TELKOM
NIKA
, Vol. 11, No. 7, July 201
3, pp. 3576 ~ 3584
e-ISSN: 2087
-278X
3576
Re
cei
v
ed
Jan
uary 7, 2013;
Re
vised Ap
ril
3, 2013; Accepted April 1
5
, 2013
Study on an Energy-aware Routing Algorithm for
Agriculture WSN
Huarui
Wu*, Chunjiang Z
h
ao, Li Zhu
Natio
nal En
gi
n
eeri
ng Res
earc
h
Center for C
enter
for Information T
e
chno
log
y
i
n
Agricu
lture, Beij
ing
100
09
7, Chin
a
Ke
y
Lab
orator
y for Information
T
e
chnol
ogi
es i
n
Agricu
lt
ure, Ministr
y
of Agri
cult
ure, Bei
i
ng
100
09
7, Chin
a
*Corres
p
o
ndi
n
g
author, e-ma
i
l
:
w
u
hr@n
ercit
a
.org.cn
*
, zhaocj@n
e
rcita.or
g.cn, zhul@
ner
cita.org.cn,
Chu
n
ji
ang Z
h
a
o
A
b
st
r
a
ct
As precis
io
n fa
rmi
ng r
equ
ires
real-ti
m
e,
accu
rate
a
nd s
u
stai
nab
le
monitor
i
ng of th
e
bioto
pe of th
e
field, e
nergy c
onsu
m
ption
an
d netw
o
rk de
l
a
y consti
tute t
he
maj
o
r facto
r
s affecting th
e perfor
m
ance
of
Z
i
gBee-
base
d
W
i
reless Se
ns
or Netw
ork (W
SN) app
lie
d
in
agricu
l
tural
pr
oducti
on. Base
d on the AOD
V
jr
alg
o
rith
m, this
pap
er pr
ese
n
ts a n
e
w
al
gorith
m
, n
a
m
ely
NS-
A
ODVjr.
F
eatu
r
ing e
m
p
hasis
on ener
gy
c
ont
rol
and
dyna
mic routin
g, as w
e
ll
as bal
anc
e b
e
tw
een en
er
gy
consu
m
pti
on
and s
hortest p
a
th routi
ng, thi
s
alg
o
rith
m r
eal
i
z
e
s
max
i
mu
m
service
ti
me
of
the W
S
N
w
h
il
e e
n
suri
ng
ti
mely
and
effecti
v
e trans
missio
n
o
f
the data i
n
the
mo
nitori
ng pr
o
c
ess. After a simu
lati
on
ex
per
iment w
i
th the
Netw
ork Simul
a
tor tool, resu
lts
show
that, co
mpare
d
to
the
ori
g
in
al
al
gorith
m
,
alth
ou
gh t
he
p
a
cket d
e
liv
ery r
a
te a
n
d
n
e
tw
ork de
lay
w
i
tness
no i
m
pr
ove
m
e
n
t, the new
alg
o
rith
m not o
n
ly
signific
antly
re
duces th
e dev
i
c
e en
ergy co
n
s
umptio
n, but
also
effectively exp
and the s
e
rv
ice
time of the net
w
o
rk.
Ke
y
w
ords
:
en
ergy-aw
a
re, W
S
N, routing a
l
g
o
rith
m, AODVjr, agricultur
a
l
Copy
right
©
2013 Un
ive
r
sita
s Ah
mad
Dah
l
an
. All rig
h
t
s r
ese
rved
.
1. Introduc
tion
ZigBee
routin
g protocol is
suitabl
e for
sma
ll-s
i
ze
d
and
c
l
os
e-
ra
n
g
e data tran
smissi
on,
and mo
st
of
t
he studie
s
on
ZigBee
p
r
ot
ocol
-ba
s
e
d
ro
uting al
gorith
m
[1] a
r
e fo
cuse
d o
n
routi
ng
policy optimi
z
ation. Fo
r example, the
study
on transmitting p
o
we
r optimizing is aimed
at
redu
cin
g
ene
rgy con
s
u
m
pt
ion for the co
mmuni
cati
on
betwe
en short-distan
ce
d n
ode
s, so a
s
to
reali
z
e better communi
cati
on with lower transmi
tting
power; the st
udy on
the topology features
and ad
dre
s
s allocation me
cha
n
ism of th
e ZigBee net
work is ai
me
d at redu
cing
the transmitti
ng
radiu
s
of the routing requ
e
s
t packet thro
ugh ada
pt
ive adju
s
tments
optimizatio
n, so a
s
to reali
z
e
lowe
r po
we
r con
s
um
ption.
For la
rge
-
sca
l
e agri
c
ultu
ral
prod
uctio
n
, as the
cro
p
gro
w
th cy
cle
is long a
nd
covers
a
wide
area, th
e bioto
pe m
o
nitoring
sy
ste
m
ba
se
d
on
WSN ha
s m
a
ny disadvant
age
s, such a
s
the
field imped
a
n
ce, hig
h
en
ergy con
s
um
ption, di
fficult maintena
nce, inequality
of node e
nergy
con
s
um
ption,
sho
r
t se
rvice
time and oth
e
r issu
es
. T
h
e system
often utilize
s
the
ZigBee routin
g
proto
c
ol
-ba
s
e
d
AODVj
r
(Ad-ho
c On
-de
m
and Di
stan
ce Ve
cto
r
ju
nior
routin
g) whi
c
h i
s
m
o
re
simplified
an
d
practi
cal
to
reali
z
e com
m
unication
b
e
twee
n
the nod
es or betwee
n
the node
s and
the netwo
rk [
2
]. Such a practice may re
sult in
the failure of a path
or
even crash of the entire
netwo
rk,
a
s
n
ode
s
with lo
w e
n
e
r
gy ma
y become
in
v
a
lid d
ue to
to
o mu
ch
en
ergy co
nsumpti
on.
Con
s
id
erin
g the effectiven
ess an
d servi
c
e time
of
the network
[3], it is
hard but nec
e
ss
ary to find
the best path
[4] while taking into acco
unt the ener
gy state of th
e node, so a
s
to redu
ce t
h
e
energy con
s
u
m
ption of the whol
e network
and p
r
olo
n
g
the network
servi
c
e time.
Con
s
id
erin
g the spe
c
ial fe
ature
s
of la
rg
e-scal
e [5] preci
s
ion
agri
c
ultural p
r
o
d
u
c
tion, this
pape
r, ba
sed
on AO
DVjr,
pre
s
ent
s a
n
e
w al
gorith
m
whi
c
h
combi
n
es p
a
th weig
hts an
d dyna
mic
routing. In o
r
der to b
a
lan
c
e the en
ergy
con
s
um
pt
ion
of different n
ode
s, and
effectively tran
smit
the data
with
limited e
nergy, this al
gorithm take
s i
n
to co
nsi
deration the
e
s
timated remai
n
i
n
g
work time
of
the nod
e dev
ice in
dete
r
m
i
ning the
final
datag
ram transmi
ssion
p
a
th, reali
z
in
g
a
dura
b
le an
d stable monito
ri
ng syste
m
for preci
s
io
n farming.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
Study on an
Energ
y
-aware Routin
g Algorithm
for Agriculture WS
N (Hu
a
rui Wu
)
3577
2. Related Works
There are two routin
g alg
o
rithm
s
in Zi
gbee
NWK la
yer [6] .One i
s
AODVj
r
, a
modified
Ad-ho
c
O
n
-Demand
Di
sta
n
ce Ve
cto
r
(AODV),
whi
c
h
is the
defa
u
lt routing
al
gorithm fo
r Zi
gBee
proto
c
ol. Th
e
other is
a
co
mparatively new al
go
rithm calle
d
Hi
erarchi
c
al Routin
g
Algorith
m
. The
se
con
d
alg
o
rithm ca
n lea
d
to un
bala
n
c
ed
ene
rgy
distrib
u
tion i
n
the net
work.
Wh
at is m
o
re,
some
no
des
requi
re to
o
much
en
ergy
con
s
u
m
ption
[7], which m
a
y re
sult in t
he cra
s
h
of the
whol
e net
work. Such a l
a
ck of
stability
makes it
unsuitable for agricult
ure moni
toring system
s,
whi
c
h re
qui
re
long se
rvice
time and high
stability.
AODV algo
rithm has b
e
e
n
widely app
lied in wirel
e
ss a
d
hoc n
e
tworks as a
routing
proto
c
ol for
mobile no
de
s. As it was n
o
t desig
ned
f
o
r wi
rele
ss sensor net
wo
rks, it did not
take
into acco
unt
the lower
energy
co
nsumption req
u
irem
ent
of wi
rel
e
ss sen
s
or
net
works
[8].
Con
s
id
erin
g t
he fa
ct that t
he n
ode
s i
n
a WS
N
are
gene
rally in
f
i
xed po
sition
s, the
AODV
jr
algorith
m
is
more
co
nci
s
e
and si
mplifie
d than the
A
O
DV alg
o
rith
m in term
s of
maximal re
du
ction
in nod
e e
nergy co
nsumpti
on. The
AODVjr algo
rith
m
inherit
s o
n
ly the dyna
mic routing featu
r
e of
the AODV al
gorithm, a
nd
omits all o
p
timization
me
a
s
ures,
su
ch
a
s
Hello me
ssage, ro
uting
error
messag
e, an
d que
ry se
rial
numbe
r, etc.
, which
are d
e
sig
ned fo
r b
e
tter nod
e m
obility, realizi
ng a
maximum
si
mplification
of the AO
DV algo
rithm.
Than
ks
to su
ch a simp
lification,
AO
DVjr
algorith
m
performs m
u
ch better than th
e AODV al
go
rithm in ene
rgy con
s
um
ption, and is
wi
dely
use
d
in vario
u
s wi
rele
ss sensor net
wo
rks.
Ho
wever,
be
cau
s
e
AO
DVjr al
gorith
m
d
oes not
take
into
co
nsi
d
e
r
ation
en
ergy
control
over the
bottleneck
node i
n
the netw
ork, it still has potential to be
further improved. The AODVjr
algorith
m
det
ermin
e
s the fi
nal routin
g pa
th by co
mpa
r
i
ng the wei
ght
s of different
paths. Th
e path
weig
ht is th
e
comm
uni
cati
on
co
st between t
w
o n
ode
s,
which is u
s
ually set
as a
co
nsta
nt value.
In fact, the routing algo
rith
m of AODVjr is
a kin
d
of Shortest Path Routing Algo
rithm.
The sho
r
te
st path
routing
algorith
m
can
effect
ively re
duce n
e
two
r
k delay, a
s
m
o
st of the
netwo
rk d
a
ta
gram
can b
e
transmitted al
ong this p
a
th and arrive at the destin
a
tio
n
node q
u
ickl
y.
But the p
r
ice
is that th
ose
node
s i
n
the
sho
r
test
path
will
con
s
um
e
a lot of
po
wer, leadin
g
to th
e
failure of
som
e
key n
ode
s
becau
se of in
sufficie
n
t
po
wer, whi
c
h i
n
turn le
ad
s to redu
ced
network
serv
i
c
e t
i
me.
3. Impro
v
ed
AODVjr Routing Algorithm
All kind
s of
monitori
ng
systems face
the
same
problem
of h
o
w to
achieve
bala
n
ce
betwe
en dev
ice ene
rgy consumption
and network
delay [9].
Therefore, we must ch
o
o
se
approp
riate routing alg
o
rit
h
m or ma
ke
further
imp
r
o
v
ements a
c
cordin
g to different mo
nitoring
obje
c
ts, different monitori
n
g
req
u
ire
m
en
ts and t
he re
ality.
In
agriculture WSN, device
e
n
e
r
gy
con
s
um
ption
is more impo
rtant than ne
twork del
ay. So, the focus of this pape
r is on h
o
w t
o
redu
ce
the
n
ode
ene
rgy
consumption
a
s
m
u
ch
a
s
p
o
ssible
an
d e
x
pand th
e
se
rvice tim
e
of
the
netwo
rk, whil
e maintainin
g
the normal p
e
rform
a
n
c
e o
f
the monitori
ng syste
m
.
Acco
rdi
ng to the above an
alysis, this p
aper p
r
e
s
ent
s a new AO
DVjr algo
rith
m base
d
routing
algo
ri
thm, which combine
s
e
n
e
r
gy co
nt
rol a
nd dynami
c
routing. In ro
uting, the ne
w
algorith
m
takes into
con
s
i
deratio
n the
estima
ted
re
maining
wo
rk time of the
node
device
in
deci
d
ing the f
i
nal datag
ram
transmi
ssion
path.
3.1. Calculati
on of Path E
n
erg
y
Weight
Many
challe
n
ges,
su
ch
a
s
the h
uge
nu
mber
of d
e
vice
nod
es, li
mited batte
ry ene
rgy
cap
a
city, and
insufficie
n
t band
width, call for
energ
y
control me
cha
n
ism to redu
ce the n
ode
energy con
s
u
m
ption and e
x
pand the se
rvice time of
the netwo
rk. The prin
cipl
e
of the energ
y
control mech
anism p
r
e
s
e
n
ted in this pape
r is a
s
follows: the final routin
g path is cal
c
ul
ated
according to the node
energy
wei
ght
, whi
c
h
represents the
probability of
the node bei
n
g
inclu
ded i
n
to
the ro
uting
set. The g
oal
of this
ene
rgy
wei
ght ba
se
d calculation
method,
whi
c
h is
the core of ou
r ene
rgy co
ntrol me
cha
n
ism, is to
conv
ert the remai
n
ing en
ergy o
f
the node de
vice
to its ene
rgy
weig
ht, whi
c
h i
s
a
key
referen
c
e f
o
r routing
calcul
ation. T
he me
cha
n
ism is
descri
bed a
s
follows:
1) Let
()
in
it
E
i
be th
e i
n
itial battery
cap
a
city of
n
ode
i
N
at th
e t
i
me
when
it j
o
ins the
Zig
B
e
e
netwo
rk, an
d the remai
n
ing
battery capa
city of node
i
N
is den
oted by
()
ho
l
d
i
E
.
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN:
2087
-27
8
X
TELKOM
NIKA
Vol. 11, No
. 7, July 2013
: 3576 – 357
7
3578
2)
In the
co
urse
of e
a
ch routi
ng, ea
ch
no
d
e
i
N
will
record it
s energy consumption
caused by
transmissio
n, rece
ption, an
d monitorin
g
activities in
()
ij
E
T
, and then e
s
timate the ene
rgy
con
s
um
ption
for the
nex
t
T
seconds
by utilizing the Ex
ponent
ial Weighted Moving
Average met
hod:
1
()
(
)
(
1
)
(
)
ij
ij
ij
E
TE
T
E
T
(1)
Whe
r
e
{
0
,
1
,
2
.
..}
j
,
01
,
is a vari
able
to adju
s
t
1
()
ij
E
T
and
()
ij
E
T
.
1
()
ij
E
T
re
pre
s
e
n
ts th
e e
s
timated
e
nergy
con
s
u
m
ption i
n
the
previou
s
T
se
c
ond
s,
()
ij
E
T
rep
r
e
s
ent th
e ene
rgy co
nsum
ption v
a
lue in the
newly
T
seconds. For the initial value,
(
)
()
()
i
j
in
it
h
o
ld
E
TE
i
E
i
.
3) Let
L
be
the estimated nu
mber of
re
si
dual workin
g
cycle of
no
de
i
N
at time
j
T
,
meanin
g
nod
e
i
N
can al
so work fo
r
T
se
con
d
s.
L
is cal
c
ul
a
t
ed as
()
()
hol
d
ij
E
i
L
E
T
.
i
Fn
represents t
he estimate
d
lifetime of node
i
N
at time
j
T
,
i
Fn
is calculate
d
as
follows
:
1
0
()
()
()
m
ho
l
d
ii
j
i
ij
E
i
Fn
E
T
ET
(2)
Acco
rdi
ng to
Eq. 2, the
long
er lifeti
m
e of
a no
de me
an
s t
he mo
re
nu
mber of
wor
k
in
g
T
. The
node with mi
nimum
i
Fn
must be the bottlen
eck node of
a
route path, whi
c
h is
comp
osed of
N nod
es. O
n
ce the b
a
ttery
of this nod
e i
s
de
pleted, th
is path b
e
co
mes invali
d a
n
d
a new p
a
th should b
e
cal
c
ulated.
4) Assu
ming
{|
1
,
.
.
.
}
i
P
Pi
m
as the set of a
ll the possibl
e route
s
bet
wee
n
given source
s
N
and d
e
stin
ation
d
N
. We u
s
e t
he minim
a
l
i
Fn
for
i
P
to re
pre
s
e
n
t the maxim
a
l lifetime of t
h
is
path, and the
averag
e maxi
mal lifetime of set
P
is
1
1
()
m
i
i
Fn
F
n
P
m
.
W
e
us
e
()
s
um
i
E
P
to repre
s
ent th
e total ene
rgy
consumed
by
each no
de of
path
i
P
for
data tran
smi
s
sion. If the powe
r
co
nsum
ption of
each
node is fixed
in the transmissi
on p
r
o
c
ess,
we
ca
n u
s
e
su
ch
a fixed
value to m
u
ltiply the nu
mb
er of
nod
es to get
()
s
um
i
E
P
. Similarly, the
averag
e valu
e of energy consumption o
f
set P is
1
1
()
m
s
um
s
u
m
i
i
EE
P
m
.
For a given
set
P
, each path
weight can b
e
cal
c
ulate
d
as follo
ws:
12
()
()
()
(
1
)
(
)
is
u
m
s
u
m
i
i
su
m
F
nP
E
E
P
Val
P
v
al
v
a
l
E
Fn
(3)
Whe
r
e
1
va
l
and
2
va
l
are bot
h weig
ht factors,
1
01
va
l
, 0<
2
va
l
<1, and
12
1
va
l
v
a
l
.
3.2. Implementa
tion of th
e D
y
namic Routing Mec
h
anism
Appro
p
riate
routing requ
e
s
t is p
r
e
r
eq
u
i
si
te for
due
perfo
rman
ce
of the path
ene
rgy
weig
ht. In old
practi
ce, the
so
urce
no
de
loo
k
s for ne
w
routin
g p
a
ths
by p
e
rio
d
i
c
ally
sub
m
itting
routing
re
que
sts.
Th
e sou
r
ce node
s do not
u
s
ually
consi
der whet
her th
ere
is b
o
ttle ne
ck
no
de in
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
Study on an
Energ
y
-aware Routin
g Algorithm
for Agriculture WS
N (Hu
a
rui Wu
)
3579
the netwo
rk,
or the influen
ce some hig
h
-
ene
rgy-
co
nsumption no
de
s. Be
ca
use routing re
que
sts
also in
cu
r e
nergy con
s
u
m
ption, more rout
ing
re
que
sts mea
n
more en
ergy con
s
umpt
ion,
therefo
r
e, it is not wise to a
pply su
ch a p
r
acti
ce in Zig
B
ee netwo
rk.
Based o
n
AODVjr, this p
aper p
r
e
s
ent
s an im
prove
d
dynamic ro
uting algo
rith
m, NS-
AODVjr
（）
Ne
w Save AO
DVjr
, whi
c
h
adju
s
ts th
e routing
req
u
e
s
t a
c
tivity by introd
uci
ng
the
con
c
e
p
t of path energy wei
ght. This dyn
a
mic
routing
mech
ani
sm is describ
ed a
s
follows:
1
1
rre
p
s
u
m
E
E
(4)
Figure 1.
Packet delivery rate
Figure 2.
Average d
e
lay
Her
e
rre
p
E
denote
s
the remai
n
i
ng battery ca
pacit
y of the node while receivin
g RREP
packet
s
;
(0
1
)
is a factor whi
c
h
can be set accordi
ng to real traffic loa
d
situation. T
he
impleme
n
tation of the algo
rithm incl
ude
s four
step
s:
1)
Each
n
ode should store
1
()
ij
E
T
to cal
c
ulate th
e value of
()
ij
E
T
and
i
Fn
a
c
c
o
r
d
ing to
formula 1, 2 for every
T
.
2)
For the initial
value,
(
)
()
()
i
j
i
n
it
h
o
ld
E
TE
i
E
i
is used only for th
e first
T
seco
n
d
.
For
each follo
win
g
T
, we
sh
ould
save th
e p
r
evious va
riab
le
()
hol
d
i
E
in the
variable
()
ol
d
h
ol
d
i
E
,
and the value
of
()
ij
E
T
is upd
ate
d
usin
g
()
(
)
ol
dh
o
l
d
i
h
o
l
d
i
EE
.
3)
Adding the p
a
thco
st and
hop
cou
n
t to store
the e
s
ti
mated lifetime of the path and routin
g
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN:
2087
-27
8
X
TELKOM
NIKA
Vol. 11, No
. 7, July 2013
: 3576 – 357
7
3580
hop count re
spectively into RREP pa
cket
.
4) A
rre
p
E
field is a
d
ded into th
e
route ta
ble.
Upo
n
re
ceivi
ng a
RREP
packet, up
dat
e
rre
p
E
according to the rem
a
ining
energy of the node.
The detaile
d pro
c
e
ss of th
e four ste
p
s i
s
sh
own as fo
llows:
Algorithm 1
:
Algorithm 2
:
Inpu
t:
T
Ou
tpu
t
:
()
ij
E
T
,
i
Fn
if
T
>= 0
the
n
Calculate
()
ij
E
T
and
i
Fn
;
if
routet
able
!
=
N
U
LL
the
n
if
（
4
）
is true
the
n
Update r
outetabl
e;
end
end
Inpu
t:
RR
EQ
Ou
tpu
t
:
RREQ, Update
r
outetabl
e
RREQ->
hopcoun
t
+= 1;
if
i
N
is dest node
th
e
n
for
i=1 to m
do
calculate
Val(P
i
)
Select a path P
i
w
i
th ma
x
i
mal Val;
Send a RREP ac
cording path P
i;
else if
i
Fn
>0
the
n
Build the reversal path;
if
RRE
Q->
pathc
ost
>
i
Fn
then
RREQ->
pathcost
=
i
Fn
;
Update r
outetabl
e of node
i
N
;
Update R
R
EQ;
end
end
Al
g
o
r
i
t
h
m
3
:
Al
g
o
r
i
t
h
m
4
:
Inpu
t:
RR
EP
Ou
tpu
t
:
rre
p
E
Select a for
w
ard
path;
if
i
N
is not a source
node
th
e
n
Update
rre
p
E
w
i
th
()
ho
l
d
i
E
;
Forw
a
r
d RREP;
if
i
N
is a dest node
then
Send date packe
ts;
end
Inpu
t:
packet
Ou
tpu
t
:
rre
p
E
Lookup the ro
ute
table entr
y
for th
e dest;
if
routet
able is Update
th
e
n
For
w
a
r
d the
packet;
Send RRE
Q;
routetable->flags
=RTF_UP;
else if
r
outetable
-
>flags=RTF_UP
then
For
w
a
r
d packet;
end
4. Simulation Experimen
t
In this
pap
er,
we
refe
r to
this
ne
w alg
o
r
ithm a
s
NS-AODVjr
(Ne
w
Save AO
DVjr).Thi
s
section will simulate
the perform
a
nce of
both
t
he
old and new
routing al
gorithms i
n
Network
Simulator en
vironme
n
t, an
d p
r
e
s
ent th
e
co
mpa
r
ison
betwe
en
NS-AODVjr an
d
AODVjr in te
rms
of pa
cket d
e
livery rate,
ne
twork
delay,
and
rem
a
inin
g
devi
c
e ene
rgy cap
a
city after
a
certai
n
perio
d of
ope
ration. It i
s
p
r
oved that, u
n
der equ
al
n
e
twork delay,
NS-AO
D
Vjr a
l
gorithm
is be
tter
able to balan
ce the po
wer consumptio
n
of each
nod
e in the network, effectivel
y extending the
netwo
rk
se
rvice time.
4.1. Simulation Model Se
tting
We in
stalled
30 sen
s
or
no
des i
n
the farmland
( an a
r
ea of 50*
50m
2). All these
sen
s
o
r
node
s
are
F
u
ll Fun
c
tion
a
l
Devi
ce(FFD). Since o
u
r experi
m
ent
s are
ba
se
d
on the
JN5
1
4
8
wirel
e
ss mi
croco
n
trolle
r p
r
odu
ced
by Je
nnic,
we
set the tra
n
sceive
r po
we
r pa
ra
meters of sen
s
o
r
node
s acco
rding to the manual in
struction
s
of
Je
nnic. Set Init-Powe
r
is 200
0J And 12 m
e
ter
transmissio
n
rang
e i
s
ad
o
p
ted. Packet
error
ra
tio i
s
set to 0.3
%
and the
p
a
cket si
ze
is 300
bytes. The si
mulation time
is 9000
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
Study on an
Energ
y
-aware Routin
g Algorithm
for Agriculture WS
N (Hu
a
rui Wu
)
3581
4.2. Simulation Res
u
lts a
nd Discu
ssi
on
Wirel
e
ss sen
s
or
netwo
rk for large area
farm
land inf
o
rmatio
n mo
nitoring i
s
se
nsitive to
device
ene
rg
y con
s
um
ptio
n. The
r
efore, routin
g
al
gorithms fo
r
su
ch sy
stems m
u
st b
a
lan
c
e t
he
netwo
rk
pe
rforma
nce an
d the devi
c
e ene
rgy co
nsum
ption. It must en
su
re the farml
a
nd
informatio
n can be tran
smitted in a timely mann
er, at the sam
e
time, the n
e
twork
can
work
stably und
er
unattend
ed si
tuation for a long pe
riod of
time.
Therefore, when we co
mp
are
th
e
e
nerg
y
co
n
s
um
ptio
n of NS
-AO
D
Vjr an
d AO
DVjr, we
must
ma
ke
s
u
re t
hey
h
a
v
e
simila
r net
wor
k
p
e
rfo
r
m
ance. Base
d
on the
pe
rformance testin
g we
carrie
d out fo
r NS
-AO
D
Vjr
and AO
DVjr
comp
ari
s
o
n
, packet d
e
live
r
y rate
and
th
e average
del
ay
(from 8
00s to
8000
s) a
r
e shown in Figu
re 1 and Figu
re 2 respe
c
tively.
Figure 1an
d Figure 2 sh
o
w
that the packet de
livery rate of NS-AODVj
r algo
rithm is
slightly lo
we
r
than that of A
O
DVjr
in the
l
a
tter sta
ge
of the si
mulatio
n
(from 8
0
0
s
to 8000
s). Th
e
differen
c
e i
s
so sli
ght that
it can not affect
the perfo
rman
ce of ag
ricultu
r
al m
o
n
i
toring
syste
m
.
The ave
r
ag
e
delay time
s o
f
both alg
o
rit
h
ms
are
ba
si
cally
t
he
sa
m
e
.
The
r
efore, the
NS
-AODVjr
algorith
m
is similar with A
O
DVjr in p
a
cket
delivery rate and average del
ay. They can me
e
t
the
requi
rem
ents of ag
ricultural monito
ring
system
s.
In
o
t
her words, b
o
th
algo
rithm
s
can gua
ran
t
ee
the timely transmi
ssi
on of the dat
a colle
cted by the sensor no
de.
Figure 3 and
Figure 4 sh
o
w
the rem
a
ini
ng ene
rg
y ca
pacity of all the device no
des afte
r
2500
s an
d 50
00s d
u
rin
g
the simulatio
n
.From Fig
u
re
3 we can see
that every node (from 1 to
30)
has roug
hly the
sam
e
re
maining
en
ergy. This is
b
e
ca
use the
runnin
g
time
of the
network i
s
short and all
the nodes
still have sufficient
remai
n
i
ng energy
. Both algorithm
s
use the
same
routing m
e
tho
d
, namely sh
ortest p
a
th ro
uting of AODVjr, for data routing.
Figure 3. Re
maining e
nergy of nodes a
t
2500s
Figure 4. Re
maining e
nergy of nodes a
t
5000s
Evaluation Warning : The document was created with Spire.PDF for Python.
e-ISSN:
2087
-27
8
X
TELKOM
NIKA
Vol. 11, No
. 7, July 2013
: 3576 – 357
7
3582
Whe
n
the n
e
twork
run
n
ing
time is
sho
r
t
,
the node
s
h
a
ve suffici
ent
remai
n
ing
e
nergy, the
pa
th
-
energy weig
h
t
metric of NS-AODVj
r ha
s little infl
uen
ce on p
a
th ro
uting cal
c
ul
ation, so the
NS-
AODVjr an
d AODVjr alg
o
ri
thms have no
signific
ant differen
c
e in terms of ene
rgy con
s
um
ption.
From Fi
gu
re
4, we
can
se
e that NS
-AO
D
Vjr
al
gorith
m
is b
e
tter th
an AO
DVjr al
gorithm i
n
optimizin
g th
e rem
a
inin
g
energy of the
bottlene
ck
n
ode afte
r 50
0
0
s in th
e
sim
u
lation. Th
e
main
rea
s
on i
s
tha
t
with the run
n
ing of sy
ste
m
, the
influe
nce
of ene
rg
y weight on t
he calcul
atio
n of
path be
com
e
s mo
re an
d more o
b
viou
s, and the
effect of the bo
ttleneck nod
e
beco
m
e
s
m
o
re
importa
nt. The NS-AO
D
Vjr algorithm
ca
n balan
ce
the
energy
con
s
umption of b
o
ttleneck no
d
e
s
according to
netwo
rk co
mmuni
cation
load a
nd
t
he devi
c
e e
nergy
con
s
u
m
ption fa
ctor, by
adju
s
ting the
route
s
involvi
ng bottlene
ck node
s,
re
pla
c
ing th
e bottlene
ck
node
s
by other n
o
d
e
s
with mo
re re
maining
ene
rgy, and effect
ively redu
cin
g
the ene
rgy
con
s
um
ption
of the bottlen
eck
node
s.
As the sim
u
l
a
tion goe
s o
n
, the bottleneck no
de i
n
the AODVj
r
algo
rithm
will stop
runni
ng b
e
ca
use its
power is exh
a
u
s
ted, then
it b
e
com
e
s
a ‘d
ead’ no
de in
the netwo
rk. The
simulatio
n
e
x
perime
n
t re
sults
sho
w
that
the number of ‘dead
’ nodes
cau
s
ed by AODVjr
algorith
m
is
more tha
n
that cau
s
ed by
the NS-A
O
D
Vjr algorith
m
. Therefo
r
e, the lifetime of the
network using NS
-AODVjr al
gorit
hm will certai
nly be
longer
than
that
of the net
work using
AODVjr algorithm.
The re
mainin
g power of th
e WSN i
s
often a ke
y fa
ctor affectin
g the asse
ssme
nt of the
servi
c
e time of the network. As more a
nd more
n
o
d
e
s be
com
e
‘dead’ with the decrea
s
e of the
remai
n
ing
po
wer of the n
e
t
work, the
WSN will
co
ntinue to carry o
u
t re
con
s
tru
c
t
i
on, and
sel
e
ct a
new
ro
uting,
so a
s
to
en
sure th
at the
data of e
a
ch
node
co
uld b
e
se
nt to the
cent
ral n
ode
. In
ca
se th
e
dat
a of m
o
st
of
the no
de
s
ca
n not
be
re
co
nstru
c
ted
a
n
d
sent th
rou
g
h
the
net
work to
the cent
ral no
de, the ZigBe
e
netwo
rk
will
be decl
a
re
d '
dead'.
Figure 5. Re
maining e
nergy of the network no
de
s
Figure 6. Re
maining n
ode
numbe
r
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
e-ISSN:
2087
-278X
Study on an
Energ
y
-aware Routin
g Algorithm
for Agriculture WS
N (Hu
a
rui Wu
)
3583
Figure 5
sh
o
w
s differe
nt remainin
g e
n
e
rgy of th
e
same n
e
two
r
k und
er t
w
o
d
i
fferen
t
algorith
m
s. A
nd Fi
gure 6
comp
are the
remai
n
ing
no
de n
u
mbe
r
of two
differe
nt algo
rithm
s
.
As
the sim
u
latio
n
progresse
s, the rem
a
ini
ng en
er
gy of the net
wo
rk and
rem
a
ini
ng no
de
nu
mber
also
cha
nge
s. The remain
ing ene
rgy o
f
the WS
N u
nder the
NS-AODVjr alg
o
rithm is alway
s
highe
r than t
hat und
er th
e AODVjr
alg
o
rithm,Mea
n
w
hile th
e nu
mber
of rem
a
ining n
ode
s wa
s
increa
sed,
p
a
rticul
arly, a
s
the
time g
oes,
su
ch
a
n
adva
n
tage
of the
NS-AODVjr
algo
rithm
become
s
more notable, sh
owin
g an lon
ger servi
c
e time of the network.
Acco
rdi
ng to
the si
mulatio
n
result, NS-AODVjr alg
o
rithm can
exp
and th
e
effective are
a
of network
coverag
e
, and
extensive farmlan
d
ca
n b
e
long-te
rm
monitori
ng wi
th real-time
a
n
d
su
staina
bility in the unatten
ded environm
ent.
5. Conclusio
n
This
pap
er i
n
trodu
ce
s
a
cal
c
ulatio
n m
e
thod of p
a
th ene
rgy
wei
ght, whi
c
h
combine
s
energy co
ntro
l and AO
DVjr algorith
m
. Base
d on th
e
shorte
st path
routing meth
o
d
appli
ed by t
he
origin
al AO
DVjr algo
rithm,
the ne
w
met
hod
de
cide
s
wheth
e
r to
ta
ke
a no
de i
n
to an
active
ro
ute
or n
o
t a
c
cording to th
e e
s
timated
rem
a
ining
po
wer of the d
e
vice nod
e. Ba
sed on
the
pa
th-
energy wei
g
ht metric m
e
thod an
d dynamic
ro
utin
g mechani
sm, this pap
e
r
presents
a
new
routing al
gori
t
hm, which is adapted for
the ZigBee
a
g
ricultural monitorin
g
syst
em. Simulation
experim
ents
are
condu
cte
d
for the
ne
w al
gorith
m
, NS-A
ODVj
r, in
Network
Simulator. T
h
e
simulatio
n
re
sults
sh
ow th
at the lifetime
of t
he ZigBe
e
agri
c
ultu
ral
monitori
ng
system u
s
ing
NS-
AODVjr
algo
ri
thm ca
n b
e
ef
fectively expa
nded
by expa
nding t
he lifet
ime of the
bot
tleneck
node
s
in the net
work. T
he p
a
c
ket delivery
rate an
d n
e
twork d
e
lay
can
suffici
ently meet the
requi
rem
ents
of agricultural
monitorin
g
systems.
Ackn
o
w
l
e
dg
ments
This p
ape
r is sup
porte
d b
y
Beijing Nat
u
ral S
c
ien
c
e
Found
ation (4122
034
), the Nation
al
Scien
c
e a
nd
Tech
nolo
g
y Major Sp
eci
a
l
Proje
c
ts
of
China (201
0ZX
0104
5-0
01-0
04), the
Natio
nal
Scien
c
e a
n
d
Technolo
g
y Suppo
rt Progra
m
(
2011
BAD21B02
),
the Scie
nce Fou
ndatio
n for
Young Sci
e
n
t
ists of the
Nation
al Nat
u
ral S
c
ien
c
e
Found
ation of
China (61
1021
26) and
the
Youth Scien
c
e Found
ation
of Beijing Aca
demy
of Agricultural an
d Fo
rest
ry Scien
c
es.
Referen
ces
[1]
Hua
ng CM, Ku
HH, an
d Ku
ng
HY. Efficient
po
w
e
r- cons
u
m
ption-
base
d
l
oad-s
hari
ng to
pol
og
y co
ntro
l
protoco
l
for har
sh envir
onme
n
t
s in
w
i
r
e
l
e
ss sensor n
e
t
w
orks
.
Commu
n
ic
ati
ons
. 200
9; 3: 859-8
70.
[2]
Liu Yu
n-h
uai,
Z
hang Qia
n
, and N
i
LM. Opportun
i
t
y
-
b
ased to
pol
og
y control i
n
w
i
reless se
nso
r
net
w
o
rks.
IEEE Transactions
on Parallel
and Distributed S
ystem
s
. 20
10; 21: 405-
41
6.
[3]
W
u
M, Xu J,
T
ang X, et al.
T
op-
k monito
ring i
n
w
i
r
e
l
e
s
s
sensor
net
w
o
rk.
IEEE Transaction on
Kn
o
w
l
e
dg
e
and
D
a
ta
En
gi
nee
ri
ng
. 200
7; 19
: 962-97
6.
[4]
Z
h
i Ang
Eu
a, H
w
ee-P
i
nk T
anb, a
nd W
i
nst
on K
G. Se
ah,
Opportu
nistic
routin
g i
n
w
i
r
e
less s
ens
or
net
w
o
rks p
o
w
e
red b
y
amb
i
ent
energ
y
h
a
rves
ting.
Co
mp
uter
Netw
orks
. 2010; 54(17): 2
943
-296
6
[5]
Uster H and Li
n Hui. Integrat
ed topo
log
y
co
ntrol an
d routin
g in
w
i
re
less s
ensor n
e
t
w
orks
for prolong
e
d
net
w
o
rk lifetime.
Adhoc Netw
orks
. 2011; 9: 835-
851.
[6]
Park K
y
un
g-jo
on, Kim L
a
e
y
o
ung, a
nd
Hou
JC. A
daptiv
e p
h
y
s
i
cal
carri
er
sense
in to
pol
og
y-co
ntroll
e
d
w
i
reless net
w
o
rks.
IEEE
Transactions on Mobile Com
p
uting
. 2010; 9: 87-9
7
.
[7]
B Nefzi and Y
Q
Song. Performance a
nal
ys
is
and impr
ove
m
ent of zi
gbe
e
routing pr
otoc
ol.
7th IF
AC
Internatio
na
l
C
onfere
n
ce on F
i
eld
buses
&
Netw
orks in I
ndustri
a
l & E
m
b
e
d
ded
Systems
, France.
200
7: 351-
369.
[8]
Kandr
is D, T
s
agkar
opo
ul
o s
M
, and
Pol
i
tis
I, et al. En
erg
y
efficie
n
t an
d
perc
e
ive
d
Qo
S a
w
ar
e v
i
de
o
routin
g over
w
i
reless multim
e
d
ia se
nsor n
e
tw
o
r
ks.
Ad hoc
Netw
orks
. 201
1; 9: 591-6
07.
[9]
Han J,
Le
e H,
Park KR. R
e
m
o
te-contro
lla
bl
e a
nd E
ner
g
y
-
S
avin
g R
oom
Architecture
ba
sed
on Z
i
gb
e
e
C
o
mmu
ni
ca
tion
.
IEEE Transactions on Cons
umer Electronics
. 2009; 55: 2
64-2
68.
Evaluation Warning : The document was created with Spire.PDF for Python.