Indonesian J
ournal of Ele
c
trical Engin
eering and
Computer Sci
e
nce
Vol. 1, No. 2,
February 20
1
6
, pp. 390 ~
398
DOI: 10.115
9
1
/ijeecs.v1.i2.pp39
0-3
9
8
390
Re
cei
v
ed Se
ptem
ber 21, 2015; Revi
se
d Jan
uary 8, 2016; Accept
ed Ja
nua
ry 2
8
, 2016
Prolonging the Lifetime of Wireless Sensor Networks
using LPA-star Search Algorithm
Ahm
e
d A.
Alkathm
a
w
e
e
*
1
, Lusong Fe
ng
2
, Imad S.
Alsha
w
i
3
1,2
School of computer scie
n
ce
and tech
nol
og
y,
Hu
azho
ng U
n
iversit
y
, W
u
h
an, Hub
e
i, Chi
n
a
3
Computer Sci
ence a
nd Infor
m
ation T
e
chno
lo
g
y
, Un
iversit
y
of Basra, Basra, IRAQ
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: ahmed
ad
el1
949
@gma
il.co
m*
1
, lusongfe
n
g
@h
ust.edu.cn
2
,
emad
alsh
a
w
i
@
gmai
l.com
3
A
b
st
r
a
ct
Since
sens
ors
hav
e l
i
m
ited
pow
er r
e
so
u
r
ces,
en
ergy consu
m
ption
has beco
m
e a
critic
a
l
chall
e
n
g
e
to W
i
reless
Se
nsor
Netw
orks (W
SNs). Most of t
h
e ro
uting
pr
oto
c
ols
prop
ose
d
to trans
mit
dat
a
packets thro
ug
h paths w
h
ic
h consu
m
e low
e
nergy a
i
m si
mply to red
u
ce b
a
ttery pow
er consu
m
ption. T
h
i
s
can l
ead t
o
l
ead to
netw
o
rk partitio
n
a
nd re
duce
ne
tw
ork lifetime.
T
herefor
e, to
bal
ance
en
e
r
gy
consu
m
ption
a
nd ext
end
n
e
tw
ork lif
eti
m
e
w
h
ile
mi
ni
mi
z
i
ng p
a
cket
del
i
v
ery de
lay; th
i
s
pap
er
prop
o
s
es a
new
ener
gy-ro
uting pr
otoco
l
usin
g the
life
l
o
ng pl
an
nin
g
A-
star (LPA-star) search al
gor
ithm. T
h
is a
l
g
o
ri
th
m
is used to fin
d
an opti
m
u
m
forw
arding p
a
th b
e
tw
een
the so
urce no
de a
nd
the sink. T
he o
p
timu
m path c
a
n
be sel
e
cted
de
pen
din
g
o
n
hi
ghest res
i
du
al
sensor
ener
gy
, the shortest d
i
stance to th
e
sink an
d l
o
w
e
st
traffic load. S
i
mu
lati
on res
u
lt
s ind
i
cate th
at the pr
opos
ed
protoco
l
i
n
cre
a
sed th
e l
i
feti
me
of the
net
w
o
rk
compar
ed w
i
th the A-star
routi
ng (EERP) prot
ocol.
Ke
y
w
ords
: LP
A-star algor
ith
m
, netw
o
rk lifet
ime,
routin
g, w
i
reless se
nsor n
e
tw
orks
Copy
right
©
2016 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Gro
w
ing
adv
ances in
wireless
com
m
unication a
n
d
ele
c
tro
n
ics
make th
e
Wirel
e
ss
Senso
r
Net
w
orks
(WSNs) have come
u
nder
inten
s
iv
e
rese
arch
o
v
er the
la
st years. it h
a
ve
the
enormou
s
a
pplication p
o
tentials in
defen
se
a
nd civilian related
ap
pli
c
ation
s
su
ch
as
environ
menta
l
monitori
ng,
military field
and bi
omedi
cal ob
se
rvation [1-3]. In
su
ch a
ppli
c
atio
ns,
WSN is a di
stributed wi
rel
e
ss network comp
ri
si
ng of
many numb
e
r of low-co
st devices
call
ed
sen
s
o
r
no
de
s, which a
r
e i
n
telligent e
n
o
ugh to
sen
s
e
,
calculate a
nd commu
nication with
ea
ch
other to
colle
ct and t
r
an
sm
it the data fro
m
the env
iron
ment and
pa
sse
s
it at a ce
ntric
node
kn
ow
as the ba
se
station (sin
k n
ode) [4].
The sen
s
or n
ode
s in WS
Ns have resou
r
ce
co
nstraint
s, inclu
d
ing
short commun
i
cation
rang
e, limited pro
c
e
ssi
ng
, small mem
o
ry, low ban
dwidth, and
limited energ
y
resou
r
ce [5].
Indeed, sen
s
or
b
a
tterie
s
powere
d
by finite
ene
rgy
pre
s
e
n
t difficultie
s in
bei
ng repla
c
ed
or
recha
r
ge
d. Therefo
r
e, ene
rgy resou
r
ce con
s
trai
ned a
nd scarcity is an acute ch
alleng
e in WSN
appli
c
ation
s
. On the other hand, tran
smissi
on an
d reception inf
o
rmatio
n sp
e
nd most en
e
r
gy
quickly du
rin
g
co
mmuni
cation an
d m
a
y be con
s
id
ered th
e maj
o
r
sou
r
ces
o
f
sen
s
o
r
en
e
r
gy
con
s
um
ption
[6]. In one
-ho
p
tra
n
smi
s
sio
n
, the
sen
s
o
r
nod
es lo
cate
d farth
e
r
awa
y
from the
si
nk
drain
high
er
energy due t
o
the long tra
n
smi
ssi
on d
a
ta ran
ge, and
may die out first. On the
other
hand, m
u
lti-h
op tra
n
smissi
on
which exp
end
s le
ss
po
wer at e
a
ch
hop i
s
m
o
re
energy-effici
e
n
t
than di
re
ct
transmissio
n
[7]. Ho
wev
e
r, multi-
hop
tran
smi
ssi
o
n
is u
s
ed
i
n
many
-to-o
n
e
comm
uni
cati
on, and in thi
s
communi
ca
tion pattern, t
he clo
s
e
r
sen
s
or to th
e sin
k
may ru
n ou
t of
energy soon
er d
ue to
ha
ving the
high
est lo
ad
of
p
a
ckets,
and
thus there i
s
Uneve
n
Ene
r
gy
Con
s
um
ption
(UEC) am
o
ng
sen
s
o
r
n
o
des [8].Fi
gure1
sho
w
s a
n
example
of
UEC proble
m
in
multi-ho
p tra
n
smi
ssi
on.Th
e se
nsors n
eare
s
t to
th
e sin
k
a
r
e o
v
erbu
rde
ned
becau
se of
the
expulsi
on of
other
se
nso
r
s data to the
sink. As
a con
s
eq
uen
ce, n
e
a
re
r sen
s
ors
con
s
um
e mu
ch
more
en
ergy
and
die m
u
ch
faste
r
, which
re
sult
s in
net
work
pa
rtition
pro
b
lem,
en
ergy h
o
le
s, a
n
d
incom
p
lete
p
a
cket d
e
livery
[8]. Therefore, UE
C
i
s
a
n
innate
dra
w
b
a
ck d
e
scri
be
d by a
ma
ny-one
traffic pattern and multi
-
hop
routin
g
which re
d
u
c
e
s
the lifetime of a wirele
ss n
e
twork
signifi
cantly.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Prolon
ging th
e Lifetim
e
of
Wirel
e
ss Sen
s
or
Networks usin
g… (Ahm
ed A. Alkathm
a
wee)
391
Figure 1. Une
v
en energy
consumption i
n
WSN multi
-
hop tran
smi
s
sion
The ro
uting
algorith
m
s in
WSNs are
desi
g
ne
d to prolo
n
g net
work lifetime
and to
enha
nce bala
n
ce
d ene
rgy con
s
um
ption
throug
h data
forwa
r
din
g
[9]. As seen i
n
Figure 2, the
routing
proto
c
ol
s sel
e
ct th
e best p
a
ths
for forwar
din
g
the data pa
cket from the
sou
r
ce nod
e
to
the si
nk.
Furt
herm
o
re, th
e
same
route
p
a
th is u
s
ed
continuo
usly t
hen th
e exi
s
ting n
ode
s
on t
hat
route path
will die, which results in network
partitioning. Therefore,
the bal
anci
ng energy
con
s
um
ption
of sen
s
o
r
n
o
des
with
an o
p
timizing
ro
uting meth
od i
s
a ne
w
chall
e
nge p
r
e
s
e
n
te
d
in many routi
ng proto
c
ol
s for WS
Ns.
Figure 2. Dea
t
h the centri
c
node
s in WS
N lead to Network pa
rtition
proble
m
In this
pap
er, a ne
w
ene
rgy efficie
n
t
routing
p
r
oto
c
ol,
calle
d t
he LPA-sta
r
routin
g
proto
c
ol, i
s
propo
sed. It
uses
an
optimal
agg
re
gat
ed
co
st fun
c
tion,
and
the
lifel
ong
plan
ning
A-
star (LPA-sta
r) search al
g
o
rithm to find
an
optimum
forwa
r
di
ng p
a
th betwe
en
the sou
r
ce n
o
d
e
and th
e d
e
sti
nation. T
he
L
PA-star alg
o
rithm is an
in
cre
m
ental
ve
rsio
n of
A-sta
r
al
gorithm
th
a
t
reu
s
e
s
information
from previou
s
sea
r
ch path
s
to find a ne
w routing path
without re-ru
n
nin
g
from the
sta
r
t nod
e. The
propo
sed
proto
c
ol ta
ke
s the
followin
g
pa
rameters i
n
to
con
s
id
eratio
n
to
sele
ct an
opti
m
um routing
path:
i) hi
ghe
st re
sid
ual
se
nso
r
n
ode
en
ergy; ii) l
o
we
st traffic lo
ad
;
and iii) minim
u
m hop count
s.
The remain
der of the
pa
per i
s
organi
zed
as follo
ws: Section
2
pre
s
ent
s relat
ed work
for the
routin
g algo
rithm
o
n
improving t
he maxi
mu
m
netwo
rk lifetime. The
pap
er int
r
odu
ce
s
and
provide
s
a
bri
e
f backg
rou
n
d
to the LPA-star al
go
rithm
in Section
3. Section 4
prese
n
ts the
L
PA-
star routin
g
p
r
otocol.
The
perfo
rman
ce
evolution
i
s
d
i
scusse
d in
Section
5. Fi
nally, Section
6
pre
s
ente
d
the
con
c
lu
sion to
this pape
r.
2. Relate
d Work
In the traditio
nal routing
al
gorithm
s, the
inte
re
st in
mi
nimizin
g
e
n
e
r
gy co
nsumed
witho
u
t
con
s
id
eratio
n
to bal
ance e
nergy
co
nsu
m
ption fo
r
th
e whole
network lea
d
to
a
netwo
r
k pa
rtition
probl
em and
hence
will reduce net
wo
rk lifetime. Therefore, the el
i
m
ination of
UEC in order t
o
prolo
ng the n
e
twork lifetim
e has b
e
com
e
a criti
c
al issue in WSNs.
In the re
ce
nt past, ma
ny
energy-effici
e
n
t ro
utin
g al
gorithm
s h
a
ve bee
n p
r
op
ose
d
to
improve and
extend
the n
e
twork
lif
etim
e of WS
Ns. In [10] an
d [1
1], the autho
rs
propo
se t
w
o
different ap
proache
s that
distrib
u
te the
traffi
c loa
d
throu
gh the
n
o
de
s which
have re
maini
ng
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 2, February 201
6 : 390 – 398
392
energy to en
sure the
stab
ility of sen
s
or node
s fo
r ex
tending
wirel
e
ss net
wo
rk l
i
fetime. In [12],
the authors p
r
opo
se a rout
ing method to redu
ce t
he
hop co
unt extent of a
routing path so a
s
to
redu
ce
the p
o
we
r
con
s
um
ption of e
nd
to end t
r
an
smissi
on. All o
f
the above
routing p
r
oto
c
ols
usin
g fixed routing p
a
ths
focu
s on
obt
aining
ene
rgy
effic
i
enc
y
. As
a
res
u
lt of
this
, trans
m
itt
i
ng
data via spe
c
i
f
ic routing p
a
ths might d
r
ai
n node
s po
wer qui
ckly.
In [13], WSNHA-GA
HR i
s
a gree
dy and
A-star
h
euri
s
tic routing
alg
o
rithm for
WSNs in
home autom
ation. The algorithm mini
mize
s the nu
mber of hop
s for data tran
smissio
n
by usin
g
the greedy fo
rwa
r
di
ng te
ch
nique. F
u
rt
he
rmore, the p
r
otocol
uses t
he A-star
se
a
r
ch
alg
o
rithm
to
overcome
th
e lo
cal mi
ni
mum p
r
obl
e
m
. The
WS
NHA
-GAHR
proto
c
ol
doe
s n
o
t con
s
i
der
remai
n
ing
en
ergy in
se
nso
r
no
de
s whe
n
sel
e
ct
in
g th
ese
nod
es fo
r ro
uting
data
packet
s
. Th
us,
this may depl
ete sele
cted
node
s en
erg
y
quickly. Re
na et al. [14] prop
osed an
efficient routi
n
g
algorith
m
whi
c
h
used
an
A-sta
r
sea
r
ch
algorith
m
to
pick o
u
t an
o
p
timum
routi
ng p
a
th from
the
sou
r
ce n
ode i
n
to the
sin
k
, whi
c
h
wa
s ba
sed
on
a no
d
e
’s mi
nimum
energy level
value to exte
nd
the lifetime
o
f
WSNs.Thi
s
routing
p
r
oto
c
ol
defin
e
d
t
he th
re
shold
as th
e mi
nim
u
m e
nergy le
ve
l
value se
nsor
node
s. Co
nseque
ntly, a node will b
e
ex
clud
ed from t
he routin
g op
eration
whe
n
the
energy value
of this n
ode i
s
le
ss than th
e thre
sh
old v
a
lue. Th
e aut
hors of [15]
p
r
opo
se
d OF
F
I
S
(optimi
z
ed fo
rwa
r
di
ng by fuzzy inferen
c
e sy
stems)
for flat sensor n
e
two
r
ks to extend the
lifetime of WSNs. T
h
is protocol
fa
v
our
s
crit
e
r
ia ju
st
as
sh
ort
e
st
p
a
t
h
wit
h
sm
al
l hop
s,
max
i
mum
remai
n
ing b
a
ttery powe
r
i
n
ord
e
r to se
lect the
be
st
can
d
idate n
o
de inthe forwardin
g
path
s
.
In
[16], the auth
o
rs
propo
sitio
n
is of
a
hybrid multi-h
op routing p
r
oto
c
ol so
as to
o
p
timize e
n
e
r
gy
con
s
um
ption
and to exte
nd the
WSNs lifetime by
combi
ned
h
i
era
r
chical a
nd flat multi-hop
routing
al
gori
t
hms
with
a
data
colle
ctio
n sch
e
me. In
[17], the
aut
hors
pre
s
e
n
ted a
n
o
vel
sl
eep
sched
uling
m
e
thod to
ma
ximize
netwo
rk lifetim
e
ca
lled Virtu
a
l B
a
ckbo
ne S
c
h
edulin
g (VBS
).
This u
s
e
s
on
ly backb
one
s to forward d
a
ta packet
s
, and the othe
r remai
n
ing n
ode
s halt their
radio
s
so as
to con
s
erve
and save en
ergy. Al G
haffari et al. [18] propo
se
d an
energy effici
ent
routing
(EE
R
P) p
r
oto
c
ol t
o
extend
WSNs lifet
ime
a
nd b
a
lan
c
e
e
nergy
con
s
u
m
ption. In th
eir
prop
osed m
e
thod, the rout
ing meth
od u
s
ed
an A
-
st
a
r
sea
r
ch al
gorithm to achie
v
e an o
p
timu
m
forwa
r
di
ng p
a
th betwe
en
sou
r
ce nod
e
and si
nk by
developin
g
p
a
ram
e
ters to cho
o
se the n
e
xt
hop in the ro
uting ope
rati
on. This ro
uting proto
c
ol
rerun
s
the se
arch algo
rith
m from the start
node
every round to
find
an optim
um
path with
out
usin
g info
rma
t
ion from th
e
previou
s
sea
r
ch
operation.
M
o
reove
r
,
thi
s
approa
ch do
es not
in
clu
d
e
a
w
ay of
ch
ecking
its
cu
rrent
path to
select
a new
path, whi
c
h will
result in
increased search
eff
o
rt. Hence, the inevitable
consequence is
highe
r co
nsu
m
ption of ene
rgy.
From the revi
ewe
d
literature, it can be conc
l
ude that a numbe
r of different metrics a
r
e
use
d
to avert netwo
rk p
a
rtition and
prolo
ng WS
Ns lifetime, su
ch a
s
bal
anci
ng ene
rgy
con
s
um
ption,
sho
r
te
st pat
h to the
sin
k
and b
a
lan
c
in
g load
traffic.
In this p
ape
r, we p
r
op
ose
a
new
ene
rgy-efficient ro
uting protoc
ol u
s
ing
LPA-sta
r
sea
r
ch alg
o
rithm. The LP
A-sta
r
algo
rit
h
m
repe
atedly uti
lize
s
a
simila
r optim
um forwarding
path
from
sou
r
ce
node to
si
nk.
It recalculat
es
the new o
p
timum path u
s
ing d
a
ta fro
m
its prev
io
us sea
r
ch re
sults. Th
e propo
sed p
r
oto
c
ol
sele
cts th
e o
p
timum routin
g path
depe
n
d
ing o
n
the
a
bove criteri
a
(resi
dual
ene
rgy sen
s
o
r
n
o
de,
distan
ce
to t
he
sin
k
of
a
node
an
d tra
ffic l
oad
in th
e no
de
queu
e). Fu
rthe
rm
ore, the
routi
ng
proto
c
ol
bala
n
ce
s
betwee
n
the
s
e
pa
ra
meters to
pr
o
t
ract th
e n
e
twork lifetim
e
with lowes
t
effort
and time as f
a
r as p
o
ssibl
e
.
3. Lifelong Planning A-star Search
Algorithm
Lifelong Pla
n
n
ing A-sta
r
(LPA-sta
r) i
s
an in
cr
e
m
ent
al version
of the A-star a
l
gorithm
introdu
ce
d in
200
1 by S
v
en an
d Ma
xim as a
co
mbination
of
the h
e
u
r
isti
c A-sta
r
sea
r
ch
algorith
m
, wh
ich
sp
eed
s
u
p
sho
r
test
st
atic p
a
th pl
a
nning
an
d
Dynamic SWSF
-FP in
creme
n
t
a
l
sea
r
ch algo
rit
h
m, which sp
eed
s up re
pla
nning [19, 20]
LPA-sta
r
em
ploys h
euri
s
ti
c kno
w
led
ge
to fo
cu
s the
search a
nd
sel
e
ct the o
p
tim
u
m path
from the
sta
r
t node
to the
goal n
ode
wh
ile the e
n
viro
nment dyn
a
m
i
cally chan
gin
g
over time [2
0].
Furthermore,
LPA-star
utilize
increm
ental techni
que to
reus
e inform
ation from
previous
sea
r
che
s
op
e
r
ation to find
a ne
w optimu
m
path
wi
tho
u
t rerunni
ng t
he alg
o
rithm f
r
om
scrat
c
h. In
fact, LPA-sta
r
is repeat
edl
y select
s sim
ilar opt
imu
m
path every time the nod
e
s
state remai
n
s
without ch
an
ges an
d alwa
ys efficiently recal
c
ul
at
es a
n
optimum pa
th from only those node
s that
have cha
nge
d to the goal node rather t
han re
com
p
u
t
ing the path from the start
node [19]. As a
result, LPA-star de
crea
se
s the nu
mbe
r
of sta
r
t no
des
whi
c
h n
eed to be
recom
puted
a
nd
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Prolon
ging th
e Lifetim
e
of
Wirel
e
ss Sen
s
or
Networks usin
g… (Ahm
ed A. Alkathm
a
wee)
393
decrea
s
e
s
th
e sea
r
ch effort for determining t
he n
e
w path to the goal. The
LPA-star
se
arch
algorithm uses the key val
ue (k(n))
with two co
m
ponents to sel
e
ct
the next node whi
c
h will
be
expand
ed ba
sed o
n
the followin
g
equati
ons:
))
(
2
),
(
1
(
)
(
n
k
n
k
n
k
(1)
)
(
))
(
),
(
min(
)
(
1
n
h
n
RHS
n
g
n
k
(2)
))
(
),
(
min(
)
(
2
n
RHS
n
g
n
k
(3)
Whe
r
e g
(n
)
and h
(n) va
lues
are th
e
actual
co
st p
a
th from the
start no
de to
node n
and t
h
e
estimated
co
st of the optimum path
from node
n to the target no
de
(d
estinatio
n no
de)
respe
c
tively. Also, the
Rig
h
t-Ha
nd
-Side
(RHS
(n
)) va
lue i
s
the
min
i
mum valu
e o
f
the path
costs
comp
uted
fro
m
the
nod
e’s neig
hbo
ring
node
n.
T
he
RHS
-value
is a
way fo
r
no
de that
chan
ged
its key value
to notify nei
ghbo
ring
no
d
e
s. the
com
pone
nts
of the
keys (k1
and
k2
)
can
be
thought of as identical to the f-value
s
a
nd g-valu
es u
t
ilized by the A-sta
r
algo
rithm re
spe
c
tively,
becau
se b
o
th
the g-val
u
e
s
and
RHS
-val
ues fo
r the
compon
ent of
keys
(k1 a
nd
k2
) in th
e LP
A-
star
algo
rith
m co
rrespon
d to the
g-val
ues u
s
ed i
n
t
he A-star alg
o
rithm
and th
e h-val
u
e
s
of
the
LPA-sta
r
alg
o
rithm
corre
s
pond
dire
ctly to the h-
val
ues i
n
the
A-sta
r
algo
rit
h
m [21-25]. As
con
s
e
que
ntly, the key(k(n
)
) evaluati
on f
unctio
n
s
can
be rep
r
e
s
e
n
ted as:
))
(
2
),
(
1
(
)
(
n
k
n
k
n
k
(
4
)
)
(
)
(
)
(
1
n
h
n
g
n
k
(5)
)
(
)
(
2
n
g
n
k
(6)
Gene
rally, LPA-star alg
o
rithm maintain
s a prio
rity queue to ke
ep
track of the node
s to
detect the
ne
xt node with
pick the
small
e
st key val
ue.
More
sp
ecifi
c
ally, the no
d
e
expan
ds i
n
the
prio
rity queu
e
whe
n
it ha
s
smalle
st
k1-v
alue a
nd if th
ere
are
many
node
s
with
same
smalle
st
k1
value, finally,
the node
sele
cted when
it has the
small
e
st k2
-value.
4. LPA-s
t
ar
Rou
t
ing Protocol
In
this
pap
er,
the WSN
to
pology
i
s
ab
stract
s
a
s
a
di
recte
d
gra
ph G
(N, D), wh
ere
N
is
the set of n
node
s a
nd
D is the
set of
dire
ct
radio
comm
uni
cati
on lin
ks bet
ween the
no
de
s.
A
dire
ct link d
=
(v, u), d
∈
D exists if the Euclide
an
distan
ce b
e
twee
n node v
and u insid
e
th
e
domain
of
r,
whe
r
e
r i
s
the
radi
us of the
tran
sm
i
ssi
on
ran
ge
of no
d
e
s. T
he Ba
se
station
(BS) i
s
respon
sibl
e f
o
r
sen
s
o
r
y d
a
t
a gathe
red
from all
othe
r n
ode
s in
the
n
e
twork
within
its tran
smi
s
si
o
n
rang
e [13], [26-27]. The p
r
oce
dure of finding an opt
im
um path from
the source n
ode to the BS is
rega
rd
to
so
me pa
ram
e
te
rs i
n
e
a
ch
se
nso
r
n
ode
like the
re
sidu
a
l
ene
rgy, traff
i
c lo
ad a
nd t
h
e
distan
ce to the sink. The B
S
must kno
w
the curr
ent le
vel for the criteria of each sensor no
de so
as to find the optimum path. Therefo
r
e, the BS
send queri
e
s to all sen
s
o
r
nod
es in the network
at the first ro
und an
d ea
ch sen
s
o
r
nod
e sen
d
s
a
b
o
v
e param
eter to the BS.
At the remai
n
ing
roun
d, only the sen
s
o
r
has
data to send
appe
nd
s it
s param
eters wit
h
the dat
a pa
cket toward the
BS. The BS
sho
u
ld
cal
c
ul
ate an
d b
r
o
a
d
ca
st a
n
o
p
timum
routin
g
sched
ule to
e
a
ch
sen
s
o
r
n
ode
[17].
The mod
e
l of the prop
osed
method a
s
su
mes
that WS
N ha
s the followin
g
pro
pert
i
es:
1.
All sensor n
o
des that rand
om
ly distrib
u
ted insi
de the
area.
2.
All se
nso
r
no
des have
the
same
initial
e
nergy
and
th
e same
maxi
mum tran
smi
ssi
on
rang
e.
3.
Each sen
s
o
r
node
kno
w
s the location of
BS as well a
s
its own and
its neigh
bors.
4.
Each
sen
s
o
r
node
ha
s the
different qu
a
n
tities of traffi
c loa
d
in
sid
e
its qu
eue
when
the traffic loa
d
ma
de
by the a
ppli
c
atio
n an
d the
n
o
de h
a
s al
rea
d
y co
mmitted
to
forward.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 2, February 201
6 : 390 – 398
394
The mai
n
g
oals
of this pape
r d
e
si
gn a n
e
w e
fficiency rout
ing
p
r
otocol
that
will
enha
ncement
and
extend t
he lifetime
of the WS
Ns t
h
rou
gh limitin
g ene
rgy
co
st as
well
as the
egalitari
an di
stributio
n of
power
con
s
u
m
ption wi
th
minimizi
ng d
a
ta pa
cket d
e
livery time. To
reali
z
e thi
s
, the ne
w m
e
th
od u
s
e
s
the L
PA-star
algo
ri
thm with all
a
f
oreme
n
tione
d ro
uting
crite
r
ia
(re
sid
ual ene
rgy, distan
ce
to the BS and traffi
c loa
d
amount
) for each se
nso
r
node to fin
d
an
optimum forwardin
g
path from the sou
r
ce node to
the
BS. The can
d
idate no
de to rep
r
e
s
ent the
optimum
routi
ng p
a
th de
pe
nds on
the la
rge
s
t ke
y-val
ue
whi
c
h u
s
e
d
the
evaluati
on fun
c
tion
(k1
(n)) to find th
e larg
est
k1
-value.The
nod
e ha
s t
he la
rgest
k2-val
ue
sele
cted
wh
en there is
m
o
re
than one n
o
d
e
with the sa
me larg
est k1
-value. The e
v
aluation fun
c
tion
s we u
s
e
d
are given a
s
:
))
(
2
),
(
1
(
)
(
n
k
n
k
n
k
(
7
)
))
(
/
1
(
)
(
)
(
1
n
h
n
g
n
k
(
8
)
)
(
)
(
2
n
g
n
k
(9)
Whe
r
e h
(n
) i
s
the di
stan
ce from n
ode
n to the BS a
nd g (n) i
s
the
co
st f
o
r nod
e
n, whi
c
h
take
s valu
e [
0
…1] a
nd
d
e
termin
e by
the fitnes
s fu
nction.
The f
i
tness eval
ua
tion fun
c
tion
is
con
s
id
ere
d
for the amou
nt of residu
al energy
and
the traffic loa
d
of node n to determin
e
the
optimum cost
value for the node n. The
cost functio
n
g
(n) we used i
s
given a
s
:
)
(
)
(
)
(
)
Re(
)
(
n
Bt
n
Bf
n
Et
n
n
g
(
1
0
)
Whe
r
ea
s
Re
(n) an
d Et(n
) a
r
e
refe
r to the initia
l and
re
sid
u
a
l ene
rgy
of nod
e n
respe
c
tively, and Bf(n) an
d Bt(n
) a
r
e
refer to
the
cu
rre
nt an
d m
a
ximum traffic
load
s of
nod
e n
r
e
spec
tively.
α
and
β
a
r
e consta
nt value
s
.
The candi
dat
e node th
at h
a
s the m
a
ximize po
we
r an
d lowe
st traffi
c load
amou
n
t
as well
as the shortest di
stance to t
he BS (mi
n
i
m
um hop
counts)
will be
selected as the better node for
the ro
uting p
a
th to the BS. As a
re
sult, these pa
ra
m
e
ters play a vit
a
l role
for b
a
l
ance the e
n
e
r
gy
con
s
um
ption
and the
app
ropriate
di
strib
u
tion of n
e
twork loa
d
for
a
ll node
s in
order to
prolog
the
netwo
rk lifeti
m
e.
Our
ro
uting p
r
otocol fo
r th
e propo
se
d
model
ch
eck
the existin
g
n
ode
s in th
e o
p
timum
path
with thei
r nei
ghbors
after
every packet send
s and re-use t
he
same path while its nodes still
have the
la
rg
est
key-val
u
e
.
Furthe
rmo
r
e, ou
r m
e
tho
d
get
s to
be
n
e
fit from th
e
previou
s
path
to
recalculate th
e new o
p
timu
m path from
curre
n
t node
that have ch
ange
d in whi
c
h key-valu
e to
the BS by select its neigh
b
o
rs t
hat have
the largest key-value with
out re-run
nin
g
from the start
node
s.
The p
r
op
ose
d
metho
d
d
epen
ded
on
the thre
sh
old value fo
r no
de e
nergy. The
comm
uni
cati
on between t
he se
nsor n
o
de and th
e BS through
ro
uting path
will brea
k while
the
sen
s
o
r
no
de
energy is le
a
s
t than the
th
reshold val
u
e
.
Therefo
r
e, t
he propo
se
d
algorith
m
trie
s to
find alte
rnate
ro
ute p
a
th
b
y
run
n
ing
the
algo
rithm
fro
m
the
sta
r
t n
ode to
avoi
d
cho
o
se the
l
o
w
energy node
s in orde
r to prolong the n
e
twork lifet
ime.
The flow
cha
r
t of propo
sed
method to find
optimum ro
uting path from
start nod
e to the BS is sho
w
n in Figu
re
3.
5. Performan
ce Ev
aluation
In this
se
ctio
n, the pe
rformance of the
pr
o
p
o
s
ed
al
gorithm i
s
in
vestigated
ag
ainst the
EERP protocol in [1
7] to d
e
mon
s
trate
o
p
timization
m
e
thod
s in
terms
of averag
e resid
ual
en
ergy
as well as ma
ximizing net
work lifetime.
5.1. Simulation Ev
aluation
We a
dopt M
A
TLAB for executin
g the
si
mulation.Th
e
simulatio
n
a
r
ea is
200
× 2
0
0
2
m
with 50
se
nsor n
ode
s ran
domly depl
oyed in thi
s
top
ogra
phi
cal a
r
ea. The i
n
itial ene
rgy for e
v
ery
sen
s
o
r
nod
e in the netwo
rk is set to 5 Jo
ules, with ma
ximum sen
s
e
d
transmissio
n rang
e at 80
m.
The traffic l
o
ad is
gen
erat
ed ra
ndo
mly betwe
en [0...10] for e
a
ch
sen
s
o
r
no
de.
There is
a si
ngle
base station
located at
the t
op right
-han
d co
rn
er of the
simulated area
(200
m, 200
m).
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Prolon
ging th
e Lifetim
e
of
Wirel
e
ss Sen
s
or
Networks usin
g… (Ahm
ed A. Alkathm
a
wee)
395
Furthe
rmo
r
e;
a radio mod
e
l discu
sse
d in [28]
is use
d
for the pro
posed metho
d
. Based on
this
model; th
e
se
nder no
de
ex
pend
s
ene
rgy
to run
ra
di
o
electroni
cs a
nd a
po
we
r
a
m
plifier. O
n
t
h
e
other h
and, the re
ceive
r
n
ode spen
ds
energy to run
the radio el
e
c
troni
cs. In orde
r to tran
smit
numbe
r
of bit
s
p
e
r
pa
cket (k)
over the
dista
n
ce (d) between
se
nder an
d
re
ceiver n
ode
s,
the
dissipate e
n
e
r
gy of a node
is ch
ara
c
te
rized by:
2
)
(
d
k
E
k
E
k
T
E
amp
elec
n
(
1
1
)
And the power co
nsumpti
on of a node
for re
ceivin
g this messa
ge is charact
e
rized by
the expre
ssi
o
n
s:
k
E
k
R
E
elec
n
)
(
(
1
2
)
Figure 3. Flow-cha
rt of the LPA-star
sea
r
ch al
go
rithm
The ele
c
troni
cs
ene
rgy Eel
e
c i
s
affected
by seve
ral factors such a
s
filtering, mod
u
lation,
digital codin
g
,
and
spreadi
ng of the
sig
nal. In addi
tio
n
, the amplifi
e
r e
nergy Ea
mp is i
n
fluen
ced
by facto
r
s like the
dista
n
ce d
bet
ween
sen
der
an
d receive
r
n
ode
s
a
nd acce
ptable bit-e
rro
r rate.
Simulation
s a
r
e m
ade
u
s
in
g value
s
100
pJ/bit/m
2
an
d
50 n
J
/bit for
Eelec
and
Ea
mp respe
c
tively.
The sim
u
latio
n
para
m
eters are presente
d
in Table 1
whi
c
h follo
ws.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 2, February 201
6 : 390 – 398
396
Table 1. Simulation Para
meters
Parameter Value
Topographical Area
200 × 200 m
2
Location of BS (meters)
(200 × 200
)
Number of
nodes
50
Transmission range (mete
rs)
80
Initial energ
y
of n
ode
5 J
Eamp
100 pJ/bit/ m
2
Eelec 50
nJ/bit
Data Packet size
4k bit
Number of
trans
mission packet
s
2 × 10
4
Maximum traffic
node’s
10
α
,
β
0.65,
0.35
5.2. Simulati
on Results
The n
u
mbe
r
of node
s
still l
i
ve is give
n a
s
a fu
nctio
n
o
f
roun
ds by u
s
ing th
e two
different
approa
che
s
p
r
esented in F
i
gure 4. The
grap
h indi
cat
e
s that the propo
s
ed meth
od outpe
rforms
the EERP protocol.
Whe
n
all pa
ckets a
r
e
sent, the
n
e
twork lifetim
e increa
se
d
with the LPA-star
algorith
m
by roughly 2
5
% o
v
er that obtai
ned fro
m
EERP
protocol.
Furthe
rmo
r
e, we can se
e
that
the numbe
r
of active nod
es of the LPA-sta
r
algo
rithm is co
nsi
s
t
ently higher t
h
an that of the
EERP proto
c
ol.
The different
duratio
n of time co
rre
sp
o
nding
to the first dea
d nod
e, compute
d
usin
g two
different a
p
p
r
oache
s is list
ed in
Tabl
e
2. Obviou
sly, the time
bef
ore th
e first
node
die
s
in
th
e
prop
osed met
hod is mu
ch
greate
r
than t
hat taken
by the first nod
e to die in the EERP method.
From Fig
u
re 4 and Tabl
e
2, it is clear t
hat the pro
p
o
s
ed met
hod
outperfo
r
m
s
the EERP
method in terms of maximi
zation of net
work
lifetime
and bal
a
n
c
in
g energy con
s
umptio
n.
Figure 4. Number of no
des still alive as a function of ro
unds based on the
two approaches
(LPA-s
t
ar and EERP)
Table 2. Nu
m
ber of Ro
und
s with the First Dea
d
No
de
Approaches
EERP
LPA-star
Round first node
dies
2743
5895
The simul
a
tio
n
results in Figure 5 sho
w
the avera
ge remainin
g ene
rgy of the network for
both p
r
oto
c
ol
s a
s
a fun
c
ti
on of t
r
an
smi
ssi
on
rou
n
d
s
. Ou
r p
r
op
ose
d
metho
d
al
ways re
peate
d
ly
re-uses
opti
m
um path
s
with minim
u
m hop
cou
n
t while th
e ene
rgy nod
es i
n
the route
pat
h are
better tha
n
th
eir alte
rnative
node
s to
ro
ute the
d
a
ta
packet
s
. The
r
efore, the p
r
opo
sed
meth
od
perfo
rms
better than the E
E
RP proto
c
ol
in term
s of
saving n
ode
energy. This
indicates that
a
good e
nergy balan
ce i
s
achieved in
a WSN throu
gh o
u
r propo
se
d method.
Many appli
c
a
t
ions in WS
Ns req
u
ire a timely
respon
se, includin
g
for example fi
re ala
r
m
system
s.The delay
in
tran
smissi
on
of a data pa
cket from the sen
s
i
ng nod
es to the ba
se stati
o
n
depe
nd
s on
the tra
n
smi
s
si
on time, a
nd
it shoul
d b
e
sho
r
t, whi
c
h
lead
s to a
fast respon
se
wi
th
high
quality o
f
se
rvice.
The
co
mpa
r
ison
betwe
en t
w
o different app
r
oache
s
i
s
sh
own
in Figu
re
6.
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Prolon
ging th
e Lifetim
e
of
Wirel
e
ss Sen
s
or
Networks usin
g… (Ahm
ed A. Alkathm
a
wee)
397
it can be
see
n
that the pro
posed alg
o
rit
h
m ha
s
the shorte
st delay
more 5
0
% than that obtai
ned
by the EERP proto
c
ol.
Figure 5. Average n
e
two
r
k remai
n
ing en
ergy
as a fun
c
tion
of transmi
ssi
on rou
n
d
Figure 6. dela
y
in transmi
s
sion of a data
packet in the
two app
roa
c
h
e
s (LPA
-sta
r
and
EERP)
6. Conclusio
n
Energy re
so
u
r
ce
con
s
traint
s are o
ne of the mo
st critical challenges
for wirele
s
s
sen
s
o
r
netwo
rks
(WSNs). In spit
e of the fact
routing
sel
e
ction i
s
quite
difficult as
a
result of en
ergy
con
s
trai
ned,
but thi
s
i
s
strongly
rel
a
ted to
extendin
g net
w
ork lifetime.
Un
even
en
ergy
con
s
um
ption
is an
inn
a
te p
r
oble
m
in
WS
N which le
ad
to red
u
ce the
netwo
rk lifetime. To p
r
olo
n
g
the lifetime of the wh
ole
netwo
rk
an
d eliminate t
he chan
ce
o
f
network
pa
rtition, we h
a
ve
prop
osed a n
e
w efficient routing
protocol called LPA
–
star p
r
oto
c
o
l
. The new m
e
thod u
s
e
s
the
LPA-sta
r
al
g
o
rithm to
re
peatedly fin
d
an
optimu
m
ro
uting
pat
h.The
optimu
m
path
can
b
e
sele
cted
by favoring
the h
i
ghe
st re
sid
u
a
l ene
rgy,
mi
nimum h
o
p
count an
d lo
west traffic loa
d
.
The LPA
-
sta
r
proto
c
ol
al
ways reu
s
e
s
t
he p
a
rt
of th
e previou
s
search
ope
rati
on to
re
comp
uted
the ne
w o
p
timum routing
path with
out runnin
g
the al
gorithm f
r
om
the scratch.Si
mulation
re
su
lts
sho
w
the app
rop
r
iaten
e
ss and better p
e
r
forma
n
ce of new p
r
oto
c
ol again
s
t EERP protocol wi
th
rega
rd
s to en
han
ceme
nt of the WSNs lifetime and
b
a
l
anced en
ergy con
s
um
ption
with fast dat
a
transmissio
n across the ne
twork.
Referen
ces
[1]
Cha
ndir
a
sek
a
ran D, T
Jay
a
barath
i
. W
e
rel
e
ss s
enser N
e
t
w
o
r
ks No
de
Local
izatio
n-A
Performan
c
e
Comp
ariso
n
of Shuffle
d
F
r
o
g
Le
api
ng
a
nd
F
i
refl
y Al
gorith
m
in
La
bVIEW
. T
E
LKO
M
NIKA Indo
nes
i
a
n
Journ
a
l of Elec
trical Eng
i
ne
eri
ng.
201
5; 14(3
)
: 516-52
4.
[2]
Itu Snigdh, Ni
sha G
upta
. Energy An
d Evalu
a
ting Q
u
asi
Rand
o
m
Dep
l
oy
me
nt in Z
i
gbe
e Bas
e
d
W
i
reless S
ens
or N
e
tw
orks
. 10th T
E
LKO
M
NIKA Indo
nesi
a
n Jo
urna
l
of
E
l
ectrical
En
gin
eeri
ng. 2
0
1
5
;
15(1): 14
2-1
5
0
.
[3]
Rani
a Kh
ad
im
, Mohamme
d
Erritali, Ab
de
lhakim
Maa
d
e
n
. Ran
g
-F
ree
Loca
lizati
on
Schemes fo
r
W
i
reless S
ens
or N
e
t
w
orks.
T
E
LKO
M
NIKA Indo
nesi
a
n
Jo
u
r
nal
of El
ectric
al E
ngi
ne
erin
g
. 20
15;
16(2)
:
323-
332.
[4]
Soro S, W
end
i B He
inze
lma
n
. Cluster
He
ad El
ectio
n
T
e
chn
i
qu
es F
o
r
Cover
age Pr
eservati
on
I
n
W
i
reless Se
ns
or Net
w
orks.
A
d
Hoc Netw
ork
s
, 2009; 7: 955
-972.
[5]
Selcuk
O
,
Cel
a
l O
,
Derv
is K
a
rab
oga.
A C
o
mp
arativ
e Stu
d
y o
n
D
i
fferent
ial Ev
ol
ution
B
a
sed
Ro
utin
g
Im
plem
entations for Wire
less
Sensor Netw
o
r
ks.
IEEE
Int.
Conf. Innovati
o
ns in Intelli
ge
nt S
y
stems and
Appl
icatio
ns (INIST
A); 2012: 837-
844.
[6]
Z
hang
H, Sh
e
n
H, T
an Y.
O
p
timal E
ner
gy
Bala
nced
Data
G
a
therin
g i
n
W
i
reless S
ens
or Netw
orks
.
IEEE Int. Conf. Parall
el an
d Di
stributed Proc
e
ssing S
y
mp
osi
u
m (IPDPS). 2007.
[7]
W
u
X, Che
n
G
,
Das S. Avo
i
din
g
Ener
g
y
Holes i
n
W
S
Ns
w
i
th N
o
n
unif
o
rm Nod
e
Dist
ributi
on.
IE
EE
T
r
ansactio
n
s o
n
Paral
l
el a
nd
Distribut
ed Sys
t
ems
. 20
08; 19
: 710-72
0.
[8]
Jia J, Z
h
a
ng
G
,
W
u
X, Ch
e
n
J. O
n
the P
r
obl
em of En
e
r
g
y
Bal
anc
ed
Rela
y
Se
nsor
Placem
ent i
n
W
i
reless Se
ns
or Net
w
orks.
In
ternatio
nal J
o
u
r
nal of Distri
but
ed Sens
or Net
w
orks.
2013.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
1, No. 2, February 201
6 : 390 – 398
398
[9]
W
Xu, et a
l
. D
i
stribute
d
Opti
mal Rat
e
-Re
lia
bilit
y-L
i
fetime
T
r
adeoff In
T
i
me-Var
yin
g
W
i
reless S
enso
r
Net
w
orks.
IEEE Transactions
on Wireless
Comm
uninction.
2
014; 13(
9): 483
6-48
47.
[10]
Park J, Sahni
S. An Online
Heuristic F
o
r
Maxi
mum Lifet
i
me Routi
ng i
n
W
i
reless Sen
s
or Net
w
ork.
IEEE Transactions on Com
p
uters
. 2006; 5
5
(8
): 1048-1
0
5
6
.
[11]
W
u
C, Yuan
R, Z
hou H.
A
Nove
l Lo
ad
Bala
nced
And
Lifeti
me M
a
xi
mi
z
a
t
i
o
n
Ro
uti
ng Protoc
ol
in
WSNs
. Procee
din
g
IEEE Vehi
cular T
e
chnol
o
g
y
C
onf. Sing
a
pore. 20
08: 11
3-11
7.
[12]
T
s
ai M, Yang H, Huan
g W
.
Axis-Based
Virtual C
oord
i
nate Assi
gn
ment Protoco
l
And De
livery-
Guarante
ed R
outin
g Protoco
l
in W
S
Ns
. Proc. IEEE Int. Conf. Comput. Commun. 200
7: 2
234-
224
2.
[13]
Li X, Hon
g
S, F
ang
K.
W
S
N
H
A–GAHR: A
Greed
y A
nd
A
He
uristic R
o
u
t
ing A
l
gor
ithm
F
o
r W
i
reles
s
Sensor N
e
t
w
or
ks in Home Aut
o
matio
n
.
IET C
o
mm
. 20
11; 5(
13): 179
7-1
805
.
[14]
Ran
a
K, Z
a
veri M. A-star
algorithm for ener
g
y
effici
ent rou
t
ing in
w
i
r
e
less
sensor net
w
o
r
k
.
T
r
ends in
Netw
ork and C
o
mmunic
a
tio
n
s
. Springer. 20
1
1
: 232-2
41.
[15]
MJ T
s
ai, HY
Yang, W
Q
Hu
ang.
Axis-
bas
ed virtu
a
l co
or
din
a
te a
ssi
gn
me
nt protoc
ol
and
del
ivery-
guar
ante
ed rou
t
ing protoc
ol i
n
W
S
Ns
. Proc. IEEE Int. Conf.
Comp
ut. Commun. 200
7: 22
34-2
242.
[16]
Abdu
lla A, Ni
shi
y
am
a H, Y
ang J, Ansari
N, Kato N.
H
y
m
n
: A Novel H
y
bri
d
Mult
i-Hop R
outi
n
g
Algorit
hm T
o
Improve T
he Long
evit
y
Of W
S
Ns.
IEEE Tr
ansactions on Wireless Comm
unications
.
200
7; 11(7): 25
31-2
541.
[17]
Z
hao Y, W
u
J,
Li
F
,
Lu
S. On Ma
ximizi
ng
T
he Lifetime
Of W
i
reless
Sen
s
or N
e
t
w
orks
Using
Virtu
a
l
Backbo
ne Sch
edu
lin
g.
IEEE Transactions on Parallel
and Distributed System
s
. 20
12;
23(8): 15
28-
153
5.
[18]
Ghaffari A. An Energ
y
Effici
e
n
t
Routing Pr
oto
c
ol for W
i
reles
s
S
ensor Net
w
orks usin
g A-star Algor
ithm.
Journ
a
l of App
l
ied R
e
searc
h
a
nd T
e
chn
o
l
ogy
. 2015; 12(
4): 815-8
22.
[19]
Koen
ig S, Likh
achev M.
In
crem
en
ta
l
A
. Proc
eed
ings
of the Neura
l
Information Proc
essi
ng S
y
stems
.
Vanco
u
ver, Bri
t
ish Col
u
mbi
a
, Can
ada. 2
002.
[20]
Liu Y, Koen
ig
S, David F
. Spee
din
g
Up T
he Calc
ul
ation
Of Heuristics F
o
r Heuristic
Search-B
as
e
d
Plan
nin
g
. Proc. the Nation
al C
onfere
n
ce o
n
Artificial Intel
lig
e
n
ce. 200
2: 484
-491
.
[21]
Koen
ig S, Likh
achev M,
F
u
rcy D. Lifel
o
n
g
Pl
ann
ing A.
Artificial Intelligence
. 2004; 1
55(2
004): 93-
14
6.
[22] Yao J, Lin C, Xi
e X, W
a
n
g
A, Hung C.
Path
Plann
ing F
o
r
Virtual Hu
man
Motion Usi
ng I
m
pr
ove
d
-Star
Algorit
h
m
.
Seventh Intern
atio
nal C
onfere
n
c
e
on Informa
ti
on T
e
chnol
og
y: Ne
w
Ge
nerat
ions (IT
N
G).
201
0: 115
4-11
58.
[23]
Aine S,
Lik
hac
hev M.
T
r
uncated Incr
ementa
l
Search: F
a
ste
r
Rep
l
an
ni
ng B
y
E
x
plo
i
tin
g
Su
b Optimal
i
t
y
.
T
r
uncated Incr
ementa
l
Searc
h
: F
a
ster R
epl
ann
ing
by Expl
oitin
g
Sub
opti
m
a
lity
. 201
3.
[24]
Yoon
S, Shim
D SLPA. Sh
ap
e-A
w
ar
e L
i
felo
ng
Pl
an
nin
g
A
*
for D
i
fferenti
a
l W
h
e
e
l
ed V
e
hicles.
IEEE
T
r
ansactio
n
s o
n
Intelli
ge
nt Systems
. 20
15; 1
6
(2): 730-
74
0.
[25]
Lu Y,
Hu
o
X,
Arslan
O, T
s
iotras P.
Multi-S
c
ale
LPA
*
w
i
th
Low
W
o
rst-C
a
se
Co
mp
lexit
y
Guara
n
tees
.
IEEE/RSJ Internatio
nal C
onfer
ence o
n
Intell
ig
ent
Rob
o
ts and
S
y
stems (IROS). 2011: 35
07
-351
2.
[26]
Park J, Sah
n
i
S. An Onli
ne
Heuristic
F
o
r
Maxi
mum
Lifet
i
me R
outi
ng i
n
W
i
reless
Sen
s
or Net
w
o
r
ks
.
IEEE Trans. On Com
p
uters
. 200
6; 55(8): 10
48-1
056.
[27]
W
u
C, Yuan
R, Z
hou H.
A
Nove
l Lo
ad
Bala
nced
And
Lifeti
me M
a
xi
mi
z
a
t
i
o
n
Ro
uti
ng Protoc
ol
in
WSNs
. Proc. IEEE Vehicu
lar
T
e
chnolog
y
C
o
nf
erenc
e. Sing
apor
e. 200
8: 113-1
17.
[28]
Heinz
e
lm
an W
,
Chan
drak
asa
n
A, Bal
a
krish
nan
H.
Ener
g
y
Efficient C
o
mmu
n
icati
on
Protocol F
o
r
Wireless Micro Sensor Networks
. Proc. 33rd Ann. Ha
w
a
ii Int
.
Conf. S
y
st. Sci. 2000: 1-1
0
.
Evaluation Warning : The document was created with Spire.PDF for Python.