TELKOM
NIKA
, Vol.13, No
.2, June 20
15
, pp. 469 ~ 4
7
7
ISSN: 1693-6
930,
accredited
A
by DIKTI, De
cree No: 58/DIK
T
I/Kep/2013
DOI
:
10.12928/TELKOMNIKA.v13i2.961
469
Re
cei
v
ed Au
gust 18, 20
14
; Revi
sed
Jan
uar
y 5, 2015;
Acce
pted Fe
brua
ry 1, 201
5
Optimization of Sensor Network Topology in Deployed
in Inhomogeneous Lossy Media
Ron
y
Teguh
1
*, Hajime Igarashi
2
1
F
a
cult
y
of En
gin
eeri
ng, Res
earch C
enter o
f
Info
rmation Scienc
e for Peat
lan
d
Deve
lo
pm
ent, Univers
i
t
y
Pala
ngkar
a
y
a, Indon
esi
a
2
Graduate Sch
ool of Informati
on Scie
nce a
n
d
T
e
chnolo
g
y
,
Hokkai
do U
n
iv
ersit
y
, Ja
pan
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: ronn
yteg
uh@
gmail.c
o
m
A
b
st
r
a
ct
T
h
is pap
er pr
esents o
p
ti
mi
zation
of w
i
rele
ss sensor n
e
tw
ork (W
SN) topo
logy for fo
rest fire
detectio
n
. T
h
e
sensors f
o
r th
is pur
pose
are
dep
loy
ed i
n
f
o
rest, grassl
an
d an
d o
p
e
n
sp
ace, w
h
ich
ha
ve
different
attenu
ation
pr
operti
e
s
in
el
ectro
m
a
gnetic
w
a
ves.
F
o
r this r
eas
o
n
, routers
w
h
ic
h rec
e
ive
si
gn
als
from s
ens
ors a
nd s
e
n
d
the
m
to the
bas
e st
ation
must
be
dep
loye
d c
onsi
deri
ng th
ese
di
fferences. In
th
i
s
w
o
rk,
w
e
devel
op an
opti
m
i
z
at
ion
meth
od for
W
S
N topol
o
g
y base
d
on si
mul
a
ted an
ne
ali
n
g
consid
erin
g th
e
differenc
es i
n
the atte
nuati
o
n pr
operty. T
h
e ve
getati
on
d
a
ta ar
e tak
e
n
from L
ands
at
data.
Usin
g t
h
e
prese
n
t metho
d
, the nec
essa
ry nu
mber
of routers for
full
conn
ectio
n
of the se
nsors d
e
p
loy
ed i
n
div
e
r
s
e,
irregu
lar e
n
viro
nments can b
e
estimat
ed.
Ke
y
w
ords
: W
i
reless Se
nsor
Netw
ork, Electromag
netics,
W
a
ve Prop
ag
atio
n, Simulate
d Anne
ali
n
g
1. Introduc
tion
Wildfire
s cau
s
ed by
lightin
g,
spo
n
taneo
us
comb
ustio
n
, human
act
i
vities and
so
on are
seri
ou
s p
r
obl
ems e
s
p
e
ci
all
y
in North A
m
eri
c
a, Si
be
ria and Ind
o
n
e
sia.
Wildfire
s can give
rise to
signifi
cant he
alth, econ
omi
c
and e
n
viron
m
ental dam
a
ges. In Kalim
antan an
d Sumatra, a few ten
thousand
wild
fire event
s are dete
c
t
ed a
year by M
O
DIS [1]. Becau
s
e initial
dete
c
tion of
wildfires
is of im
po
rta
n
ce
for
effect
ive extinction
, a det
e
c
tion
system f
o
r In
done
sia
n
wil
d
fires ha
s
be
en
develop
ed [2]
-
[4]. One
of th
e mo
st p
r
omi
s
ing
dete
c
ti
o
n
sy
stem
s i
s
t
hat ba
se
d o
n
wirel
e
ss
se
nsor
netwo
rk
s
(
W
S
Ns
). In the
WS
N det
e
c
t
i
on, many
se
nso
r
s a
r
e
de
ployed i
n
the
target
area
to
measure envi
r
onm
ental da
ta su
ch a
s
te
mperatur
e
an
d humidity. The mea
s
u
r
ed
data are the
n
aggregate
d
to the base st
ation throu
gh
wirel
e
ss com
m
unication.
This sy
stem can reali
z
e fast and dire
ct detec
tio
n
of wildfire
s. The
WSNs, ho
wever have
some
p
r
obl
e
m
s. Th
e m
a
j
o
r
pro
b
lem
i
s
relatively
short
lifetime
of
se
ns
ors which
are usu
a
lly
driven by batt
e
rie
s
. Beca
use it needs
gre
a
t efforts
to make frequ
ent repla
c
e
m
ent
of the batterie
s
in se
nsors d
eployed in
wide area
s, prolong
ation of
the lifetime
is strongly
re
quire
d. For t
h
is
rea
s
on,
there have
be
en
many
studi
es fo
r
ex
tens
ion of
WSN lifetimes [5]. In the
LEACH
comm
uni
cati
on proto
c
ol [
6
], sensors a
u
tonomo
u
sly
con
s
titute clu
s
ters ea
ch of
which has o
n
e
clu
s
ter he
ad.
The
data
me
asu
r
ed
by
se
nso
r
s a
r
e
gat
here
d
by the
clu
s
ter he
ad
s and
tra
n
sfe
r
red
to the ba
se
station. A
clu
s
ter
hea
d
is dyn
a
mi
c
a
lly
sele
ct
ed
f
r
om
sen
s
o
r
s i
n
t
he
cl
ust
e
r
con
s
id
erin
g e
nergy loa
d
ba
lance.
In the Zi
gbe
e
t
echn
ology [7
], WSNs
are
comp
osed
of
se
nsors,
rou
t
ers an
d b
a
se statio
n.
The se
nsors sen
d
the me
asu
r
ed d
a
ta to the nearest
router o
r
directly to the base
station. The
route
r
s
send
the aggregat
ed data to
the base
statio
n. Deployme
nts
of the se
nso
r
s a
nd ro
uters
have sig
n
ifica
n
t influence on lifetime, coverag
e
and
co
nne
ctivity in Z
i
gbee
-ba
s
e
d
WSN
system
s.
The se
nsors
woul
d be dep
loyed to maximize thei
r co
verage [8]. On the other h
and, the rout
er
positio
n
sho
u
l
d be
dete
r
m
i
ned to
maxi
mize th
e lifet
ime an
d
con
nectivity. The
autho
rs hav
e
prop
osed op
timization m
e
thod of po
sition of
the
routers to
maximize the lifetime and
c
o
nnec
tivity
of WSNs
for fores
t
fire detec
t
i
on
co
nsiderin
g differences i
n
the
elevation u
s
ing
geneti
c
algo
ri
thm [4].
In this
wo
rk,
we
develop
a
n
optimi
z
atio
n metho
d
b
a
s
ed
on
sim
u
l
a
ted a
nneali
n
g (SA) for
deployme
nt
of route
r
s of
Zigbe
e-ba
se
d WS
N
wh
o
s
e
wo
rki
ng f
r
eque
ncy i
s
i
n
UHF
ban
d. In
particular, we consider the WSNs locat
ed in
inhomogeneous lossy media
such as forest and
gra
ssl
and,
which h
a
ve no
t been di
scussed in
othe
r
works. We
consi
der th
e d
i
fferences i
n
the
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 2, June 20
15 : 469 – 47
7
470
prop
agatio
n cha
r
a
c
teri
stics in the me
dia into a
c
count. This p
aper i
s
o
r
ga
nize
d as foll
ows:
prop
agatio
n of
electroma
gnetic (EM) wave
s
will
b
e
discu
s
sed
in the next section. Th
en
the
optimizatio
n
method
will b
e
de
scribe
d i
n
the third
se
ction. The
co
nclu
sio
n
will
be follo
wed b
y
the
nume
r
ical re
sults rep
o
rte
d
in fourth secti
on.
2.
Propaga
tion of EM Wav
es in Forest
2.1.
Propaga
tion Modes in Fo
rest
There are th
ree EM
wav
e
co
ntributio
ns to
the fiel
d [9],[10]: geometri
c opti
c
al wave
s
prop
agate
di
rectly or reflectively
from th
e source
to th
e si
nk thro
ug
h the t
r
ee
tru
n
ks a
nd
ca
no
py.
The
sky
wav
e
s h
a
ve lo
ng
triang
ular pa
th who
s
e
vert
exes a
r
e th
e
sou
r
ce, si
nk
and io
no
sph
e
r
e.
More
over, th
e lateral
wav
e
s
pro
pag
ate
along
t
he
canopy-air i
n
terface. We
can di
scard t
h
e
se
con
d
wave
s
fo
r WSNs becau
se
th
ey
use UHF
wa
ves whi
c
h do
not
h
a
ve reflection
from
the
iono
sph
e
re.
T
he first a
nd th
ird
wave
s va
ry with di
stan
ces
as
and
res
p
ec
tively.
Hen
c
e it dep
end
s on the
distan
ce
a
nd
the attenuatio
n con
s
tant
of the medium
which wave i
s
domina
n
t. A full wave an
al
ysis ba
se
d on
four layer
m
odel of the forest
con
c
lud
e
s that the former
is d
o
mina
nt
above 1
00
M
H
z if the
co
mmuni
cation
distan
ce i
s
shorte
r tha
n
3
km [1
1]. In
the
WSNs
for fores
t
fire detec
t
ion, the
c
o
mmunicati
on di
stan
ce of
the
sen
s
o
r
s
a
nd routers wo
uld be
s
u
ffic
i
ently s
h
orter than 3
k
m
. For this
reas
on,
we
consi
der
only
the first waves in thi
s
stu
d
y.
More
over, for simplicity, we only con
s
id
er the direct
wave
s.
2.2.
Electromag
n
etic Wav
es in Absor
b
ing
Media
Let us con
s
id
er ele
c
trom
ag
netic waves i
n
inhomo
gen
eou
s lossy di
electri
c
me
di
a, which
are gove
r
n
e
d
by the Maxwell equation
s
(1a)
,
(1b)
whe
r
e
are ele
c
tri
c
field, magnetic field, a
ngula
r
freq
ue
ncy and p
e
rm
eability in vacuum.
More
over
is the com
p
lex p
e
rmittivity defined by
(2)
whe
r
e
are pe
rmittivity in
vacuum, relative permittivity
and cond
ucti
vity, and
j
denotes the
imagina
ry uni
t.
We ma
ke he
re followin
g
assumptio
n
s o
n
the dielectri
c
prope
rty of the mediu
m
:
(a) T
he loss i
s
domin
ant so that
.
(b) The relative permittivity
is uniform while conductivity
varies
with positio
n.
(c) T
he
spatia
l scale
of
is
sufficiently sm
aller th
an th
e
wavele
ngth
o
f
UHF
wave
. That i
s
,
a
s
s
u
ming
th
at
va
r
i
es
s
i
nus
o
i
da
lly,
the magnitude o
f
is expressed
by
which i
s
sufficiently sm
aller than
.
No
w intro
d
u
c
i
ng vecto
r
pot
ential, sati
sfying
, which
ob
eys the L
o
re
ntz ga
uge,
the vector
He
lmholtz eq
uat
ion
(3)
can b
e
derive
d
from (1
), where
is the complex wave
numbe
r defin
ed by
.
(4)
Due to the ass
u
mption (a),
can be a
ppro
x
imated as
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Optim
i
zation of Sensor Net
w
ork Topology in
Deployed in Inhom
ogeneous .... (Rony Teguh)
471
(5)
For sim
p
licity
,
we here
co
nsid
er a wave radiate
d
fro
m
a dipole in
z
dire
ction p
e
rpe
ndi
cula
r to
grou
nd, whi
c
h is governed
by the one-di
mensi
onal
scalar Helmh
o
ltz equ
ation gi
ven by
(6)
whe
r
e
. It can
be sho
w
n th
at the dampe
d wave soluti
on
(7)
satisfie
s (6) unde
r
the assumptio
n
s
(a
)-(c).
No
te th
at (7
) is an
e
x
act solution
to (6
) when
is
con
s
tant. It is therefore co
nclu
ded that
the vector p
o
tential is given
by
(8)
It can al
so
find that
and
also have the
spatial
attenuatio
n of
the
form
.
3. Optim
i
za
tion
Method
3.1.
Formation o
f
Wireless Se
nsor Net
w
o
r
ks
In the optimization, we a
dopt the followi
n
g
assum
p
tions for de
terminatio
n of WSN
topology for simplicity.
(a)
The se
nsors,
routers and
base station
have a com
m
on thre
sh
old in electri
c
field
above
whi
c
h they co
mmuni
cate wi
th others.
(b)
The m
u
lti-ho
p
tran
smi
ssi
on
is
available
for
the
routers but n
o
t for sensors.The
m
agnitud
e
of
electri
c
field
whi
c
h is g
e
n
e
rated by no
de
and re
ceived by nod
e
, and vice
versa, i
s
expre
s
sed b
y
, where
and
den
ote a
strai
ght
line
con
n
e
c
ting th
ese p
o
ints a
n
d
their dista
n
c
e.
3.2.
Wireless Se
nsor Net
w
o
r
ks Deploy
ment Algo
rith
m
The WS
N top
o
logy is dete
r
mined from th
e followin
g
algorithm.
(1)
The tentative layer level, sa
y -1, is given for all the rout
ers.
(
2
) T
h
e
se
ns
or
and its
nea
re
st nod
e
which is eith
er a
router o
r
b
a
se
station a
r
e
conne
cted if
.
(3) The
route
r
a
nd its ne
are
s
t
ba
se
station
a
r
e
co
nne
ct
ed if
. The l
a
yer level
of t
h
e
con
n
e
c
ted
ro
uter i
s
set to
1 a
nd th
e
current laye
r l
e
vel
is
set t
o
1. T
hen
th
e follo
wing
pro
c
ed
ure is repeate
d
un
til there are
no r
oute
r
s which
can be
con
n
e
c
ted to the other
route
r
s.
(4) The
ro
uter
of level -1 and its nearest ro
u
t
er
of level
are co
nne
cted
if
. If router
is co
nne
cted,
then its layer is set to
.
(5)
Return to (4).
3.3.
Optimiza
tion
Using Simulated
Anne
aling
The obj
ectiv
e
function to
be maximized is
ju
st e
qual to the
numbe
r of conne
cted
s
e
ns
or
s
,
N
.
For optimi
z
at
ion of
WS
N
topology,
we
empl
oy the
sim
p
le
sim
u
lated
anne
ali
ng
who
s
e al
gorit
hm is de
scrib
ed belo
w
.
(a)
We
set
the
values of
,
, initial tem
p
erature
and maximum
iteration coun
t
.
s
e
ns
ors
a
n
d
route
r
s a
r
e
rand
omly dep
loyed in the target field
.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 2, June 20
15 : 469 – 47
7
472
(b)
One router is ran
d
o
m
ly cho
s
en
and its
positio
n
is
modified to
, where
are random numb
e
rs and
is a given co
nsta
nt. If
is outside of
, this
modification is
disc
arded. The topology of both WSNs
is
det
ermined
usin
g the
alg
o
rithm
de
scri
bed i
n
3.2,
a
nd the
nu
mb
ers of
co
nne
cted
se
nsors,
ar
e
comp
uted.
(c) If
is po
si
tive, then this mo
difica
tion is a
c
cepted. Othe
rwi
s
e,
we
comp
ut
e
. If
, then the modif
i
cation i
s
accepted (reje
c
te
d).
(d)
The temp
erat
ure i
s
de
crea
sed
by
. If iteration count i
s
smalle
r tha
n
, then retu
rn t
o
(b).
4.
Analy
s
is of Experimenta
l
Results
4.1.
Artifici
al Tes
t
Problem
We ap
ply the present opti
m
ization m
e
thod to an
arti
ficial test pro
b
lem, whe
r
e
an are
a
of
highly attenu
ation with
is lo
cated n
e
a
r the b
a
se
stat
ion. In al
l the optimization
mentione
d
belo
w
, the o
p
timization
para
m
et
ers are
set a
s
follo
ws:
. It is expecte
d that the WS
N
topology woul
d be form
ed
avoiding
the
attenuation a
r
ea. Figure 1
sho
w
s the op
timization result.
We can
see
that all the sens
ors are succe
ssfully conne
cted
to t
heir p
a
re
nt node
s and th
e
i
r
comm
uni
cati
on route
s
det
our a
r
ou
nd th
e attenuation
area, a
s
expe
cted.
Figure 1. Artificial test probl
em with atten
uation area with
near ba
se
station,
RN
=1
2
.
BS, RN SN
re
pre
s
e
n
tBase Statio
n, Router a
n
d
Senso
r
nod
e
s
4.2.
Optimiza
tion
of WSNs for
assumed a
t
tenua
tion co
nsta
nts
In this simulat
i
on, we ch
oo
se the tropical
ra
in fore
st in Central Kalim
antan, Indone
sia for
a ca
se
study of the simulat
i
on.
The ma
p
of the locatio
n
is obtai
ned
from Lan
dsat image
s. The
r
e
are forest, grasslan
d and free spa
c
e. We use
the sy
stem for automated geo
sci
entific analyses
(SAGA) [13] to extract the information
about
fore
st and vegetati
on from Lan
dsat imag
es.
We
also
perfo
rm
optimizatio
n
of a free
spa
c
e
with the same a
r
ea fo
r
comp
ari
s
o
n
. In these mod
e
ls,
the sen
s
o
r
s
and ro
uter n
ode
s are pla
c
ed rand
om
ly and the latter positio
ns a
r
e optimize
d
. We
use the same random
seed for both optimization
s.We utilize a linear inte
rpolation based method
to estimate th
e informatio
n about
fore
st, vegetation an
d free-sp
ace.
To evaluate the ele
c
tric fie
l
d received b
y
the nodes,
whi
c
h dep
en
ds on
defined in (5),
we n
eed th
e
values
of the
homog
eni
zed
permittivity
and el
ectri
c
condu
ctivity
for forest a
nd
gra
ssl
and. A
c
cordi
ng to [
7
], their rang
es a
r
e
In [8]
,
is
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Optim
i
zation of Sensor Net
w
ork Topology in
Deployed in Inhom
ogeneous .... (Rony Teguh)
473
assume
d to b
e
unity. Hen
c
e
wo
uld ran
g
e
from about 0.2
to
. T
h
u
s
w
e
ass
u
me
he
r
e
that
for forest.
In [9], the rec
e
ived power
is assu
me to be of the form
in dBm, and
the value
s
of
, determi
ne
d by expe
rim
ents, a
r
e
co
mpar
ed fo
r fore
st an
d g
r
asslan
d. At 2.45
GHz, the
re
sultant value i
s
2.8
9
for pin
e
fore
st,
an
d
it rang
es fro
m
3.55 to
4.1
3
for
gra
s
sla
nd in
long
com
m
un
ication. Sin
c
e
their de
cay
model i
s
diffe
rent
from our expone
ntial model, we
ca
nnot
evaluate
from these
re
su
lts for g
r
a
ssl
and. We a
s
su
me that the
decay for g
r
asslan
d is two
times
stro
ng
er tha
n
that
for fores
t, that is
(Fi
g
ure
2).
Note
that the
prese
n
t
optimizatio
n can be exe
c
ut
ed for arbitra
r
y values of
.
Figure 2. Sensing field in g
r
assla
nd an
d forest
In WSNs
co
nsid
er
a sen
s
or net
work
to det
ect fo
rest fire
s. Th
e se
nsor
no
des
are
assume
d to
be rand
omly
depl
oyed in
the fo
re
st.
More
over, it
assume
s th
a
t
the sen
s
or
and
route
r
no
de
s have
the co
mmuni
cation distan
ce
R
0
i
n
free
sp
ace. It is cle
a
r th
at the num
be
r of
the
se
nsors whi
c
h can communi
cate with
the
n
e
a
r
est p
a
rent n
ode d
epen
ds on the rout
er
deployme
nt. The se
nsor i
s
judged to be
con
n
e
c
ted if the co
ndition.
(9)
is satisfied,
whe
r
e
R
is t
he di
stan
ce f
r
om the
se
nsor to
the
nea
rest
route
r
in
cludi
ng the
b
a
se
station. We o
p
timize
th
e
router po
sition
s
to
m
a
ximize the
numb
e
r of conn
ecte
d sen
s
ors
usi
n
g
the simulate
d
anneali
ng (S
A). The optim
ization p
r
obl
e
m
is defined
by]
(10)
whe
r
e
N
c
den
otes the num
ber of the con
necte
d se
nso
r
s.
Figure 3 sh
ows the optimiz
ed
WSNs when RN=5. We fi
nd that the num
bers of
con
n
e
c
ted se
nso
r
s a
r
e 2
6
and 22 for the free spa
c
e and inh
o
m
ogen
eou
s area com
p
o
s
e
d
of
forest, grassl
and and fre
e
spa
c
e. Be
cau
s
e of
the
strong
er attenuatio
ns in
the forest and
gra
ssl
and,
th
e nu
mbe
r
s of
co
nne
cted
sensors
ar
e
re
duced fo
r th
e
latter
ca
se.
More
over,
we
find
the optimize
d
network topo
logy is differe
nt from each other.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 2, June 20
15 : 469 – 47
7
474
Figure 3. Optimized
re
sult
s for RN=5, (a
)
free-sp
ace e
n
vironm
ent, (b) Inhom
oge
neou
s field
comp
osed of forest, grassl
and an
d free
sp
ace
Figure 4
sh
ows the
con
v
ergen
ce
hi
stories of
S
A
f
o
r
bot
h
ca
se
s.
After the initial
fluctuation
s
d
ue to ra
ndo
m sea
r
ch at
high temp
era
t
ure, the val
ues
of the o
b
jective fun
c
t
i
on
(num
ber of conne
cted
sen
s
ors) almo
st monoton
ou
sl
y increa
se a
n
d
conve
r
ge to
the final values.
Figure 4. Optimization hi
sto
r
ies fo
r RN=5
Figure 5 an
d 6 sho
w
the co
rrespo
nding re
sults for RN=1
2. All the sen
s
ors a
r
e
connected to the network for t
he free
space, while t
here
are
st
ill 3
unconnected
sensors in
the
inhomo
gen
eo
us fiel
d. To
e
v
aluate the
n
e
ce
ssary
nu
mber of
rout
ers o
r
full
co
nne
ction
s
of
the
sen
s
o
r
s to
WSNs, we perf
o
rm optimi
z
at
ions
cha
ngi
n
g
the numbe
r of routers an
d rand
om see
d
s.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Optim
i
zation of Sensor Net
w
ork Topology in
Deployed in Inhom
ogeneous .... (Rony Teguh)
475
Figure 5. Optimized top
o
lo
gy for inhomo
gene
ou
s RN=12, (a) Free
spa
c
e, (b
) Inh
o
moge
neo
us
field comp
osed of fore
st, grassla
nd an
d free sp
ace
Figure 6. Optimization hi
sto
r
ies fo
r RN=1
2
The re
sult
s are sho
w
n in Fi
gure
7 and Fi
gure
8, whe
r
e
we can con
c
l
ude that we n
eed at
least
8
and
1
1
routers are
need
ed fo
r th
e fre
e
sp
a
c
e
and i
nho
mog
eneo
us field
resp
ectively. T
h
is
numbe
r d
epe
nds on th
e v
egetation.
We can eval
uat
e the n
e
cessary nu
mbe
r
o
f
the ro
uters
for
inhomo
gen
eo
us field with a
r
bitra
r
y distri
b
u
tion.
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 16
93-6
930
TELKOM
NIKA
Vol. 13, No. 2, June 20
15 : 469 – 47
7
476
Figure 7. Nu
mber of conn
ected
sen
s
o
r
s in free
spa
c
e field
Figure 8. Nu
mber of conn
ected
sen
s
o
r
s In
Inhomog
eneo
us field
comp
osed of forest,
gra
ssl
and a
n
d
free space
5. Conclu
sions
We have
prese
n
ted opti
m
ization
of WSNs
pla
c
e
d
in
inho
mo
gene
ou
s lo
ssy field
comp
osed
of
rai
n
fo
re
st, gra
ssl
and
an
d fre
e
spa
c
e.
We
o
p
timize
the
route
r
d
eployment
u
s
ing
the SA. The
wave
atten
uation in
the
inhom
oge
ne
ous fields is take
n into
accou
n
t in t
he
optimizatio
n.
We
ca
n eval
uate the
ne
cessary
num
b
e
r of th
e rout
ers for full conne
ction
of
the
s
e
ns
or
s
to
WSN
.
We have a
p
p
lied the pre
s
ent meth
od
to the arti
fici
al test field with re
ctang
u
l
ar lossy
area
ne
ar th
e ba
se
statio
n and
real in
homog
ene
ou
s field
co
mp
ose
d
of
rain
forest, g
r
a
s
sl
and
and free
spa
c
e in
Kalima
n
tan. The
n
e
twork
ha
s
been fo
rme
d
avoiding th
e lossy a
r
ea
as
expecte
d.
For the latte
r pro
b
lem,
we have fou
nd t
hat we
need at lea
s
t 11 routers for full
connection of
sensors. For future
work, we will eval
uate the relia
bi
lity of the present method
b
y
measuri
ng th
e perfo
rman
ce of the optimized
WSN in real field.
Evaluation Warning : The document was created with Spire.PDF for Python.
TELKOM
NIKA
ISSN:
1693-6
930
Optim
i
zation of Sensor Net
w
ork Topology in
Deployed in Inhom
ogeneous .... (Rony Teguh)
477
Referen
ces
[1]
N Yulianti, H Hay
a
sak
a
, Usup.
Recent forest
and Peat Fire Trends
in Indones
ia, the Latest Decade
b
y
MODIS Hot
s
pot Data.
Glo
bal Env
i
ron
m
e
n
tal Res
earch
,
AIRIES.
2012; 16(1): 10
5-1
1
6
.
[2]
R T
eguh,
T
H
onma, A Usop
, H Shin, H Igarash
i
.
Detecti
on an
d Verific
a
tion of Potent
ial Pe
at F
i
re
Using W
i
r
e
less
Sensor N
e
tw
ork and UAV
. Internati
o
n
a
l C
onfere
n
ce o
n
Information te
c
hno
log
y
an
d
Electrical E
ngi
neer
ing. Ind
o
n
e
sia. 20
12: 6-1
2
.
[3]
Yoon
I, No
h
DK, Le
e D, T
egu
h R,
Ho
n
m
a T
,
Shin H.
Rel
i
a
b
le
Wildf
i
re Mo
nitor
i
ng
w
i
th Sparse
ly
Depl
oye
d
W
i
rel
e
ss Sens
or Ne
tw
orks
. IEEE 2
6
th Int. Conf. A
d
v. Inf.
Net
w
.
Appl. Fukuoka.
2012: 460-
466.
[4]
R T
eguh,
R M
u
rakami,
R I
g
a
r
ashi
H.
Opti
mi
z
a
ti
on
of
Rout
er D
epl
oy
me
nt for Se
nsor
Ne
tw
orks Usi
n
g
Genetic Al
gori
t
hm
. 13th Int
e
rnati
ona
l Co
nferenc
e on
Artifi
cial Intel
l
i
genc
e an
d S
o
ft Computi
n
g
ICAISC. Polan
d
. 2014.
[5]
Sohra
b
y
K, Mi
noli
D, Z
n
ati T
.
W
i
reless s
e
n
s
or n
e
tw
ork techno
logy,
prot
ocol,
and
a
ppl
i
c
ation.
Wile
y
intersci
ence. 2
007.
[6]
Heinz
e
lm
an W
R
, Chan
drak
a
s
an A, Bal
a
kri
s
hna
n H.
Ene
r
gy-Efficient C
o
mmunic
a
tio
n
Protocol fo
r
W
i
reless Micro
s
ensor N
e
tw
orks
. Sy
st
em Scienc
es, 200
0. Procee
din
g
s o
f
the 33rd An
nua
l Ha
w
a
i
i
Internatio
na
l C
onfere
n
ce. 20
0
0
.
[7] http://
w
w
w
.
zigbee.org/
[8]
Vinh T
Q,
T
a
ku
mi M.
An Algor
ithm for se
nsin
g covera
ge pr
o
b
le
m i
n
w
i
reles
s
sensor netw
o
rks
. Sarnoff
S
y
mp
osi
u
m IEEE. 2008: 1-5.
[9]
T
T
a
mir. On R
adi
o-
w
a
v
e
Pro
pag
atio
n in F
o
rest Environm
ent.
IEEE Transactions on
Antennas an
d
Propa
gati
o
n
in
clud
es theor
eti
c
al an
d e
x
p
e
ri
mental a
d
va
nc
es in ante
n
n
a
s
.
1967; 1
5
(6):
8
06–
81
7.
[10]
T
T
a
mir. Radio
w
a
ve Pr
op
ag
ation
alo
ng m
i
xe
d p
a
ths i
n
F
o
rest Envir
o
n
m
ent.
IEEE Transactions
on
Antenn
as an
d Propa
gati
o
n
in
clud
es theoreti
c
al an
d e
x
per
i
m
ent
al a
d
vanc
es in ante
nnas
. 1977; 25(4
)
:
471
–4
77.
[11]
Le-
w
e
i, T
a
t-Soon Y
,
Pa
ng
-Shy
an
K
,
Mook-
S
eng
L.
R
adi
o
w
a
ve prop
a
gatio
n
a
l
o
ng mix
ed pat
h
s
throug
h a four-l
ayere
d
mod
e
l of rain forest: an ana
lytic ap
pr
oach.
19
98; 16
(7): 1098-
11
11
.
[12]
Jose A, et al. Peer to pe
er
w
i
reless
prop
ag
ation me
asur
e
m
ents and
pat
h-loss mo
del
in
g in ve
getate
d
envir
onme
n
ts,
IEEE trans. Antennas a
nd pr
o
pag
atio
n
, 201
3
;
61(6): 330
2-3
311.
[13] http://
w
w
w
.
sa
g
a
-gis.or
g
/en/in
de
x.html
Evaluation Warning : The document was created with Spire.PDF for Python.