TELKOM
NIKA
, Vol. 13, No. 4, Dece
mb
er 201
5, pp. 1214
~1
224
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v13i4.1895
1214
Re
cei
v
ed Se
ptem
ber 30, 2015; Revi
se
d No
vem
ber
3, 2015; Acce
pted No
vem
b
er 18, 201
5
Routing Algorithm Based on Area Division
Management of Node in Wireless Sen
s
or Networks
Gao Wei
1,*
, Song Yan
2
, Shuping Fan
1
1
Institute of
T
e
chno
log
y
, Mu
d
anji
a
n
g
Norma
l
Un
iversit
y
, Mu
dan
jia
ng
157
01
1, Heil
ong
jia
ng
, China
2
Departme
n
t of Scientific Res
earch, Mud
anj
i
ang N
o
rma
l U
n
iversit
y
, Mu
da
n
jian
g
15
70
11, Heil
on
gji
a
n
g
,
Chin
a
e-mail: 0
453
ga
o@mai
l
.com
A
b
st
r
a
ct
In order to r
e
duce th
e co
mmu
n
ic
ation
ov
erhe
ad
amon
g
sensor
no
des
, a routin
g a
l
g
o
rith
m is
prop
osed
bas
ed on
z
o
ni
ng
ma
na
ge
me
nt nod
es. T
he al
gor
ith
m
d
e
fin
e
s
the calcu
l
ati
on
meth
od of
the
netw
o
rk p
a
rtition r
adi
us
afte
r no
des
de
pl
o
y
me
nt, an
d
di
vides
mon
i
tore
d ar
ea
accor
d
ing
to th
e r
adi
us
me
anw
hi
le l
a
y
outs on
e man
a
g
e
m
e
n
t nod
e i
n
eac
h partiti
o
n
. T
hen n
odes
’ communic
a
tio
n
cost is calc
ul
ated
base
d
on the
d
i
stance a
m
on
g
nodes
as w
e
ll as nod
es
’
e
ner
gy, and fin
i
she
s
the selecti
on
of routing
nod
e
s
base
d
on the c
o
st. F
i
nally, usi
ng the Matl
ab
simulati
on e
n
vi
ron
m
e
n
t, the para
m
eters i
m
p
a
cting the
opti
m
a
l
partitio
n
rad
i
us
are disc
usse
d
,
and the
prop
osed r
outin
g al
gorith
m
is c
o
mpare
d
w
i
th exi
s
ting al
gor
ith
m
s.
T
heoretic
al
an
alysis
an
d ex
p
e
ri
ment
al r
e
sul
t
s show
that
t
he prop
ose
d
alg
o
rith
m is more bal
anc
ed on
nod
es
e
nergy consu
m
ption. T
he
al
gor
ith
m
reduc
es n
e
tw
ork traffic over
h
ead w
h
i
l
e
exte
nds the
lifeti
m
e of
the netw
o
rk.
Ke
y
w
ords
: W
i
reless Se
nsor
Netw
ork, Mana
ge
me
nt Nod
e
s,
Routin
g Alg
o
ri
thm, Co
mmu
n
i
c
ation Overh
e
a
d
,
Partition
Copy
right
©
2015 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
Wirel
e
ss Se
n
s
or Netwo
r
ks (WS
N) i
s
ge
nerally
m
ade
up of tho
u
sa
nds
of self-o
rgani
zed
sen
s
o
r
n
ode
s whi
c
h
are very small. As se
nsor
nod
e
ca
n al
so
se
nse
and
p
r
o
c
ess info
rmati
on
together
with
commu
nicating with oth
e
r no
de
s, people d
e
si
gn
wirel
e
ss se
nso
r
net
work to
monitor vari
o
u
s
situatio
ns or ev
ent
s, transmitting
th
e no
de
data
to termin
al
u
s
ers.
Ho
wev
e
r,
althoug
h the
r
e is limited
strength
of se
nso
r
n
ode
in
pra
c
tical a
p
p
licatio
n an
d
WSN will
sa
ve
stren
g
th
as m
u
ch
a
s
po
ssi
ble by
ap
plying g
r
e
a
t
net
work
top
o
logy and be
st
routi
ng strate
gy.
But
becau
se of the differen
c
e
between tra
d
itional net
work a
nd Ad h
o
c net
work, route in WSN are
paid mo
re an
d more attenti
on [1, 2].
Dep
endin
g
o
n
the event
-driven n
a
ture
of WSN, th
e main fu
ncti
on of sen
s
o
r
node i
n
different ap
pl
ication
s
i
s
to forward th
e perce
pt
ion
data from t
a
rget
zon
e
to ba
se
station.
Therefore, d
e
sig
n
ing effi
cient routing
algorit
hm t
o
build hig
h
quality path
is one of
most
importa
nt issues i
n
WSN [3, 4]. For
the inhe
re
nt
feature
s
of
WSN, routi
ng
in WSN is
a
chall
engin
g
p
r
oble
m
. In re
cent ye
ars, schol
ars h
o
me
and
ab
roa
d
prop
osed th
at there
a
r
e m
a
inly
two routing
a
l
gorithm
s [5,
6, 7]: sin
g
le
path rout
ing
algorith
m
a
n
d
multipath
routing
algo
rithm.
The fo
rme
r
i
s
e
a
sy to
be
reali
z
e
d
a
n
d
go
od in
e
x
pansi
b
ility, but ha
s
high
network e
n
e
rgy
con
s
um
ption,
and thus it cannot meet th
e requi
re
m
e
n
t
s of reso
urce
-co
n
st
rain
ed
WSN. Multipa
t
h
routing
algo
rithm is a te
chn
o
logy that ca
n be u
s
ed to
sub
s
titute ro
u
t
ing techn
o
lo
gy. Throug
h the
paths it ch
oo
se
s, the data
can be tra
n
s
mitted fr
om
the sou
r
ce to the target. Comp
ared wi
th
singl
e path routing, low
delays,
load
balanci
ng a
nd high lin
k quality of
multipath rou
t
ing
techn
o
logy h
a
s
signifi
can
t
ly improved
WSN
p
e
rfo
r
man
c
e
and
attracted
m
o
re a
nd mo
re
attention.
For the WS
N of multipath routing, a ne
w ki
n
d
of mul
t
i-hop routing
algorithm b
a
s
ed o
n
partition man
ageme
n
t nod
e is improve
d
in order to furthe
r save the node e
nergy and impro
v
e
the routin
g ef
ficien
cy of the Internet. Th
e algo
ri
thm raise
s
the
defi
n
ition of "Part
i
tion radi
us"
a
nd
provide
s
co
mputational method
of
th
e
partition
ra
dius a
nd di
scu
s
ses the
radiu
s
value
after
netwo
rk sen
s
or node ra
ndom
de
ploy
ment.
To
re
alize the a
r
ea divisio
n
of netwo
rk,
then
distrib
u
te a
p
l
ace
d
ma
nag
ement n
ode
i
n
ea
ch
pa
rtition, an
d p
o
int
out the
division p
r
o
c
e
s
s
of
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Routin
g Algorithm
Based on Area Di
vi
si
on Mana
gem
ent of Node i
n
Wirel
e
ss …
(Gao Wei
)
1215
netwo
rk p
a
rtition, in the end give routin
g pro
c
e
ss of the node. Through the the
o
retical analy
s
i
s
and sim
u
latio
n
experime
n
t, the result shows t
hat the netwo
rk ex
pan
sibility of the propo
se
d
routing
algo
rithm is g
ood,
while
red
u
ci
n
g
ene
rgy con
s
umptio
n of n
ode an
d bal
a
n
cin
g
nod
e lo
ad,
lifetime of network is d
e
lay
ed.
2. Rese
arch
Metho
d
A lot of ro
uting poli
c
y i
s
raised i
n
recent y
ears, in
ord
e
r to
pro
v
ide the u
s
e
ratio of
netwo
rk
ene
rgy. Apply the traditional ro
uting al
go
rith
m CR [3]. Wh
en nod
e dete
c
ts the eve
n
ts, it
will broad
ca
st
them to all the node
s (th
e
se node
s
a
r
e
called nei
ghb
o
r
ing n
ode
s)
within the ran
g
e
of telecommunication. The
neighbor
i
ng nodes will
al
so
br
oadcast the in
formation to
other nodes.
The process
will be going
until
th
e event reaches base
station
(BS),
whi
c
h result
s in over
con
s
um
ption
of node st
ren
g
th in netwo
rk and a
d
d
s
th
e confli
ct in wirele
ss tran
smissi
on. In order
to improve the utilization of netwo
rk st
rength, many routing st
rate
gies are put forward in
recent
years [4
-12].
LEACH [4], whose exec
ution process is perio
dical, i
s
the typical routing alg
o
rit
h
m
in WSN. Every circul
atio
n con
s
i
s
ts o
f
stage
of cluster
esta
bli
s
hme
n
t and
stage of d
a
ta
telecom
m
uni
cation. The
routing algo
rithm
divide
s
n
e
twork
into many clu
s
ters
in
net
wo
rk
by
rand
omly cre
a
ting cl
uste
r
head, from
which it
se
nd
s, re
ceive
s
an
d
com
b
ine
s
d
a
t
a to red
u
ce t
h
e
stren
g
th co
nsumption of co
mmon no
de
s. But in
the proce
s
s of clu
s
ter hea
d directly transmittin
g
data, LEACH algo
rithm
d
oes not
co
n
s
ide
r
the
a
c
t
ual p
h
ysi
c
al
distan
ce
fro
m
nod
e to
b
a
se
station, whi
c
h lead
s to n
ode con
s
um
ption’s
rapi
d
increa
sing al
ong with
big
ger di
stan
ce
from
base
station.
The
r
efo
r
e, t
h
is
algo
rithm
doe
s
not fit
in
WSN’
s l
a
rge
-
scale
e
n
vironm
ent.
The
schola
r
s late
r come
up
with the
imp
r
oved
algo
rithm like LEA
C
H-TE [5] a
nd LEA
C
H-CE[6]
agains
t the dis
advantages
of LEACH. In c
h
oos
i
ng c
l
us
ter head,
the
former c
o
ns
iders
remained
stren
g
th of n
ode an
d average
stren
g
th
in the wh
ole
netwo
rk. T
h
e
latter intro
duc
es
th
e
co
nce
p
t
of ba
ckup
cl
uster he
ad.
Both of the
algorithm
s calcul
ate the
best
clu
s
ter
num
bers fro
m
the
prin
ciple
of le
ast st
rength.
Com
pared wi
th LEACH,
th
ese t
w
o alg
o
rithm
s
are b
e
tter. Eight pe
o
p
le
inclu
d
ing
Wa
ng Kun
c
hi
[7
] and
Du
an
Wenfa
ng [8]
com
e
up
wit
h
the
routin
g
algo
rithm
ba
sed
on“m
i
nim
u
m
hop
”
.
The basic id
ea of this algo
rith
m
is to establish “m
inim
u
m
hop”
and tim
e
ly add
new no
de
s
by
se
ndin
g
m
e
ssag
e
s wi
th ea
ch
othe
r am
ong
the
nod
es.
The
two
algo
rith
m
s
optim
ize the
pro
c
e
s
s of
da
ta tran
sm
issi
on a
nd
use A
C
K a
s
well
a
s
m
e
ssage
ca
c
he
me
ch
ani
sm.
In addition,th
e document [8] puts forwa
r
d Hell
o pa
cket jumping m
e
ch
ani
sm an
d provide
s
th
e
maintena
nce
pro
c
e
s
s of
route.
RDS
R [9] puts fo
rward a
nothe
r kin
d
of
ro
uting id
ea,
whi
c
h
divides the n
e
twork into several area
s based on th
e
relative dire
ction of transferri
ng data. Then
the node
will
rand
omly sel
e
ct rel
a
y nod
es in all
are
a
s
a
c
cordi
ng t
o
the rel
a
tive positio
n dista
n
ce
from ba
se
sta
t
ion. The n
o
d
e
ro
ute proce
ss
of u
s
ing th
is alg
o
ri
thm i
s
ea
sy, but
when the
divid
ed
area
s a
r
e rel
a
tively large in netwo
rk, ro
uting circ
le
wi
ll be cre
a
ted i
n
netwo
rk b
e
c
au
se of no
d
e
’s
rand
om sel
e
ction. Do
cum
ent
[10]
a
nd
[11] put fo
rward
ro
uting
p
a
th to find
n
ode
by creati
ng
routing tabl
e for sho
r
tcomi
ngs of RDSR. Docu
m
ent[1
2] put forward ERIDSR al
gorithm, whi
c
h
introdu
ce
s “minimum ho
p” an
d add
s the factors
of telecomm
unication di
stance. In ord
e
r t
o
further
save t
he strength o
f
node an
d improve the
routing efficie
n
cy of network, a kin
d
of n
e
w
routing al
go
rithm is create
d
base
d
on are
a
division ma
nagem
ent of node.
3. Routing
Algorithm Bas
e
d on Are
a
Div
i
sion Managemen
t
of Node
Similar to LEACH al
gorith
m
, the execut
ion proc
e
s
s o
f
routing alg
o
r
ithm is p
e
rio
d
ical, but
every cycle i
s
divided into three
stage
s: se
tting net
wo
rk, dividing a
r
ea and tra
n
sf
erri
ng data.
3.1. Producti
on of Ne
t
w
o
r
k Topolog
y
It is assume
d that 200 sensor n
ode
s in net
wo
rk
are rand
omly arra
nge
d at the two-
dimen
s
ion
a
l
area
with th
e
area
of 20
0m
*200m. Ba
se
station i
s
lo
ca
ted on th
e left side
bel
ow t
h
e
monitori
ng area. The
places
will
not be changed af
ter a
ll nodes
are arranged.
As Figure
1 is
sho
w
n
the n
e
t
work top
o
log
y
, the hori
z
o
n
a
l axis
and
vertical
axis are u
s
ed to
ma
rk th
e g
r
ap
hical
positio
n of no
de (a
bsci
ssa
and o
r
din
a
te). Commo
n se
nso
r
no
de i
s
see
n
a
s
ci
rcl
e
sig
n
“
○
” in t
he
figure. Ba
se
station is l
o
cat
ed at the
origi
n
of
two
-
dim
e
nsio
nal
coo
r
d
i
nate a
s
the t
r
iangle
sig
n
“
△
”
in Figure 1.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 121
4 – 1224
1216
Figure 1. Structure of n
e
twork a
r
rang
em
ent.
3.2. Area Div
i
sion Proces
s of Ne
t
w
o
r
k
3.2.1. Calcul
ation of Sub
a
rea Semi-Diameter
It is assum
e
d
that sen
s
o
r
n
ode in
network
can dyn
a
mi
cally sen
s
e th
e rem
a
ine
d
st
rength
and po
sition i
t
self as
well a
s
cal
c
ul
ate the dista
n
ce from ba
se
station or
other
n
ode
s by usi
n
g I-
ERIDSRalgo
rithm. The i
subarea
se
mi-diamete
r in
n
e
twork i
s
Ri
(1
,
)
it
t
i
Z
, of whic
h tt
rep
r
e
s
ent
s th
e total n
u
mb
er
of suba
rea
se
mi-di
a
met
e
r. Th
e
cal
c
u
l
ation formula
is
as Equ
a
tion
(1).
11
22
21
1
,1
((
1
)
/
)
,
1
i
cd
i
R
i
ii
c
d
R
i
(
1
)
In Equation
(1), d
1
i
s
the t
e
lecommu
nication threshol
d and
it is
se
t that all nod
es in
the
rang
e of d1a
way from n
o
de i
(1
)
in
are the
neighb
ori
n
g
node
s. The
n is the n
u
m
ber of
comm
on
sen
s
or no
de in
netwo
rk. In f
o
rm n
e
tw
o
r
k the unifie
d
para
m
eter c1
and
c2
are
two
importa
nt on
es i
n
establi
s
hing
su
barea
structu
r
e
by
setting
different value
s
i
n
practi
cal
use
to
flexibly chan
ge the divi
sio
n
of network
area. Pa
ram
e
ter
c1i
s
u
s
e
d
to de
cide t
he si
ze
of se
mi-
diamete
r
R1 i
n
the first sub
a
rea, which indire
ctly
affects the si
ze o
f
semi-dia
met
e
r in other a
r
eas
and finally
wil
l
affect the a
r
ea in all
su
ba
rea
s
. Parame
ter c2 affect
s the se
mi-di
a
meter valu
e
o
f
other
sub
a
re
as ex
cept for
R1, which
ma
ke
s the value
s
of pa
ramete
r c1
and
c2in
R1a
s
0.2 a
n
d
3.
The valu
e of
Rican
be
se
e
n
in
Figu
re
2. Value
of min
i
mum
sub
a
re
a semi-diame
ter in
netwo
rk
depe
nd
s on the furthe
st no
de from ba
se
station in net
work.
We
will further study the function
of parameter
c2. It i
s
not ha
rd to
see from Figure 2
and
cal
c
ulatio
n formula
(1) th
at the whol
e se
nso
r
net
work arra
nge
ment
is in the first
quad
rant. Th
e
sub
a
re
a
semi
-diam
e
ter i
s
a qu
arte
r of t
he
circle
sem
i
-diam
e
ter.
When i
>
2
and
a,
m
X
and
m
Y
and
rep
r
e
s
ent the
scope
of m
onitorin
g
a
r
e
a
. It can
be
kno
w
n f
r
om f
o
rmul
a (2) th
at the value
of
para
m
eter c2
dire
ctly affect
s the
si
ze
of
other area
s e
x
cept fo
r
se
ction 1.
The
area S
of i
sub
a
re
a
is:
22
21
21
1
22
22
2
1
21
21
1
1
22
2
2
22
2
1
2
21
2
22
2
2
1(
1
)
(
1
)
(2
)
4
1
(
1)
(
1
)
(
1)
(2
(
0
.
2
3
)
)
4
1
(
1)
(
1
)
(
1)
(0.
4
6
)
4
i
i
k
i
k
ii
Sc
d
c
d
R
ii
ii
k
cd
cd
d
d
ii
k
ii
k
cd
c
ii
k
(2)
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Routin
g Algorithm
Based on Area Di
vi
si
on Mana
gem
ent of Node i
n
Wirel
e
ss …
(Gao Wei
)
1217
Figure 2. Top
o
logy stru
ctu
r
e of network
dividing area
3.2.2. Div
i
sio
n
of Ne
t
w
o
r
k
Area
Base station,
whi
c
h will
be regarded as
c
enter i
n
network,
will
calcul
ate the
network
sub
a
re
a semi
-diam
e
ter. Rii
s
the se
mi-di
a
meter
t
o
p
r
o
duce ma
ny virtual
circle
s. In order to ma
ke
it convenient
for instruct
ion, it is
shown i
n
the
dotted li
ne of
Figure 2
that
base station will
divi
de
netwo
rk i
n
to several fan shape a
r
ea
s
according
to
these
circle
s. After dividing areas, b
a
se
station
will broadcast information
to
node of arrangement area.
Wh
en node i
s
received, it
will
cal
c
ulate
the
are
a
n
u
mbe
r
itself
acco
rding to
th
e d
i
stan
ce from
base
st
ation.
The
process of
netwo
rk dividi
ng are
a
and
calcul
ation pro
c
e
ss of no
de
area n
u
mb
er
are a
s
followed:
(1)
The broad
ca
sting informa
t
ion packet
HELLO
(IDi
,(N(i
)
.xd,N(i
)
.yd),d1
)
of nod
e i in netwo
rk
inclu
d
e
s
the
identifier IDi with n
ode i,
coo
r
dinate(N(i).xd
,
N(i).yd) of
node i and
telecom
m
uni
cation
thre
sh
old d
1
, After
anothe
r n
ode
j
re
ceive
s
pa
ckets,
it will calcul
ate
th
e
absolute val
u
e
ij
d
accordi
ng t
o
the
dista
n
ce from
no
de i.
If
1
ij
dd
, node i
will
rega
rd
no
de
j as neig
hbo
ri
ng one a
nd st
ore it in routin
g table.
(2)
Base
station
reg
a
rds R1
as tel
e
comm
unication
se
mi-diam
e
ter
and
se
nd
s t
he p
a
cket
Messag
e (n
+1,(N(n
+
1).xd,
N(n
+
1
)
.yd ),Ri), i
dentifier with base
station in message, ba
se
station
co
ordi
nate a
nd
cal
c
ulation fo
rmul
a of
sub
a
re
a
semi
-diam
e
te
r. After no
de
i re
ceiv
e
s
messag
e in
area
1, it
will
cal
c
ul
ate th
e ab
sol
u
te v
a
lue
of di
sta
n
ce
from
ba
se statio
n a
n
d
store
Ri acco
rding to ba
se
station coo
r
d
i
nat
e. Later n
ode i will se
n
d
data packet
Message
1
(i,n+1,
(
N(n
+
1
)
.xd,N(n
+
1
)
.yd ),d1) to the
neigh
bori
ng n
ode.
(3)
After neighb
o
r
ing no
de j in
node i is received, it
will calcul
ate the a
b
sol
u
te value
of distance
from ba
se
st
ation a
c
cordi
ng to the me
ssage,
ch
a
n
g
e
the nod
e id
entifier i to j i
n
Messa
ge1
and later
will forward to other neig
hbo
rin
g
node
s.
(4)
Rep
eat step (3) until all no
des h
a
ve ca
l
c
ulated the di
stance from ba
se statio
n.
(5)
Base statio
n
compa
r
e
s
with the area semi
-diam
e
te
r acco
rding t
o
the distan
ce absol
ute
value from th
e furthe
st no
de. If
1
_
tt
RM
A
X
D
i
s
R
base st
ation will gai
n the total num
ber t
of divided are
a
in netwo
rk.
(6)
Node i will cal
c
ulate the area number N(i
)
.Z it
self according to the abs
olute value
of distance
from ba
se sta
t
ion. The cal
c
ulation form
ul
a is
：
1?
1
If
1
iB
S
dR
, then N(i).Z
=1;
2?
2
Otherwise,if
1
,(
2
)
ii
B
S
i
Rd
R
i
,then N(i).Z
=i.
The n
e
two
r
k
stru
cture is shown in Fi
gu
re
2
acco
rdi
n
g to the p
r
o
c
edures ab
ove after
netwo
rk
divid
e
s a
r
ea
s, in
whi
c
h net
wo
rk is divid
ed i
n
to seve
ral f
an shape
are
a
s. The
nea
rest
sub
a
re
a from
base
station
is area Z1.
As the area
of this part
is rel
a
tively small, it is not
indicated i
n
Figure 2.
Zi i
n
the fig
u
re
repre
s
e
n
ts th
e i
sub
a
re
a
(
2,
it
t
i
Z
). R1——
R5
in
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 121
4 – 1224
1218
hori
z
ontal
axi
s
represent
s subarea
semi-diam
e
te
r.
Base
station will arrange a managem
ent
node in
every
area
after di
viding area
s. As nod
e “*
”is
sho
w
n in fig
u
r
e, this
kind
o
f
node po
sitio
n
in area i
s
fixed and the st
re
ngth is out of limit.
3.3. Routing
Process o
f
Node
In order to fu
rther imp
r
ove
the
routin
g e
ffi
ciency
of n
ode, th
e al
g
o
rithm
(recorded
as I-
ERIDSR) fully consi
ders the su
barea
numbe
r an
d
stren
g
th of node, dista
n
ce from nod
e
to
manag
eme
n
t node
or b
a
se station i
n
selectin
g ro
uting no
de, whi
c
h n
ode i
will
cho
o
se from
all
neigh
bori
ng node
s
j
(1
,
)
jn
j
i
accordin
g to the size of telecom
m
uni
cat
i
on co
st. Th
e
telecom
m
uni
cation
co
st N(j).co
s
t of nod
e j is defined
as
：
22
(
)
/
(
).
,
(
).
1
()
.
()
.
c
o
s
22
(
)
/
(
).
,
(
).
1
sqrt
d
d
N
j
re
N
j
z
ij
j
N
i
h
e
a
d
Nj
t
sqrt
d
d
N
j
r
e
N
j
z
ij
j
B
S
(
3
)
It is not h
a
rd
to find from
formul
a (3) th
at
the big
ger
the st
rength
of node
j, the
nea
rer
distan
ce from
manag
eme
n
t node
or bi
gg
er p
r
ob
ability to forwa
r
d
da
ta as the
next node. In o
r
d
e
r
to ensure the effectiveness of
algorithm, the nodes nea
r base
station
will directly send data to
the station. T
he pro
c
e
s
s of node
(1
)
ii
n
in sele
cting ro
ute is
con
c
lu
ded b
e
l
ow:
The 1
-
4 line
of the algo
rithm above i
s
an init
ial p
r
o
c
ess of pa
ram
e
ter, of whi
c
h
MNnu
m
rep
r
e
s
ent
s the numb
e
rs of
manag
emen
t node in n
e
twor
k. As the
definition of i
dentifier in b
a
se
-----
-
-
-
1?
1
f
o
r i
=
1
:
1:(
n
+M
Nnu
m
+
1
)
2?
2
N(
i)
.
f
l
ag_m
u
lt
i
H
o
p
=
1
;
3?
3
N(i
)
.n
e
x
t
h
op
==
0
4?
4
e
nd f
o
r
5?5
fo
r
i
=
1
:
1:
n
6?6
i
f
(N
(i
).
Z=
=
1
)
7?
7
N(
i)
.next
h
op
=
n
+
1
;
8?
8
N(
i).
f
la
g_m
ultiH
o
p
=
0;
9?
9
e
nd if
10
?10
i
f
(
N
(
i
)
.
f
l
a
g_m
ultiHop
=
=
1
&
&
N(
i)
.ne
i
bNum
>
0
)
11
?11
f
o
r
j=
1:1:N(
i)
.
n
e
i
b
N
um
12
?12
COM
P
UT
E
N(tem
p
He
ads
(
j)).c
o
s
t
13
?
1
3
end
14
?14
i
f
(N(t
e
m
p
H
e
a
d
s
(j)).
Z<
N
(
i
)
.
Z
)
15
?
1
5
i
f
(N(t
e
m
p
H
e
a
d
s
(j
)).
co
s
t
==
m
i
n
c
o
s
t
)
16
?16
N(i)
.
n
ex
tho
p
=
t
em
pHe
ads
(
j
)
;
17
?
1
7
e
n
d
i
f
18
?
1
8
e
l
se
i
f
(N(t
e
m
p
H
e
a
d
s
(
j
)).
Z
=
=N
(i
).
Z&&
N
(
i
)
.
n
e
x
t
h
o
p
=
=
0
)
19
?19
i
f
(N(t
e
m
p
H
e
a
d
s
(j)).
co
st
=
=
m
i
n
c
o
s
t
1
)
20
?
2
0
N(
i
)
.
n
ex
thop
=
t
em
p
H
e
ads
(j
)
;
21
?
2
1
e
n
d
i
f
22
?
2
2
els
e
if
(
N
(
t
em
p
H
ea
ds
(
j
)
)
.
Z
>
N(i)
.Z
&
&
N(
i)
.next
hop
=
=
0)
23
?
2
3
i
f
(N(t
e
m
p
H
e
a
d
s
(j
)).
c
o
st
==
m
i
n
c
o
s
t
2
)
24
?
2
4
N(i)
.
n
ex
tho
p
=
t
em
pHe
ads
(
j
)
;
25
?
2
5
e
n
d
i
f
26
?
2
6
end
if
27
?
2
7
end
if
28
?
2
8
e
nd f
o
r
29
?
2
9
N(
n+
2)
. ne
x
t
S
t
o
p
=
n
+
1
;
30
?
3
0
N(
n+
2)
. f
l
a
g_m
ultiH
o
p
=
0;
31
?
3
1
f
o
r i
=
3
:
1
:
(M
Nnu
m
+
1
)
32
?
3
2
i
f
(N(n
+
i
).f
l
a
g
_
m
u
l
ti
Ho
p==
1
)
33
?
3
3
N(
n+
i)
.n
e
x
tSt
o
p
=
n
+
i-
1;
34
?
3
4
e
nd if
35
?
3
5
en
d f
o
r
-----
-
-
-
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Routin
g Algorithm
Based on Area Di
vi
si
on Mana
gem
ent of Node i
n
Wirel
e
ss …
(Gao Wei
)
1219
station i
s
n
+
1, the identifi
e
r of ma
nag
ement
no
de
will begi
n fro
m
n+2.
N(i
)
. flag_multiHop
is
use
d
to in
dicate if no
de i
a
pplie
s
way of
multi-jumpi
n
g to
trans
f
er data. Initiati
ve value of 1
refers
to use
way of
multi-jumpi
n
g and
N(i
)
.ne
x
thop is u
s
e
d
to store th
e i
dentifier of n
e
x
t jumping ro
ute
in node i with
initiative value of 0; the5-9
line
of the algorithm i
s
used to judge t
he are
a
num
ber
size of
nod
e
(1
)
ii
n
.If are
a
n
u
mbe
r
e
qual
s
1, n
ode i
do
es n
o
t nee
d to
forward
data
by othe
r
node
s but to dire
ctly send
data to
base station. At this time, the
multi-jumpi
ng identifier vari
a
b
le
of node i will be set as 0; the 10-27 line
is use
d
to
use way of multi-jumpi
ng to send data of n
ode
i routing p
r
o
c
e
ss, of whi
c
h the 10
-1
3line is
u
s
e
d
to calculat
e all teleco
mmuni
cation
cost
N(tem
p
Head
s(j
)).
cost
of n
e
ighb
orin
g n
o
de in
nod
e i.
The 1
4
-27 lin
e u
s
e
s
multi-j
u
mping
ro
utin
g
strategy: at first node i
will
choose the m
i
nimum
neighboring node j
as th
e next j
u
mping route in
all neig
hbo
rin
g
nod
es wh
ere area n
u
mb
er i
s
the
smal
lest. If there i
s
no
such ro
u
t
ing nod
e in t
h
e
14-1
7
line
of
the algo
rithm,
node i
will
re
spe
c
tively sel
e
ct a
c
cording
to are
a
nu
m
ber
with e
q
u
a
l
and bi
gge
r
seque
nce fro
m
the n
e
ighb
oring
no
de in
co
rrespondi
n
g
area
s b
a
se
d on th
e p
r
in
cipl
e
of minimum t
e
lecommu
nication co
st wh
ich
can b
e
se
en at 18
-26 li
ne; the 29
-30
line is u
s
ed t
o
define ro
uting
node of man
ageme
n
t nod
e in sectio
n 1
as base stati
on and set the multi-jumpi
n
g
identifier of
manag
eme
n
t node a
s
0; the 31
-35 li
n
e
is used to d
e
fine the ro
uting nod
e of o
t
her
manag
eme
n
t nod
es excep
t
for
sectio
n
1. The
next j
u
mping
in thi
s
kind
of n
o
d
e
is sm
alle
r t
han
the manag
e
m
ent node of
area num
be
r. It is not
hard to see fro
m
the routing
node sel
e
cti
on
algorith
m
defi
ned in thi
s
text that node will pr
efer to
sele
ct the ne
are
r
area nei
ghbo
ring
nod
e
from b
a
se
st
ation a
s
relay
node. In
this way, it
will
redu
ce the
te
lecom
m
uni
ca
tion co
st of
n
ode
and save the stren
g
th of no
de, whi
c
h is
cruci
a
l to wirel
e
ss se
nsor wi
th limited stre
ngth.
4.Simulation Experiment
4.1.Experiment Env
i
ron
m
ent and Pa
ramete
r Setti
ng
Imitate opera
t
ing environ
ment of network in
the M
a
tlab simul
a
tion platform b
y
setting
experim
ent p
a
ram
e
ter. It i
s
d
e
fined
the
be
st calcula
t
ion form
ula
of su
barea
semi-di
a
mete
r in
routing
algori
thm(I-ERI
D
S
R
) of thi
s
text accordi
ng t
o
the experi
ment result.
We
will also
make
comp
arative analysi
s
bet
ween I-E
R
IDS
R
alg
o
rithm
a
nd RDSR, E
R
IDSRalgo
rit
h
m. It is assu
med
in expe
rimen
t
that there
are
200
co
mmon
se
n
s
o
r
no
de
rand
omly scattered in the
two
-
dimen
s
ion
a
l area of 20
0m
*200 m in wh
ich the initial stren
g
th of node is 0.5
J
a
nd si
ze of da
ta
p
a
c
k
e
t is
40
0b
its
.
The exp
e
rim
ent value
of t
e
lecommu
nication th
reshol
d d1
is 30m.
Part of oth
e
r
variabl
e
value in expe
riment is
referred to do
cum
ent [13].
4.2. Experiment Result a
nd Theor
y
Analy
s
is
4.2.1.Cer
t
ain
t
y
of Parameter in Subare
a Semi-Diam
e
ter Fo
rmula
In ord
e
r to
p
r
operly
divide
the net
work
a
r
ea
and
re
du
ce tel
e
comm
unication
con
s
umptio
n
of node
s to t
he up
pe
r limit, it is stu
d
ied
the be
st
valu
e of pa
ram
e
ter
c1 a
nd
c2
in su
ba
rea
se
mi-
diamete
r
by
simulation
exp
e
rime
nt of
wh
ich th
e ex
pe
ri
ment result is se
en
at Figu
re 3.
Figu
re
3
provide
s
five
different valu
es
of 0.1,0.2,
0.4,0.
6,0.8 when
c2
=1 th
e
stren
g
th
con
s
umption t
r
en
d
of
all node
s in every turn of network. The
experim
ent
indicate
s that whe
n
c1i
s
valued 0.1 and
0.2,
the stre
ngth
con
s
um
ption
and wave
of node in e
v
ery turn are relatively small. The no
de
con
s
um
ption
of every turn is balan
ced
althoug
h whe
n
c1is valu
ed
0.1, the node con
s
um
ptio
n
stren
g
th
com
b
ination i
n
ev
ery turn
is th
e sm
alle
s
t. But in this co
n
d
ition, the
su
bare
a
(se
c
tion
1)
near the
ba
se statio
n i
s
t
he
smalle
st
and th
e
n
o
d
e
num
be
r in
the area i
s
t
he lo
we
st. T
hese
node
s
whi
c
h
are
nea
r
ba
se statio
n h
a
ve mo
re
bu
rde
n
an
d
will di
e
ea
rly be
cau
s
e of
run
n
ing
out
of strength. T
h
is will ma
ke
the “blind si
de” [14] in
this su
barea. In orde
r to better solve the
“hot
spot”
[15] problem nea
r the
ba
se
sta
t
ion, the b
e
st value of
c1 is 0.2 i
n
t
he exp
e
rim
e
nt
environ
ment
of routing alg
o
rithm provid
ed in this text.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 121
4 – 1224
1220
Figure 3. Net
w
ork con
s
um
ption wh
en p
a
ram
e
ter c1h
a
s differe
nt value
s
Figure 4 p
r
ovides fo
ur
different val
ues
of c
2
=
1
,2,3,4 when c
1
=
0
.2 the s
t
rength
con
s
um
ption
trend of all node
s in every turn of
network. The ex
perim
ent re
sult indicate
s that
whe
n
c2i
s
va
lued bet
wee
n
1 and 3, th
e wh
ole con
s
umptio
n of
node
will g
r
a
dually re
du
ce
in
every turn
of
netwo
rk. T
h
is is be
ca
use
when
c2i
s
valu
ed too
small
(c2
=
1 o
r
2
)
a
n
d divided
are
a
s
in netwo
rk
are relatively many, the area
of
small num
ber
will be lo
wer
and the
node
s are fe
wer
too. But these no
des ofte
n ne
ed to fo
rward the
dat
a of big
g
e
r
a
r
ea
numb
e
r
whi
c
h a
r
e
se
nt by
node. A
s
a
re
sult, the n
ode
con
s
u
m
ption
will be
more
and
will die
e
a
rly. The
nod
e co
nsumptio
n
of all area
s a
r
e not bal
an
ced, whi
c
h will
make
the
whole pe
rform
ance of network
go do
wn.
It
also
can b
e
seen from Fi
g
u
re 4 that wh
en c2 i
s
valu
ed more like
4, the netwo
rk co
nsumptio
n is
increa
sing. T
h
is is be
ca
use when
c2i
s
valued mo
re, the suba
rea
numbe
r in ne
twork is too little
and n
u
mb
er
of mana
geme
n
t nod
e is not
much a
nd th
e no
des of all
are
a
s except
se
ction
1
se
nd
data, it n
eed
s n
o
t ma
ny
manag
eme
n
t nod
es to
send, whi
c
h
will re
sult
in
the
fo
rwardi
ng
circulatio
n a
nd mo
re
con
s
umptio
n in
netwo
rk.
Th
e
r
efore, the
b
e
st valu
e of
c2
is 3 i
n
t
he
experim
ent e
n
vironm
ent of routing alg
o
ri
thm provide
d
in this text.
Figure 4. Net
w
ork con
s
um
ption wh
en p
a
ram
e
ter c1 has different values
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Routin
g Algorithm
Based on Area Di
vi
si
on Mana
gem
ent of Node i
n
Wirel
e
ss …
(Gao Wei
)
1221
4.2.2. Comparison of Str
e
ngth
Cons
u
m
ption
In Figu
re
5
we ma
ke
co
m
pari
s
on
of th
e no
de
co
nsumption i
n
n
e
twork wh
en
the three
algorith
m
s
of
RDS
R, ERIDSR an
d I-E
R
IDSR
ope
rate
i(i=1,2,…
1
0
)
turn
s. Simulati
on expe
rime
n
t
has be
en do
ne for 10 turns and the n
ode in every turn tran
sfer
only one dat
a. It is not ha
rd to
see
from
Fig
u
re
5 th
at the
co
nsumptio
n
of
RDS
R
al
g
o
rithm i
s
the
bigge
st
with t
he in
crea
sing
of
turns
and th
e co
rrespon
ding
stren
g
th
curve i
s
n
o
t
smooth. T
he no
de
strength oth
e
r
two
algorith
m
s ke
eps ste
ady g
r
owth
and th
e
con
s
um
pti
on
of routin
g al
g
o
rithm I-E
R
IDSR p
r
ovided
i
n
this text is the sm
allest. T
h
is i
s
because in
the
routi
ng p
r
o
c
ess
of usin
g RDS
R
algo
rithm no
de
will ra
ndomly
sele
ct nod
e
as relay no
de
s in all
sub
a
reas
acco
rdin
g to are
a
nu
mbers
whi
c
h
can
not ensure the optimizatio
n of
node. T
h
is will m
a
ke
some
node
s transfe
r data
in long di
sta
n
ce
and
con
s
u
m
e
big n
ode
strength. In a
d
d
i
tion, if
node
arrang
ement
in RDSR
alg
o
rithm i
s
in t
h
e
subarea with bigger ar
ea number and far away from m
anagem
ent node, node will produce
routing
ci
rcle
in this a
r
ea
which
will ma
ke the nod
e ru
n out of stren
g
th and di
e e
a
rly. Thu
s
, using
the whole n
e
twork of RDSR alg
o
rit
h
m w
ill have big algori
t
hm and imbalan
ce
d no
de
con
s
um
ption.
The
r
e
are t
h
ree
fa
ctors
among
jum
p
i
ng n
u
mb
er,
area
n
u
mbe
r
and
n
ode
in
the
pro
c
e
s
s of
choo
sing
rel
a
y node
in E
R
IDSR
alg
o
ri
thm.
Com
pared with RDS
R
al
gorith
m
, the
whol
e con
s
u
m
ption of ERIDSR algo
rith
m will be lo
wer; and I-ERI
D
SR alg
o
rith
m divides are
a
of
netwo
rk th
ro
u
gh expe
rime
n
t
to define th
e be
st ca
l
c
ul
ation form
ula
of sub
a
re
a
semi-di
a
mete
r in
orde
r to
reali
z
e th
e
goal
of re
du
cing
telecommu
nication
con
s
u
m
ption of
nod
e
.
More
over, t
h
is
algorith
m
con
s
ide
r
s the a
r
ea numb
e
r a
nd teleco
mm
unication co
st in the proce
ss of nod
e ro
ute.
The cal
c
ulati
on
of
tele
co
m
m
unication cost con
s
id
er
s
t
w
o f
a
ct
or
s
o
f
st
ren
g
t
h
a
n
d
di
st
an
ce,
w
h
ich
make
s st
ron
ger directio
n
to destination node in
time of send
ing node d
a
ta. As a result,
comp
ared wit
h
RDS
R and
ERIDSR alg
o
r
ithm, the wh
ole con
s
u
m
pt
ion of I-ERIDSR algorith
m
is
lower.
Figure 5. Co
mpari
s
o
n
of Network Co
n
s
umptio
n
4.2.3. Comparison of
Netw
o
r
k Life
In ord
e
r to te
st the
su
rviving time
of ne
twor
k,
we m
a
ke
com
p
a
r
iso
n
of n
ode
de
ath after
usin
g three al
gorithm
s of n
e
twork to
ope
rate 5
0
turn
s.
As I-ERI
D
S
R
alg
o
rithm
h
a
s n
o
t app
ea
red
death n
ode i
n
the first 2
5
turns, th
e co
mpari
s
o
n
of
node
death
n
u
mbe
r
in thre
e algo
rithm
s
will
begin fro
m
the twenty-sixt
h turn. Experiment re
sult
is se
en at Fi
gure 6, from
whi
c
h it can
be
see
n
that after network o
perate
s
26 tu
rns, we will u
s
e RDSR alg
o
rithm with m
o
re than 50 n
ode
death numb
e
r
s.
T
he node
death numb
e
r
of
E
R
IDSR algor
ith
m
is a
bout 3
0
, but i
n
the
sam
e
t
u
rn
the nod
e d
e
a
t
h numb
e
r
of
I-ERIDS
R
alg
o
rithm
wh
i
c
h is
obvio
usly smaller
th
an RDSR,
E
R
IDS
R
algorith
m
. Wi
th the g
r
o
w
th
of ea
ch
al
gorithm’
s
op
erating
turn,
the no
de
de
ath nu
mbe
r
of I-
ERIDSR
algo
rithm in
network in
cre
a
ses slo
w
e
r
t
han
RDS
R, ERIDSR alg
o
rithm.
This is
be
ca
use
the co
nsu
m
p
t
ion of I-ERIDSR al
gorith
m
is
mo
re b
a
lan
c
ed a
n
d
obviously le
ss th
an RDS
R
,
Tur
n
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No
. 4, Decem
b
e
r
2015 : 121
4 – 1224
1222
ERIDSR al
go
rithm in the con
d
ition that
three algo
rit
h
ms have th
e same
com
m
on expe
rim
ent
para
m
eter. T
herefo
r
e, the
death node
numbe
r of this
algo
rithm in netwo
rk i
s
less with lon
ger
surviving time
of network.
Figure 6. Co
mpari
s
o
n
of death nod
e nu
mber
4.2.4. Comp
arison of
Netw
o
r
k Extens
ion
In Figure 7
we will u
s
e the node
numbe
r of three al
go
rith
ms in netwo
rk when
n=1
00,20
0,…
1000 a
nd ma
ke
comp
ari
s
o
n
of netwo
rk
stren
g
th con
s
umption. Th
e experim
ent h
a
s
been
don
e ten time
s. Each
algo
rith
m ope
rate
s
10 turns i
n
every expe
ri
ment. We
m
a
ke
comp
ari
s
o
n
o
f
the g
ene
ral
stren
g
th
co
nsumed
in te
n t
u
rn
s. It i
s
n
o
t
hard
to
se
e from Fig
u
re 7
t
he
whol
e network strength
co
nsum
ption of
three al
gor
it
hms ten
d
to
increa
se
with
the gro
w
th
of
node
num
bers. And in
the
con
d
ition that
node
num
be
rs
are
the
sa
me in n
e
two
r
k, the
con
s
u
m
ed
stren
g
th of I
-
ERIDS
R
alg
o
rithm i
s
ob
vious
ly lo
wer than RDSR, ERIDSR al
gorithm,
whi
c
h
indicates
th
at
I-ERIDSR al
gorithm network ha
s
go
od
extensio
n. This al
gorithm
fits for WS
N of
large
-
scale n
ode.
Figure 7. Net
w
ork st
ren
g
th
con
s
umptio
n
trend after n
ode
s are a
d
d
e
d
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Routin
g Algorithm
Based on Area Di
vi
si
on Mana
gem
ent of Node i
n
Wirel
e
ss …
(Gao Wei
)
1223
4.2.5. Comp
arison of Lo
ad Balan
c
e
In order to te
st the
nod
e l
oad
bala
n
ce
of th
re
e al
gorithms i
n
n
e
twork,
we
will
compa
r
e
the stre
ngth
varian
ce of n
ode in net
wo
rk in the
con
d
ition that three algo
rithm
s
have the sa
me
comm
on exp
e
rime
nt pa
ra
meter. Th
e a
pproxim
ate va
lue of va
ria
n
ce
is
ro
und
ed to fou
r
d
e
c
ima
l
places. Every algorithm e
x
perime
n
t ha
s be
en do
ne
50 turn
s a
n
d
we
analyze
the varian
ce
of
gene
ral
stren
g
th of netwo
rk con
s
umptio
n on eve
r
y
turn of e
a
ch algorith
m
na
mely in 1-1
0
,11-
20,…41
-
50 tu
rn. Experime
n
t result is se
en at T
able 1
and we
can
see from the
data in Tabl
e
1
by usi
n
g
RDSR al
gorith
m
the
co
nsume
d
st
ren
g
th
of
node
in
every
network tu
rn
varie
s
be
cau
s
e
of ra
ndom
selectio
n of
routing
nod
e. As
a
re
su
lt
, varian
ce
value i
s
l
a
rge
r
while
ERI
D
SR
algorith
m
has strong
er di
re
ction in the p
r
oc
ess of no
de se
nding d
a
ta to manag
ement nod
e or
databa
se. Th
erefo
r
e, com
pare
d
with
RDSR al
go
rith
m, the con
s
u
m
ed st
ren
g
th
in ea
ch turn
of
netwo
rk n
o
d
e
will
be
rel
a
tively balan
ced
with
sm
all vari
ance. Co
mpa
r
ed
with the
first two
algorith
m
s, varian
ce of I-ERIDSR al
gori
t
hm has li
ttle fluctuation. T
h
is is be
ca
use in the process
of using n
ode
routing alg
o
rithm in this text,
the remai
ned strength
of routing no
de is con
s
ide
r
e
d
whi
c
h will bal
ance node lo
ad and ma
ke
s nod
e stre
n
g
th
con
s
um
ption in every turn of network
more stable.
Table 1. Co
m
pari
s
on of st
rength varia
n
ce in three alg
o
rithm
s
1-10
11-20
21-30
31-40
41-50
RDSR
0.3323
0.3533
0.0218
0.0313
0.0199
ERIDSR
0.1061
0.0022
0.0164
0.0002
0.0006
I-ERIDSR
0.0008
0.0019
0.0000
0.0000
0.0001
5. Conclusio
n
For th
e p
r
obl
em ab
out big
stren
g
th
con
s
umpti
on
of ro
uting alg
o
rith
m, we
com
e
up with
a
kind
of routin
g alg
o
rithm I
-
ERIDSR ba
sed o
n
a
r
ea
d
i
vision m
ana
gement
nod
e
.
This
algo
rit
h
m
con
s
i
s
ts of three
stag
es:
setting net
work, divi
din
g
are
a
and
transfe
rri
ng
data. I-ERIDS
R
algorithm will
divide network into
several
subareas
according to
subarea semi-di
a
meter. Later a
manag
eme
n
t node
will be p
r
odu
ce
d in ev
ery su
barea.
In the process of node
sen
d
i
ng data n
ode
stren
g
th an
d
telecomm
un
ication di
sta
n
ce
a
m
on
g node
s a
r
e consi
dered
which m
a
kes
the
algorith
m
in this text have good p
e
rfo
r
m
ance in t
he whole. At last we ma
ke
co
mpari
s
o
n
bet
wee
n
I-ERIDS
R
al
gorithm a
nd
existing routi
ng algo
ri
thm.
The re
sult i
ndicates the
node of routi
n
g
algorith
m
ha
s bala
n
ced l
oad, whi
c
h fit
s
for
WS
N of
large
-
scal
e
node. Thi
s
al
gorithm
red
u
c
e
s
node
stren
g
th
con
s
umptio
n
and extend
s t
he netwo
rk li
fe at the sam
e
time.
Ackn
o
w
l
e
dg
ements
This
wo
rk wa
s
sup
porte
d
by Hei
Lon
gji
ang n
a
tural scien
c
e fo
und
ation (QC201
3C0
67);
Tech
nolo
g
y
Found
ationof
Mu
Danjia
n
g
No
rmal
University (Q
G201
400
3);
Open
found
a
t
ion
proje
c
t of
Hei Lon
gjian
g
Intelligen
ce
edu
cation
a
nd info
rmatio
n engi
nee
rin
g
key l
abo
ra
tory
(SEIE2013-0
5
).
Referen
ces
[1]
Samra B, Mohammed B. A
Novel stren
g
th
Effici
ent and Lifetime Ma
xi
mizatio
n
Routi
ng Protoco
l
i
n
Wireless Sens
or Net
w
orks.
W
i
reless Pers
o
nal C
o
mmun
ic
ations
. 20
13; 7
2
(2): 133
3-1
3
4
9
.
[2]
Radi
M, Dezfo
u
li B, Ab
u B
a
k
a
r K & L
ee M.
Multip
ath ro
uti
ng i
n
w
i
r
e
l
e
ss
sensor
net
w
o
rk
s: Surve
y
a
nd
researc
h
chal
le
nges.
Jour
na
l of Sensors
. 20
12; 12(1): 6
50–
685.
[3]
Liu Y, Xio
ng, N, Z
hao Y, Vasilakos AT
, Ga
o J
& Jia Y. Multi-la
ye
r clus
tering ro
utin
g alg
o
rithm f
o
r
w
i
reless vehicular sens
or net
w
o
rks.
IET Com
m
u
nications
. 2
010; 4(7): 8
10–
816.
[4]
Marjan R, Be
hnam D, Kam
a
lrul
niz
a
m AB, et.al.
IM2PR: interferenc
e-
minimiz
ed mul
t
ipath routi
n
g
protoco
l
for
w
i
r
e
less se
nsor n
e
t
w
o
r
ks.
Wireless Network
. 2014; 20(
7): 180
7–1
82
3
[5]
Youn
is O, F
a
hm
y
S. He
ed:
A h
y
br
id, en
erg
y
-effici
ent, distrib
u
ted clu
s
tering a
ppr
oa
ch for ad-h
o
c
sensor n
e
t
w
ork
s
.
IEEE
Transactions on Mobile Com
p
uting
. 2
004; 3(4): 6
60-
669
Variance
A
lgorithm
Turn
Evaluation Warning : The document was created with Spire.PDF for Python.