a
r
o
u
t
ing
,
th
e
CH
s
c
o
ns
u
m
e
d
exc
e
ss
ive
en
er
g
y
co
ns
u
m
ptio
n
and
highly involve
d
in ro
utes
a
d
justme
nt proce
s
s. Unli
ke
mo
st
of
existing
schem
es,
those
p
e
rfo
r
m
re-ele
ction m
e
ch
ani
sm on
periodi
c ba
sis and lea
d
to additional
netwo
rk ove
r
head
s with h
i
gh
comm
uni
cat
i
on co
st
.
i-ECBR
rotate
s the role of ele
c
ted CH
s ba
sed on certain
event i.e. when
the resi
dual e
nergy of cu
rrent CH
is d
r
o
p
s to ce
rtain threshold, an
appro
p
riate
node is
sele
cted
as n
e
xt CH
by usin
g ste
p
3.
Afterwards, ne
wly el
ected
CHs flood
s thei
r st
atus info
rmat
ion
among m
e
mb
er nod
es a
nd
update the transmi
ssion
sche
dule for d
a
ta dissemin
ation.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Im
proved En
erg
y
Awa
r
e Cluster b
a
sed
Data Routing
Schem
e for
WSN (K
halid
Ha
see
b
)
535
i-ECBR
sche
me improve
d
energy
con
s
ervation an
d routing p
e
rfo
r
mance as
co
mpared
to
existing
solution
s du
e
to som
e
a
s
p
e
cts.
Firstl
y, distin
ct re
gio
n
s
are
con
s
tructed
ba
se
d
on
node
s
distri
b
u
tion rathe
r
t
han
pre
defin
ed n
e
two
r
k p
a
rtitions. In
a
ddition, o
p
timized
candid
a
tes
are sele
cted for the role o
f
CHs with m
i
nimu
m overh
ead
s and en
ergy co
nsum
ption. Secon
d
ly,
based on a
d
aptive routing
mecha
n
ism, interme
d
ia
te
node
s are ap
pointed a
s
n
e
xt-hop for d
a
ta
forwa
r
di
ng th
ereby imp
r
ov
ed network throu
ghp
ut. Moreove
r
, to b
a
lan
c
e the traffic load am
ong
routing
path
s
,
the
role
of d
a
ta forwa
r
de
r nod
es a
r
e
sh
ifted ba
sed
o
n
certai
n eve
n
t. At the en
d
,
in
orde
r to
imp
r
ove net
work
con
n
e
c
tivity and lifetime,
the po
sition
of CHs is rotated in
dyna
mic
manne
r that
leads to e
nergy bal
an
cing am
ong
nodes a
n
d
prolon
ging
overall net
work
perfo
rman
ce.
3. Rese
arch
Metho
d
In this
se
cti
on
we
pre
s
e
n
t ou
r
simul
a
tion mo
del
that ha
s b
e
en u
s
e
d
in
different
experim
ents
by using
well
kno
w
n tool
netwo
rk
simu
lator (NS2) [
15]. We ra
nd
omly deploye
d
varying sen
s
or no
de
s in
sen
s
o
r
field
of 100 X 10
0 dimen
s
io
n. All sen
s
o
r
n
ode
s are lo
cation
awa
r
e via GP
S or any po
si
tioning alg
o
rit
h
m. Energy
model a
s
sum
ed as
being
u
s
ed in [4] an
d
a
con
s
id
ere
d
free
spa
c
e
radi
o p
r
opa
gation
mod
e
l.
The
system p
a
ra
m
e
ters that
hav
e be
en
used i
n
experim
ents are sh
own
i
n
Tabl
e
2. We com
p
a
r
e
d
i-E
C
BR
scheme with relavent clu
s
tering
approa
che
s
i
.
e. LEACH-DT and K-me
ans
with re
spect to diffe
rent pe
rform
ance evaluti
on
metrics
.
Table 2. System Param
e
te
rs
Parameter Value
Net
w
ork a
r
ea
100*100m
Nodes
50-300
Initial energ
y
2J
Energ
y
threshold
50%
Optimal no. of clusters
5%
Data packet size
100 b
y
t
e
s
Base station energ
y
100
Transpo
rt la
y
e
r p
r
otocol
UDP
Simulation time
1000sec
4. Results a
nd Discu
ssi
on
In this secti
on we u
s
ed
three
different pe
rform
a
nce
metri
cs:
clu
s
terin
g
o
v
erhea
ds,
averag
e end
-to-e
nd del
ay and network thro
ughp
ut
to evaluate the perform
ance of
i-ECBR
scheme
with
relavent
clu
s
t
e
ring
sch
e
me
s. The
sim
u
la
tion re
sult
s of
perfo
rme
d
e
x
perime
n
ts a
r
e
discu
s
sed in
sub
-
sequ
ent se
ction
s
.
4.1. Cluste
ring Ov
erheads
Figure 1
sho
w
s th
e cl
uste
ring ove
r
h
e
a
d
s a
gain
s
t va
rying nu
mbe
r
of node
s. O
b
viously,
the cl
uste
ring
ene
rgy in
cre
a
se
s
wh
en
n
u
mbe
r
of
nod
es
are
in
crea
se i
n
n
e
two
r
k field. Thi
s
i
s
due
to with incre
a
se in n
ode
s density, the
numbe
r of
send/re
ceive i
s
also incre
a
s
e that lead
s to
increasing cl
usteri
ng
energy dissi
pation.
The simul
a
tion results
illustrate the
i-ECBR
sc
he
me
signifi
cantly redu
ce
s the cl
usteri
ng ove
r
head
s a
s
co
mpare to existing sch
e
me
s. Unli
ke existing
scheme
s
,
i-
EC
BR
divided
the whol
e sensor area
i
n
to
unifo
rm sized
no
n-ove
r
lappi
ng network
partition
s b
a
sed o
n
n
ode
s
distrib
u
tion,
whi
c
h
re
sult
s in g
r
ou
ping
relatively cl
oser n
ode
s i
n
to
a
particula
r cl
uster. Fu
rth
e
rmo
r
e, the
numbe
r
o
f
c
l
u
s
ters in
c
r
ea
se
s
/
d
e
c
r
e
as
es
w
h
en
increa
se/de
c
rease in
net
work si
ze. In
a
ddition, th
e
e
l
ection
proce
s
s is initiate
s within
bo
und
ed
regio
n
that le
ads to
redu
ce com
m
uni
ca
tion expendi
t
u
re a
nd en
ergy con
s
um
ption. Rathe
r
th
an
perfo
rm re
-cl
u
steri
ng of whole network
field on per
i
o
dic ba
si
s, the role
of CHs
are dyna
mica
lly
rotated am
on
g membe
r
no
des.
4.2. Av
erage End-to
-End
Dela
y
Figure 2
illustrates averag
e en
d-to
-en
d
del
ay
amon
g differe
nt cl
uster ba
se
d
energy
efficient
sche
mes agai
nst
different n
u
m
ber of n
ode
s.
It is
see
n
th
at end
-to-end
delay
of all t
h
e
scheme
s
in
creased by increme
n
ting th
e numbe
r of
node
s. This i
s
due to, hig
her the net
work
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 14, No. 2, June 20
16 : 531 – 53
7
536
size incurs a
dditional n
e
twork ove
r
he
ads a
nd
con
gestio
n
duri
n
g data forwa
r
ding.
However,
prop
osed
i-E
C
BR
a
c
hieve
d
lowe
r average en
d-to
-e
nd delay
as
comp
ared to existing rel
a
vent
scheme
s
. Existing
schem
es exhi
bit a highe
r en
d-to-end
delay
in data forwarding
pro
c
e
ss
becau
se of constructin
g
n
on-o
p
timized
routi
ng
path
whi
c
h lea
d
s t
o
more ro
ute
bea
kage
s a
n
d
freque
nt re-t
ransmi
ssion
s
.
Figure 1. Clu
s
terin
g
overh
ead
s v/s num
ber of
node
s
Figure 2. Average e
n
d
-
to-e
nd delay v/s numbe
r
of node
s
4.3. Net
w
o
r
k
Throughpu
t
Network th
ro
ughp
ut mea
n
s
h
o
w ma
ny
data p
a
cket
s
have b
een
su
ccessfully
re
ceived at
receive en
d. In singl
e ho
p, node
s di
re
ctly send
th
eir
aggregate
d
d
a
ta to sin
k
n
ode a
nd du
e
to
distan
ce fa
ct
or they exce
ssively
con
s
u
m
ed en
ergy
level, whi
c
h result
s
in mo
r
e
pa
ck
et
s lo
se.
Figure 3 illustrates the n
e
t
work thro
ug
hput as a fu
nction
of
net
work
size.
It
is
ob
se
rved
that
i-ECBR
sch
e
m
e re
markly
improve
d
through
put a
s
co
mpa
r
e to ex
isting a
pproa
che
s
. Ba
sical
l
y,
existing
clu
s
ter b
a
sed e
n
e
rgy effici
ent
routin
g
sch
e
m
es determi
ne
con
s
tru
c
t sho
r
test routi
n
g
path fo
r d
a
ta
tran
smi
ssi
on
ba
sed
on
m
i
nimum
num
ber of h
o
p
s
.
Due
to thi
s
, t
hose
scheme
s
highly decrease the achi
evable
data
delivery
per
f
o
rmance because of
low
flexibility against
exhau
sted no
des, which re
sults in
lo
we
r routing p
e
rfo
r
mance.
Figure 3. Net
w
ork throu
g
h
put v/s numb
e
r of node
s
50
100
150
200
250
300
0
0.
0
1
0.
0
2
0.
0
3
0.
0
4
0.
0
5
0.
0
6
0.
0
7
0.
0
8
0.
0
9
N
u
m
ber o
f
N
odes
C
l
us
t
e
ri
ng O
v
er
hea
ds
(
j
)
i-
E
C
B
R
L
E
A
CH-
D
T
K
-
m
e
ans
C
l
us
t
e
r
i
ng
50
10
0
15
0
20
0
25
0
30
0
0.
0
1
0.
0
2
0.
0
3
0.
0
4
0.
0
5
0.
0
6
0.
0
7
0.
0
8
0.
0
9
N
u
m
b
er
of
N
o
des
A
v
er
a
ge E
nd-
t
o
-
E
nd Del
a
y
(
s
e
c
)
i
-
EC
BR
L
EAC
H
-
D
T
K
-
m
e
a
n
s
C
l
us
t
e
r
i
ng
50
10
0
15
0
20
0
250
300
50
55
60
65
70
75
80
85
90
95
10
0
N
u
m
ber o
f
N
o
des
Net
w
ork
T
h
rou
ghp
ut
(
k
b
i
t
s
/
s
e
c
)
i
-
EC
BR
LE
A
C
H
-
DT
K
-
m
ean
s
C
l
us
t
e
ri
n
g
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Im
proved En
erg
y
Awa
r
e Cluster b
a
sed
Data Routing
Schem
e for
WSN (K
halid
Ha
see
b
)
537
5. Conclusio
n
To imp
r
ove
the e
n
e
r
gy
co
nse
r
vation
while
reliabl
e d
a
ta delive
r
y a
r
e th
e m
a
in
rese
arch
aim in WS
N. The
i-E
C
BR
divides the
entire
sen
s
o
r
filed
into ba
lance si
zed
non-overl
appi
ng
clu
s
ters ba
se
d on node
s d
i
stributio
n. In addition,
the electio
n
overhead
s are minimize
d as th
ey
are
executed
insid
e
the
cl
uster bo
und
a
r
y. Mo
re
over,
optimized ro
uting
path
s
a
r
e con
s
tructe
d
towards
de
stination in te
rms of multi
-
facet m
e
thod
thereby im
proving ro
uting
perfo
rman
ce
. In
orde
r to bal
a
n
ce th
e load
distrib
u
tion a
nd imp
r
ove
n
e
twork lifetim
e, the po
sitio
n
s of fo
cal po
ints
are rotated
o
n
dem
and ba
sis with
l
east comp
utati
ona
l overh
ead
s.
At the end,
si
mulation
re
su
lts
are
discu
s
se
d and
proven
that
i-ECB
R
schem
e out
perfo
rms in t
e
rm
s of
clu
s
tering
overhe
ads,
netwo
rk th
ro
u
ghput a
nd en
d-to-end
dela
y
as compa
r
e
d
to existing
work. In future wo
rk,
we
will
further imp
r
o
v
e the p
e
rfo
r
mance of
i-
EC
BR
sche
me
by in
corpo
r
ating h
e
trog
e
nou
se
netwo
rk
cha
r
a
c
t
e
ri
st
ic
s.
Referen
ces
[1]
Potdar V, S
har
if A, Chang E.
W
i
reless s
ens
o
r
netw
o
rks: A s
u
rvey.
Adv
anc
ed Inform
ation
Net
w
orki
ng
and Ap
plic
atio
ns W
o
rkshops,
2009 W
A
INA'
0
9
Internatio
na
l Confer
ence
on
. 2009.
[2]
Yick J, Muk
her
jee
B, Ghos
al
D. W
i
reless
se
nsor
net
w
o
rk
s
u
rve
y
.
C
o
mput
er n
e
tw
orks
. 2
008;
52(
12):
229
2-23
30.
[3]
Che
n
Z
,
Xiao
Y, Li X, Li R. A Clusteri
ng P
r
otocol for W
i
reless Se
nsor
Net
w
orks Bas
ed on E
nerg
y
Potentia
l F
i
eld.
T
he Scientific
W
o
rld Jour
nal
. 201
3;
201
3.
[4]
Heinz
e
lm
an W
R
, Cha
ndrak
a
s
an A, Bal
a
kri
s
hna
n H.
En
e
r
gy-efficie
n
t co
mmu
n
icati
on
p
r
otocol for
w
i
reless micro
s
ensor n
e
tw
orks
. Sy
stem S
c
ienc
es, 200
0
Proceed
i
ngs
of the 33rd A
nnu
al Ha
w
a
i
i
Internatio
na
l C
onfere
n
ce o
n
, IEEE. 2000.
[5]
Heinz
e
lm
an W
B
, Chan
drak
as
an AP, Ba
lakri
s
hna
n H. An
a
pplic
atio
n-sp
ec
ific
pr
otoc
ol arc
h
itecture
for
w
i
reless microsensor net
w
or
ks.
Wireless Communic
a
tions
,
IEEE
Transactions on
. 2
0
0
2
; 1(4): 660-
670.
[6]
Sharma M, Sharma K.
An e
n
e
rgy effici
ent e
x
tende
d LEA
C
H (EEE LEAC
H)
. Commun
i
c
a
tion S
y
stems
and N
e
t
w
ork Techn
o
lo
gi
es (C
SNT
)
, 2012 In
ternati
ona
l Co
nfer
enc
e on, IEEE. 2012.
[7]
Park GY, Kim H, Jeon
g HW
, Youn
HY.
A
N
o
vel
C
l
uster H
ead
S
e
lecti
on Method bas
ed on
K-Me
ans
Algorit
h
m
for Energy Efficie
n
t W
i
reless Sensor N
e
tw
ork
. Advanced In
formation N
e
tw
o
r
kin
g
an
d
Appl
icatio
ns Worksho
p
s (WAINA), 2013 2
7
th
Internation
a
l C
onfere
n
ce o
n
, IEEE. 2013.
[8]
Youn
is O, F
a
h
m
y
S.
HEED:
a h
y
b
r
id,
en
er
g
y
-
e
fficie
n
t, di
stributed
clust
e
rin
g
a
ppr
oac
h for
ad
hoc
sensor n
e
t
w
ork
s
.
Mobile Computing, IEEE Tr
ansactions on
. 200
4; 3(4): 366
-379.
[9]
Yu-Quan
Z
,
L
e
i W
.
A N
e
w
Routi
ng Pr
otoc
ol for Effici
ent
an
d Sec
u
re
W
i
reless S
ens
or Net
w
o
r
ks.
T
E
LKOMNIKA Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
n
g
.
2013; 1
1
(11):
679
4-68
01.
[10]
Ye M, Li C, C
h
en G, W
u
J.
EECS: an e
ner
g
y
efficient cl
usterin
g
sche
m
e i
n
w
i
reless s
e
n
s
or netw
o
rks
.
Performanc
e, Computi
ng,
and C
o
mmu
ni
cations
C
onfe
r
ence, 2
005
IPCCC 20
05
24th IEEE
Internatio
na
l, IEEE. 2005.
[11]
Kim KT
, Lyu
CH, Moo
n
SS,
Youn
HY.
T
r
ee-b
a
sed
clust
e
rin
g
(T
BC) fo
r ener
gy effici
ent w
i
reless
sensor n
e
tw
orks.
Advanced
Information N
e
t
w
ork
i
n
g
and
A
ppl
icatio
ns W
o
rkshops (W
AINA), 2010
IEEE 24th Internatio
nal C
onfer
ence o
n
, IEEE. 2010.
[12]
Kang
SH, N
g
u
y
en
T
.
Distance
base
d
thr
e
sho
l
ds
for
cluster head select
io
n i
n
w
i
r
e
less se
nsor
net
w
o
rks.
Comm
unications Letters, IEEE
. 2012; 16(9): 1
396
-139
9.
[13]
Naru
eph
ip
hat W
,
Charnsripi
n
y
o C. An En
erg
y
-a
w
a
re C
l
u
sterin
g T
e
chniq
ue for W
i
re
less Sens
or
Net
w
orks.
InT
e
ch. 2010.
[14]
Ali SA, Sev
g
i
C.
Energy
Lo
a
d
Bal
anci
ng f
o
r F
i
xed C
l
uste
ring
in W
i
re
les
s
Sensor
Net
w
orks.
Ne
w
T
e
chnolog
ies, Mobil
i
t
y
and
S
e
curit
y
(
N
T
M
S),
2012 5th Inter
natio
nal C
onfer
ence o
n
, IEEE. 2012.
[15]
Issariy
a
kul T,
Hossain E.
An introd
uction to
net
w
o
rk sim
u
la
tor NS2. Sprin
ger. 201
2.
Evaluation Warning : The document was created with Spire.PDF for Python.