Indonesi
an
Journa
l
of El
ect
ri
cal Engineer
ing
a
nd
Comp
ut
er
Scie
nce
Vo
l.
24
,
No.
1
,
Octo
be
r
20
21
,
pp.
236
~
24
4
IS
S
N: 25
02
-
4752, DO
I: 10
.11
591/ijeecs
.v
2
4
.i
1
.
pp
23
6
-
24
4
236
Journ
al h
om
e
page
:
http:
//
ij
eecs.i
aesc
or
e.c
om
Improve
d finge
rp
rintin
g perfo
rmance in i
nd
oo
r positi
oning by
re
du
cin
g du
ra
ti
on
of
th
e training
p
hase
p
rocess
An
dik
a Mu
haram
,
Abdi
W
ahab
, Mudrik
Alaydrus
Depa
rtment
o
f
E
le
c
tri
c
al E
ngin
eering,
Univ
ersitas Merc
u
Buana,
J
aka
rt
a,
Indone
si
a
Art
ic
le
In
f
o
ABSTR
A
CT
Art
ic
le
history:
Re
cei
ved
Oct
27
, 202
0
Re
vised Jul
30
,
2021
Accepte
d
A
u
g
7
, 2
021
W
ire
le
ss
sensor
net
work
(W
SN)
ca
n
be
used
as
a
soluti
on
to
find
out
the
positi
on
of
an
ob
je
c
t
tha
t
c
anno
t
be
rea
ch
ed
b
y
globa
l
positi
on
ing
s
y
stem
(
GPS
),
for
exa
m
ple
to
f
ind
out
t
he
positi
on
of
o
bje
c
ts
in
a
room
known
as
Indoor
Pos
it
ioning.
One
m
et
hod
in
in
door
posit
ioni
ng
th
at
c
an
be
used
is
finge
rprin
ti
ng.
I
nside
the
r
e
ar
e
t
wo
m
ai
n
work
phase
s,
namel
y
tra
ini
ng
and
positi
oning.
Th
e
tra
ini
ng
phase
i
s
the
proc
ess
of
col
lecti
ng
re
ceive
d
signal
strengt
h
ind
ication
(RSS
I
)
dat
a
l
eve
ls
from
e
ac
h
sensor
Node
re
fer
e
nc
e
th
a
t
will
be
used
as
a
ref
er
enc
e
va
lu
e
for
the
positi
o
ning
phase
.
The
m
ore
sensor
Nodes
used,
the
longe
r
the
proc
e
ss
ing
ti
m
e
nee
ded
in
the
tra
ini
ng
phase
.
Thi
s
rese
arc
h
foc
uss
e
d
on
the
dura
t
io
n
of
the
training
phase
,
the
impl
ementa
t
ion
of
which
ar
e
use
d
4
sensor
Nodes,
namel
y
Z
igbee
(IE
E
E
802.
15
.
4
protoc
ol)
arr
ange
d
ac
co
rdi
ng
to
m
esh
net
work
topol
og
y
,
one
as
Node
X
(positi
oning
ta
rge
t)
and
3
as
ref
ere
n
ce
Nod
es.
The
re
ar
e
two
m
et
hods
used
in
the
tra
in
ing
phase
,
n
amel
y
f
ixe
d
t
arg
e
t
par
a
m
et
er
(FTP)
an
d
m
ov
ing
ta
rge
t
par
amet
er
(MTP).
MTP
too
k
5
sec
onds
fast
er
tha
n
FTP
in
t
e
rm
s
of
the
dura
t
i
on
of
RS
SI
dat
a
colle
ct
ion
fr
om
ea
ch
ref
ere
n
ce
Node
.
Ke
yw
or
d
s
:
Fing
e
r
pr
inti
ng
Ind
oor po
sit
io
ni
ng
RSSI
Trainin
g p
hase
W
i
reless se
nso
r netw
ork
Zigb
ee
This
is an
open
acc
ess arti
cl
e
un
der
the
CC
B
Y
-
SA
l
ic
ense
.
Corres
pond
in
g
Aut
h
or
:
Mudrik
Alay
dr
us
Dep
a
rtm
ent o
f El
ect
rical
En
gi
neer
i
ng
,
Un
i
ve
rsita
s Merc
u
B
uan
a
Me
ru
ya
Selat
a
n No.1
, Kem
ban
ga
n
,
Me
ru
ya
Selat
a
n,
J
aka
rta Bara
t, DKI Ja
ka
rta
11650,
Ind
on
e
sia
Em
a
il
:
m
ud
rik
al
ay
dr
us@m
ercu
bua
na.
ac.i
d
1.
INTROD
U
CTION
W
i
reless
netw
ork
te
le
com
m
un
ic
at
ion
is
one
of
t
he
c
ho
ic
es
of
e
ve
ry
te
ch
nolo
gy
us
e
r
who
ne
eds
e
ase
in
com
m
un
ic
ation
,
bo
t
h
in
te
r
m
s
of
inf
rastr
uc
ture,
the
i
ns
ta
ll
at
ion
process
and
it
s
pr
act
ic
al
us
e
[1]
.
W
i
r
el
ess
netw
ork
te
le
c
omm
un
ic
at
ion
s
are
i
ncr
easi
ng
ly
dev
el
op
i
ng
w
hich
are
cha
racteri
zed
by
a
c
om
bin
at
ion
of
wireless
te
le
co
m
m
un
ic
at
ion
s
netw
orks
with
m
ic
ro
el
ect
ronic
t
echnol
og
y
a
s
sens
or
Node
s
kn
own
as
wi
reles
s
sens
or
netw
orks
(
WSN)
.
WSN
ca
n
be
us
e
d
as
a
so
l
ution
to
fin
d
out
the
po
sit
io
n
of
an
obj
ect
that
ca
nnot
be
reache
d
by
g
lo
bal
posit
io
ning
syst
e
m
(G
P
S
),
f
or
e
xam
ple
to
fin
d
ou
t
t
he
po
sit
io
n
of
ob
je
ct
s
in
a
r
oo
m
cal
l
ed
indoor
posit
io
ning
[2]
.
I
n
in
door
po
sit
io
ni
ng,
eac
h
se
nso
r
Node
i
ns
ta
ll
ed
is
e
xpect
ed
t
o
pro
vid
e
i
nfo
rm
ation
need
e
d
by
othe
r
sens
ors
[
3]
,
so
that
with
a
certai
n
m
et
ho
d
can
infe
r
the
po
sit
io
n
of
an
obj
ect
[
4]
.
The
sens
or
Node
us
ed
sho
uld
hav
e
a
sm
al
l
ph
ysi
cal
siz
e,
low
powe
r
co
ns
um
ption,
lim
it
ed
processi
ng
power,
s
hort
r
ange
com
m
un
ic
at
ion
,
a
nd
hav
e
a
sm
a
ll
a
m
ou
nt
of
m
e
m
or
y
storag
e
,
su
c
h
as
that
of
Zi
gb
ee
or
wireless
loc
al
area
netw
orks
[
5]
-
[7
]
.
I
niti
al
ly
the
us
e o
f
Zi
gb
ee was
inte
nd
e
d
f
or
v
ari
ous
ty
pe
s
of
a
uto
m
at
ion
syst
em
app
li
cat
ion
s
du
e
to
e
ne
rg
y s
avin
g
a
nd sec
uri
ty
f
act
or
s
that
are q
ualifi
ed
[
8]
.
The
o
ne
m
et
hod
that
ca
n
be
us
ed
f
or
in
door
po
sit
io
ning
on
WSN
with
Zig
bee
is
fi
ng
e
rpr
inti
ng
[9
]
,
[10]
.
F
ro
m
sever
al
oth
e
r
m
eth
ods
t
hat
can
be
use
d
i
n
in
door
po
sit
io
ning
ba
sed
on
rec
ei
ved
si
gn
al
st
rengt
h
ind
ic
at
io
n
(RS
S
I
)
s
uc
h
as
tril
at
erati
on
an
d
t
riangulat
ion,
f
ing
e
rprintin
g
is
a
m
et
ho
d
th
at
is
widely
a
dopte
d
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng
&
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Impr
oved fi
ng
e
rp
rinti
ng
perf
orma
nce
in
i
ndoor
posit
ion
in
g by re
duci
ng
dura
ti
on
of the…
(
And
ik
a M
uharam
)
237
because
of
it
s h
ig
h
le
vel o
f
a
ccur
acy
[11]
. Th
e
fi
nger
pr
i
nting
m
et
ho
d
c
onta
ins
tw
o
m
ain
w
ork
phases,
nam
el
y
trai
ning
(off
li
ne
-
phase
)
[2]
and
po
sit
io
ning
(o
nline
-
ph
ase
)
[12],
[13].
I
n
the
first
ph
a
se
s
(trainin
g)
is
the
process
of
coll
ect
ing
data
s
(e
xam
ple:
RSSI)
fr
om
each
sen
so
r
Node
that
is
colle
ct
ed
in
a
database
w
hic
h
can
la
te
r
be
us
e
d
as
ref
e
ren
ce
da
ta
to
help
in
fer
the
posit
io
n
of
an
obj
ec
t
carried
ou
t
i
n
the
sec
ond
ph
a
se
(positi
on
i
ng
)
.
The
m
or
e
N
od
es
are
us
e
d,
in
wh
ic
h
the
tim
e
nee
ded
duri
ng
trai
ni
ng
pro
c
ess
is
m
uch
lo
ng
e
r
wh
ic
h wil
l reduce the
p
e
rfo
r
m
ance of th
e
s
yst
e
m
.
Ur
a
dzins
ki
et
al
.
[
14]
di
d
resea
rch
ed
on
a
dvance
d
i
ndoor
posit
io
ni
ng
us
i
ng
Zig
bee
wireless
te
chnolo
gy
with
the
finger
pri
nting
m
et
ho
d.
The
pur
po
se
of
his
re
searc
h
is
to
raise
t
he
le
vel
of
ac
cur
ac
y
durin
g
t
he
pos
it
ion
ing
ph
ase
.
To
achie
ve
t
hi
s
goal
,
Ur
a
dzi
ns
ki
un
der
ta
ke
s
tw
o
sta
ges
of
work,
t
he
fir
st
i
s
filt
ering
out
in
te
rf
ere
nce
si
gnal
s
in
the
m
easur
em
ent
area
s
o
that
r
efe
ren
c
e
data
in
t
he
da
ta
base
can
be
m
or
e
accurate.
T
he
seco
nd
ste
p
is
to
m
ake
a
com
bin
at
ion
of
the
weig
hte
d
ne
arest
neig
hbors
al
gorithm
with
the
Ba
ye
s
al
go
rith
m
.
The
sens
or
Nodes
us
e
d
a
re
4
Zig
bee
as
ref
ere
nce
N
odes
with
im
ple
m
entat
ion
hardw
a
re
.
The
m
easur
em
ent
area
is
42.5x4.96
m
with
t
he
div
isi
on
of
t
he
area/m
app
ing
a
rea
into
10
8
ref
e
re
nce
points
at
1.6 m i
nterv
al
s
. F
r
om
h
is r
ese
arch, there was
an
incr
ease i
n ac
cur
acy
whe
n add
in
g
filt
erin
g
w
hen
taki
ng RSSI
values
i
n
the
f
ing
e
rprintin
g
m
et
ho
d.
The
s
ta
nd
a
rd
de
viati
on
for
the
m
easur
em
ent
area
ov
e
r
distances
above
40 m
is 0
.51 m
.
Ou
et
al
.
[
15]
did
rese
arc
h
ed
on
in
door
pos
it
ion
ing
with
the
pro
xim
i
ty
m
et
ho
d
w
hic
h
was
ap
plie
d
us
in
g
I
EEE
802.15.4
(ZigBee
).
Zig
bee
was
chosen
beca
use
it
has
a
low
cost,
lo
w
powe
r
co
nsum
ption
,
sm
a
ll
siz
e
and
is
easy
to
us
e
com
par
e
d
to
W
iFi
,
so
it
is
ver
y
su
it
able
for
s
hort
distance
wireless
tra
nsm
issi
on
syst
e
m
s.
The
par
am
et
ers
us
ed
as
m
easur
ing
m
a
te
rial
are
tim
e
of
arr
iva
l
(TO
A
),
ti
m
e
diff
e
ren
ce
of
arr
ival
(TDO
A)
,
a
ngle
of
ar
rival
(AOA),
an
d
recei
ved
si
gn
al
stre
ng
t
h
ind
ic
at
io
n
(RSSI).
Wh
e
n
colle
ct
ing
RSSI,
o
u
us
es
a
m
od
ific
at
ion
of
the
prox
im
ity
m
e
thod,
w
hich
is
tria
ngulati
on
betw
een
the
two
ini
ti
al
ref
eren
ce
Nodes
com
bin
ed
s
o
t
hat
they
i
nters
ect
to
form
a
new
re
fer
e
nce
N
od
e
.
T
he
ti
m
e
require
d
is
faster
an
d
co
m
plex
cal
culat
ion
s a
r
e re
du
ce
d. Th
e
r
es
ulti
ng
a
ve
r
age e
rror i
s 0.42 m
.
F
onsek
a
an
d
S
andrase
gar
a
n
[
16
]
did
resea
rc
h
ed
on
in
door
po
sit
io
ning
for
IoT
ap
plica
ti
ons
us
in
g
the
fin
gerpr
i
nting
m
et
ho
d
t
hat
is
app
li
e
d
us
i
ng
W
i
Fi.
T
he
al
gorithm
us
ed
is
pro
bab
il
it
y
and
wei
gh
te
d
k
-
near
es
t
neig
hbor
(
WKNN),
as
well
as
a
co
m
bin
at
ion
of
bo
t
h.
The
aim
of
h
is
research
is
to
i
m
pr
ov
e
the
perfor
m
ance
of
t
he
fin
ge
rpr
inti
ng
m
et
ho
d
in
te
rm
s
of
preci
sion
,
acc
ura
cy
and
dura
bili
ty
.
In
the
offli
ne
phase,
al
l
W
i
Fi
sign
al
s
co
ntained
in
the
re
s
earch
area
a
re
read
inf
or
m
at
ion
su
c
h
as
sign
al
le
vel
(RSS)
,
si
gn
al
qu
al
it
y,
mo
du
la
ti
on
an
d
m
edia
m
edia
acce
ss
co
ntro
l
(
MAC
)
a
ddres
s.
RSS
val
ues
that
ha
ve
bee
n
read
a
re
entere
d
int
o
a
database
that
will
be
use
d
as
input
in
t
he
on
li
ne
phase
.
In
t
he
res
earc
h
area
of
6
m
x
6
m
,
49
ref
e
r
ence
po
i
nts
we
re
0.5
m
away,
with
a
ta
r
get
ta
r
ge
t
sta
ti
on
,
li
ne
of
sig
ht
(
L
OS
)
c
onditi
ons,
us
in
g
5
acce
ss
po
i
nt
s
and
a
com
bin
at
ion
of
pro
ba
bili
ty
and
WK
N
N
K=3
al
gorith
m
s
ob
ta
ined
a
n
erro
r
distan
ce
of
0.4
757
m
,
it
s
pr
eci
sio
n
inc
re
ases to 88%.
I
n t
he
stud
y area
of
5m
x3
m
,
the target d
e
vice m
ov
es to n
on
-
LOS
c
onditi
ons,
with
K
=
7
a
n
e
rro
r
distance
of 0.4
025
m
is obtai
ned, its
pr
eci
si
on inc
reases t
o 9
9%.
AlSh
am
aa
et
al
.
[17]
di
d
re
searche
d
on
l
oc
al
iz
at
ion
of
s
ens
or
s
in
i
ndoor
wireless
ne
tworks:
A
n
ob
s
er
vation
m
od
el
us
in
g
WiFi
RSS.
The
m
et
ho
d
us
ed
is
fing
e
rprint
ing
a
nd
cl
us
te
rin
g
with
hardw
a
r
e
i
m
ple
m
entat
io
n
in
the
WL
AN
li
ving
la
b
e
nv
i
ronm
ent
of
Tr
oyes
Un
i
ver
sit
y
of
Tech
nolo
gy,
Fr
a
nce.
The
par
am
et
er
us
e
d
is
the
R
S
S
W
iFi
of
each
acce
ss
poi
nt
(
AP
)
at
that
loc
at
ion
with
a
re
search
area
of
500
m
2
div
ide
d
int
o
19
cl
us
te
rs.
Wh
e
n
ex
per
im
enting
with
c
om
bin
ing
fin
gerpr
i
nting
a
nd
cl
us
te
ri
ng
m
et
ho
ds
without
AP
Sele
ct
io
n,
ob
ta
ine
d
an
accuracy
rate
of
88
.21%
durin
g
the
trai
ni
ng
phase
an
d
86
.
26%
du
ring
t
he
po
sit
io
ning
ph
ase. W
it
h
the
a
dd
it
io
n
of
the A
P
sel
ect
ion
m
et
hod,
the
acc
uracy
increases to
92.
78
% dur
i
ng
t
he
trai
ning
ph
ase
an
d
90.
42%
durin
g
the
pos
it
ion
ing
phase
.
H
ow
e
ver,
with
the
ad
diti
on
of
the
AP
S
e
le
ct
ion
m
et
ho
d,
it
will
ad
d
pr
ocessin
g
ti
m
e
to
the
t
rainin
g
ph
a
se.
AlSh
am
aa
al
so
m
ade
a
com
par
is
on
bet
wee
n
the
m
et
ho
ds
car
ried
ou
t
with
m
et
ho
ds
that
oth
e
rs
hav
e
done
suc
h
as:
K
-
near
e
st
nei
ghbo
rs,
Naive
Ba
ye
s,
m
ul
ti
no
m
ial
log
ist
ic
regressi
on,
ne
ur
al
netw
or
ks
a
nd
S
VM
.
From
al
l
the
m
et
ho
ds
car
ried
ou
t
e
xperim
e
nts
on
the sam
e
en
vir
on
m
ent, scen
a
r
io an
d dev
ic
e
,
t
o
get r
es
ults with h
ig
h
accu
rac
y, it
r
equ
ires m
or
e ti
m
e
d
ur
i
ng the
trai
ning
ph
as
e
in f
i
ng
e
rprintin
g
.
In
this
w
ork
,
we
pro
pose
a
stud
y
to
im
pr
ov
e
the
per
for
m
ance
of
the
le
vel
of
sp
ee
d
in
te
r
m
s
of
processi
ng
ti
m
e
o
n
t
he
trai
ni
ng
ph
as
e
of
the
fin
gerp
rint
ing
m
et
ho
d
for
indo
or
posit
ion
in
g
with
c
ertai
n
al
gorithm
s
by
hard
war
e
im
ple
m
entat
ion
us
ing
Zig
bee.
T
he
outp
ut
gen
e
r
at
ed
w
he
n
us
i
ng
certai
n
al
go
rithm
s
can
red
uce
t
he
processi
ng
ti
m
e
in
the
trai
ni
ng
phase.
Thi
s
resea
rch
was
co
nducted
in
an
act
ual
r
oo
m
with
a
n
area
of
in
door
area
that
is
3
m
x
3.
6
m
(e
m
pty
sp
ace).
The
m
et
ho
d
us
ed
is
finge
rprintin
g
for
indo
or
po
sit
io
ning
with
WSN.
T
he
com
m
un
ic
at
i
on
prot
oco
l
s
ta
nd
a
rd
us
e
d
is
IEEE
802.
15.4
(Zi
gBee)
.
Th
e
par
am
et
ers
ta
ken
in
the
fin
ge
rprintin
g
m
et
h
od
a
re
RSS
I
(
r
ecei
ved
stre
ngth
sig
nal
ind
ic
a
tor)
from
each
sens
or
Node.
Eac
h
se
ns
or
N
od
e
is
pl
aced
in
an
ide
al
conditi
on
/
LOS
,
so
that
t
he
infl
uen
ce
of
the
h
um
an
bo
dy
and
furn
it
ure
ca
n
be
igno
red
[
18
]
.
This
st
ud
y
use
d
tw
o
m
easurem
ent
m
et
ho
ds
w
hich
wer
e
c
arr
ie
d
out
duri
ng
the
trai
ning
ph
a
se.
The
res
ults
of
the
durati
on
of
the
tw
o
m
et
ho
ds
are
the
n
c
om
par
ed
s
o
w
e
get
data
to
gi
ve
a
con
cl
us
io
n.
T
he
pap
e
r
is
or
ga
nized,
s
ect
ion
2
deals
with
t
he
re
sea
rc
h
str
at
egy
perform
ed
in
this
w
ork
an
d
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
2
4
, N
o.
1
,
Oct
o
ber
20
21
:
23
6
-
24
4
238
how
t
he
proce
dures
fixe
d
ta
r
get
pa
ram
et
er
(F
TP
)
a
nd
m
ov
in
g
ta
r
get
para
m
et
er
(MTP)
are
im
ple
m
ented
in
fin
gerpr
i
nted
m
et
ho
d.
Sect
ion
3
de
scribe
s
the
m
easur
e
m
ent
resu
lt
s
f
or
var
i
ou
s
sce
nar
i
o
an
d
it
s
analy
sis.
Sect
ion
4 giv
es
the c
on
cl
us
io
n o
f
the
p
a
per.
2.
RESEA
R
CH MET
HO
D
Fo
r
t
he
propos
ed
m
od
el
,
we
us
e
zi
gbee
devi
ces
that
are
c
oupled
with
a
m
esh
netw
ork
topolo
gy
to
form
a
wireles
s
sen
sor
netw
ork
.
WSN
cat
e
gory
us
e
d
is
str
uctu
re
WSN,
the
se
nsor
N
od
e
is
placed
i
n
a
fixed
po
sit
io
n
pe
rm
a
nen
tl
y,
so
as
t
o
re
du
ce
netw
ork
savi
ngs
an
d
m
anag
em
ent
costs
[19]
.
I
n
gen
e
ral,
eac
h
sens
or
Node
co
ntains
4
su
b
-
syst
em
s,
nam
el
y:
Co
ntr
oller
sub
-
sy
stem
;
Co
m
m
u
nicat
ion
s
ub
-
sy
stem
;
Sens
in
g
sub
-
syst
e
m
;
and
Power
S
upply
sub
-
syst
em
s
[2
0],
[21]
.
Zigb
ee
can
be
us
e
d
as
a
sensor
N
od
e
because
each
s
ens
or
do
e
s
no
t
re
quire
wide
ba
ndw
idth,
but
requi
res
low
la
te
nc
y
and
low
en
e
rg
y
co
nsum
pti
on.
Zigbe
e
has
been
introd
uced
by
IEEE
a
s
a
n
IE
EE
80
2.15.4
c
omm
un
ic
at
ion
sta
nd
a
rd
[
22]
that
w
orks
at
2,
4
Gh
z
f
reque
nc
y
[23]
.
The
re
a
re
seve
ral
netw
ork
to
po
l
og
ie
s
that
c
an
be
use
d
on
WSN,
one
of
them
is
the
Mesh
To
polo
gy
that
is
of
te
n
us
ed
be
c
ause
it
has
a
high
redu
nd
a
nc
y
[24]
.
T
he
pa
ram
et
er
us
ed
from
each
N
ode
is
RS
SI
w
hi
ch
is
app
li
ed
t
o
indo
or
p
os
it
io
ning.
RSSI
is
short
f
or
Re
cei
ve
d
Si
gn
al
Stre
ngth
I
nd
ic
at
or, w
hic
h
m
eans
an
indi
cat
or
of
th
e sig
nal
st
rength
posses
s
ed
by
a w
irel
es
s
tran
sm
issi
on
dev
ic
e
bet
wee
n
on
e
d
evice
a
nd
an
othe
r
[
25]
.
RSSI
is
al
so
a
sta
nd
ard
pa
ram
et
er
in
wireless
co
m
m
un
ic
at
ion
that
is
of
te
n
use
d
by
oth
e
r
re
searche
rs
[
26
]
.
In
this
stud
y
RSS
I
w
as
cho
se
n
bec
ause
it
can
be
us
ed
for
the
f
ing
e
rprintin
g
m
et
ho
d
with
the
distance
from
each
sens
or
Node
th
at
is
cl
os
e
enough.
RS
SI
m
ea
su
rem
ents
can
ref
le
ct
a
dis
ta
nc
e
betwee
n
eac
h
sen
sor
N
ode.
The
far
the
r
th
e
dist
ance
betwee
n
t
he
se
nsor
N
odes,
the
sm
al
le
r
the
RSS
I
valu
e
obta
ined
.
T
he
m
a
in
obj
ect
i
ve
of
this resea
rch is
to
a
pp
ly
the
RSSI
data c
ollec
ti
on
m
et
ho
d
i
n t
he
trai
ni
ng ph
ase w
hich
is
m
or
e
ef
fecti
ve
s
o
as t
o
r
ed
uce
the
dur
at
ion
of
the
processin
g
ti
m
e
i
n
the
trai
ni
ng
ph
a
se.
T
his
sect
ion
will
disc
us
s
incl
ud
i
ng
r
esearc
h
fo
c
us
,
researc
h
flow,
fin
gerp
r
inti
ng
m
et
ho
ds
in
indoor
posi
ti
on
in
g
an
d
RSSI
data
colle
ct
ion
m
et
ho
ds
in
the
trai
ning
ph
as
e.
2.1.
Rese
arc
h st
r
ate
gy
W
i
reless
se
nso
r
net
wor
k
(
W
SN
)
ha
s
m
any
ap
plica
ti
on
s
t
hat
can
be
ap
plied
in
it
,
su
c
h
as
i
ndoo
r
po
sit
io
ning.
D
et
erm
inati
on
of
the
l
ocati
on
of
ob
j
ect
s
in
a
n
indoor
a
rea
ca
n
be
do
ne
with
a
va
riet
y
of
m
et
hods
,
on
e
of
wh
ic
h
is
fing
e
rprintin
g.
T
her
e
are
s
ever
al
m
et
ho
ds
that
can
be
us
e
d
for
in
door
posit
ion
in
g/i
ndoor
local
iz
at
ion
sy
stem
s
(I
LS
)
[
27
]
,
su
c
h
as
pro
xim
it
y,
trian
gula
ti
on,
fin
ge
rprintin
g,
dea
d
re
ck
onin
g.
I
n
the
fin
gerpr
i
nting
m
et
ho
d,
the
re
are
two
m
ai
n
ph
ases
,
nam
e
ly
the
trai
nin
g
ph
ase
an
d
th
e
po
sit
ion
i
ng
ph
a
se.
The
f
oc
us
of
this
resea
rch
is
to
optim
iz
e
the
trai
nin
g
phase
in
the
fin
gerp
r
inti
ng
m
et
ho
d.
In
si
de
there
a
r
e
two
m
et
ho
ds,
nam
el
y
fixed
ta
rg
e
t
par
am
et
er
(F
TP
)
an
d
m
ov
ing
ta
r
get
pa
ra
m
et
er
(
MTP
)
wh
ic
h
is
a
m
eth
od
for
retrievin
g
RSS
I
data
in
the
trai
nin
g
ph
a
se.
Af
te
r
knowin
g
the
fo
cu
s
of
t
he
resear
ch
to
be
co
nducted
,
the
nex
t
ste
p
is
to
m
ake
a
research
flo
wch
a
rt
to
achi
eve
the
ob
j
ect
ives
of
the
rese
arch.
The
re
se
arch
fl
ow
c
ha
rt
can
be
seen i
n
Fi
gure
1
.
Figure
1
.
Re
se
arch flo
wc
har
t
S
T
A
RT
F
IN
IS
H
L
IT
E
RA
T
U
RE
RE
V
IE
W
CO
N
F
IG
U
RIN
G
W
S
N
H
A
RD
W
A
RE
W
IT
H
Z
IG
BE
E
A
N
A
L
Y
S
IS
A
N
D
CO
N
CL
U
S
IO
N
S
t
a
rt
A
ppl
i
c
a
t
i
on
of T
he
F
i
nge
rpri
nt
i
ng M
e
t
hod
A
P
P
L
ICA
T
IO
N
O
F
T
H
E
F
IN
G
E
RP
RIN
T
IN
G
M
E
T
H
O
D
M
a
ppi
ng T
he
M
e
a
s
ure
m
e
nt
A
re
a
M
e
a
s
ure
m
e
nt
w
i
t
h T
he
F
i
xe
d T
a
rge
t
P
a
ra
m
e
t
e
r (F
T
P
) S
c
he
m
e
M
e
a
s
ure
m
e
nt
w
i
t
h T
he
M
ovi
ng
T
a
rge
t
P
a
ra
m
e
t
e
r
(M
T
P
)
S
c
he
m
e
D
one
N
ode
X
M
e
a
s
ure
s
T
he
RS
S
I of
A
l
l
S
e
ns
or N
ode
s
i
n e
a
c
h M
a
ppi
ng
A
re
a
A
l
l
N
ode
S
e
ns
ors
M
e
a
s
uri
ng
RS
S
I
N
ode
X
w
he
n M
ovi
ng
RS
S
I M
e
a
s
ure
d R
e
fe
re
nc
e
N
ode
s
i
n A
l
l
M
a
ppi
ng
A
re
a
s
M
e
a
s
ure
m
e
nt
D
ura
t
i
on
M
T
P
<
F
T
P
?
Ch
a
nge
i
n M
ove
m
e
nt
S
c
he
m
e
N
ode
X
Y
e
s
No
Re
fe
re
nc
e
D
a
t
a
ba
s
e
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng
&
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Impr
oved fi
ng
e
rp
rinti
ng
perf
orma
nce
in
i
ndoor
posit
ion
in
g by re
duci
ng
dura
ti
on
of the…
(
And
ik
a M
uharam
)
239
The
researc
h
m
et
ho
d
is
ca
rr
i
ed
ou
t
with
se
ver
al
sta
ges
in
accor
da
nce
wit
h
Fi
gure
1
.
Fi
r
st,
li
te
ratur
e
rev
ie
w
sta
ge
,
r
ead
s
om
e
of
th
e
ref
e
ren
c
es
of
pr
e
vious
jour
nal
/
researc
h
w
rite
rs
to
l
ook
f
or
pro
blem
s
that
are
deem
ed
neces
sary
to
de
v
el
op
or
co
ntin
ue
their
resea
rch.
Config
ur
i
ng
WSN
ha
rdwa
r
e
with
Zig
bee
sta
ge,
determ
ine
the
po
sit
io
n
of
eac
h
se
ns
or
Node,
in
or
der
t
o
obt
ai
n
an
i
deal
res
ult,
nam
el
y
the
li
ne
of
sig
ht
(
LO
S
)
to
the
ref
e
re
nc
e
N
od
e
(
Node
X)
.
A
ppli
cat
ion
of
t
he
fin
gerpr
i
nting
m
e
tho
d
sta
ge.
I
niti
al
ly
the
m
app
ing
of
t
he
m
easur
em
ent
area
is
carrie
d
ou
t
acco
r
ding
to
the
co
nd
it
ion
s
i
n
the
fiel
d.
T
he
pa
ram
e
te
r
m
easur
e
d
is
RSSI
(r
ecei
ve
d
si
gn
al
stren
gth
i
ndic
at
or
)
from
No
de
X
a
nd
al
l
sens
or
N
od
es
in
eac
h
m
app
in
g
area
.
RSS
I
s
ens
or
Node
r
e
fer
e
nc
e
that
is
read
by
No
de
X
is
entere
d
into
the
database
w
hich
will
be
us
ed
as
the
nex
t
refe
ren
ce
value.
m
easur
e
m
ent
of
RSSI
at
the
trai
nin
g
sta
ge
in
the
fin
gerpr
i
nting
m
e
thod
is
done
in
two
ways,
na
m
el
y
fixe
d
ta
r
get
pa
ram
et
er
(F
TP)
an
d
m
ov
in
g
t
arg
et
p
a
ram
et
e
r
(MT
P).
F
rom
the
res
ults
of
the
FTP
an
d
MT
P
m
et
ho
d,
t
he
durati
on
of
the
trai
nin
g
phas
e
is
then
co
m
par
ed
and
a
naly
zed.
T
he
n
The
A
naly
sis
an
d
Con
cl
us
io
ns
st
age.
from
the
r
esults
of
m
easur
em
ents
an
d
data
processin
g
that
has
bee
n
ob
ta
i
ned,
will
fin
d
a
con
cl
us
io
n fro
m
a stud
y.
2.2.
Fin
gerpr
inting
met
hod
In
the
fi
ng
e
r
pri
nting
m
et
ho
d,
there
are
tw
o
ph
a
ses
of
the
process
that
m
us
t
be
car
ried
ou
t,
nam
el
y
trai
ning
a
nd
posit
ion
i
ng.
I
n
t
he
trai
ning
ph
a
se,
the
N
ode
X
sens
or
is
fi
rst
m
app
ed
into
th
e
sm
app
ing
s
c
hem
e
in
accor
da
nce
with
the
m
easur
em
ent
locat
io
n
ap
plied.
N
ode
X
can
know
the
RSSI
val
ue
of
each
se
nsor
Nod
e
arou
nd
it
,
t
he
n
the
values
a
re
entere
d
i
nto
a
database
that
will
be
us
ed
as
a
re
fer
e
nce
va
lue
in
t
he
nex
t
ph
a
se.
In
this
stu
dy,
the
i
m
ple
m
entat
ion
was
car
rie
d
out
in
an
in
door
room
(em
pt
y
sp
ace)
ideal
con
diti
on
(
sens
or
Node
LO
S)
wi
th
a
siz
e
of
3
x
3.6
m
2
.
The
po
sit
io
n
of
the
ref
ere
nce
N
ode
sensor
is
placed
at
a
heigh
t
of
2
m
et
ers
and
Node
X
is place
d
on
a 60
cm
wh
eel
ed
chair
.
A
n
il
lustrati
on
of the
posit
ion
o
f
the
sens
or
N
ode
ca
n
be
see
n
i
n
Fi
gure
2
.
Figure
2
.
I
ll
us
t
rati
on of se
nso
r
N
ode
placem
ent
Node
X
is
a
se
ns
or
Node
a
s
a
ta
rg
et
locat
io
n
that
you
wa
nt
to
kn
ow
it
s
loc
at
ion
by
colle
c
ti
ng
RSS
I
inf
or
m
at
ion
from
the
sensor
ref
e
ren
ce
N
odes
that
a
re
ar
ound
it
,
nam
ely
Node
A,
Node
B,
a
nd
N
ode
C
.
Term
inals
are
connecte
d
di
re
ct
ly
to
Node
X
to
see
th
e
s
urrou
nd
i
ng
RS
SI
in
real
ti
m
e
.
I
n
th
e
FT
P
m
et
ho
d,
Node
X
is
in
an
area
t
hat
ha
s
bee
n
m
app
ed
first
.
N
od
e
X
colle
ct
s
RSSI
data
from
each
re
fer
e
nc
e
N
ode
al
te
rn
at
el
y
in
each
de
sig
nated
area.
T
he
sche
m
e
is
m
ade
in
accor
da
nce
wit
h
the
dim
ension
s
of
the
s
pace
us
e
d
as a
place o
f re
search
, as
sho
wn in Fi
gure
3
.
In
Fig
ur
e
3
(a),
t
he
locat
io
n
m
app
in
g
sc
he
m
e
was
create
d
by
di
vid
in
g
t
he
m
easur
em
e
nt
area
by
30
m
app
in
g
areas
.
Each
of
these
m
easur
em
ent
areas
is
60
x
60
cm
2
.
The
div
isi
on
of
these
areas
is
exp
ect
ed
to
i
m
pr
ove
the
posit
ion
i
ng
res
ul
ts
m
or
e
accu
r
at
el
y.
Af
te
r
th
e
m
app
ing
sc
hem
e
is
create
d,
the
nex
t
ste
p
is
to
determ
ine
the
locat
ion
of
the sen
s
or
N
ode
th
a
t
will
be
us
ed
as
a
ref
e
ren
ce
. Th
e
sc
hem
a
se
ns
or
re
fer
e
nce N
ode
placem
ent
is
seen
in Figure
3
(
b)
.
T
he
se
ns
or N
ode u
se
d
in
this
stu
dy
use
s 4
se
nsor
N
od
e u
nits, 1
unit
as N
ode
X
an
d
3
oth
e
r
un
it
s
as
sens
or
re
fer
e
nce
N
od
e
s
(No
des
A,
B,
an
d
C).
In
the
FTP
m
et
ho
d,
Node
X
in
a
sta
ti
on
ary
sta
te
in
on
e
of
the
m
app
in
g
areas
m
easur
es
the
RSSI
of
each
s
ens
or
N
ode
refe
ren
ce
(
N
od
e
s
A,
B
and
C)
agai
ns
t
N
od
e
X
.
T
hes
e
m
easur
em
ent
s
are
re
peated
on
e
by
one
i
n
each
m
app
in
g
area
(A1
-
F5).
The
n
the m
easur
ed R
SSI
value
is s
tore
d
in
the
dat
abase
w
hich w
il
l be u
se
d
as
a
ref
e
ren
ce
v
al
ue
.
In
the
MTP
m
et
hod,
N
ode
X
in
the
m
ap
ping
area
is
not
at
rest,
bu
t
there
is
m
ov
e
m
ent
with
an
aver
a
ge
s
peed
of
0.06
m
/s.
As
Node
X
m
oves,
the
refe
re
nc
e
Node
sen
sors
(
Node
A
,
B
and
C)
retrieve
data
/
m
easur
e
t
he
RSSI
N
od
e
X
val
ue
of
eac
h
se
nsor
N
od
e
re
fer
e
nce.
Il
lustrati
on
of
Node
X
m
ov
e
m
ents
perform
ed
can
be
see
n
i
n
Fi
gure
4
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
2
4
, N
o.
1
,
Oct
o
ber
20
21
:
23
6
-
24
4
240
(a)
(b)
Figure
3
.
Tar
ge
t posi
ti
on
a
nd
sens
or
N
ode re
fer
e
nce
m
app
i
ng sc
hem
e
Figure
4
.
I
ll
us
t
rati
on of
Node
X
m
ov
em
ent o
n
MTP
in
t
rain
ing
phase
Figure
4
il
lust
r
at
es
the
m
ov
em
ent
of
N
ode
X
in
the
MT
P
m
et
ho
d
wh
ic
h
si
m
ultaneou
sly
each
sen
sor
Node
re
fer
e
nc
e
(
N
od
e
A,
B
and
C)
ta
ke
s
the
le
vel
of
RS
SI
N
od
e
X
to
the
sens
or
Nod
e
ref
ere
nce.
La
te
r
the
RSSI
val
ue
is
proc
esse
d
an
d
com
par
ed
to
m
at
ch
the
RSSI
value
in
the
FTP
m
et
ho
d.
The
ap
pro
pr
ia
t
e
RSSI
value wil
l be
used as
the
ne
xt
ref
e
ren
ce
v
al
ue
for
t
he posi
ti
on
in
g ph
a
se.
3.
RESU
LT
S
A
ND
D
IS
C
USS
ION
In
t
his
te
sti
ng
and
m
easur
em
ent
phase,
it
use
s
two
m
et
ho
ds
as
e
xpla
ine
d
in
t
he
pre
vio
us
cha
pter
,
nam
ely
the
fi
xed
ta
r
get
para
m
et
er
(F
TP)
and
m
ov
in
g
ta
rg
et
pa
ram
eter
(MTP)
m
eth
ods.
B
oth
of
these
m
et
ho
ds
a
re
use
d
du
rin
g
the
trai
ning
phas
e
in
the
fin
ge
r
pr
i
nting
m
et
hod
to
c
ollec
t
R
SSI
data
from
each
sens
or
N
od
e
r
e
fer
e
nce a
nd it
s v
al
ue
is e
ntere
d
int
o
the
data
base in t
he fo
r
m
o
f
a ref
e
ren
c
e v
al
ue
table t
ha
t wil
l
be use
d
at
the
po
sit
io
ning sta
ge.
3.1.
Measure
ment o
f
t
he
fixed t
arg
e
t
p
ar
amet
er
met
hod
Me
asur
i
ng
ste
ps
a
re
ca
rr
ie
d
ou
t
i
n
acc
orda
nce w
it
h
t
he
m
easur
em
ent
m
e
thod p
la
nne
d
i
n
the
p
re
vious
chap
te
r.
T
he
fi
rst
m
easur
e
m
e
nt
m
e
tho
d
is
done
by
the
FT
P
m
e
tho
d.
N
ode
X
is
in
one
of
the
a
reas
t
hat
has
been m
app
ed b
efore
(ex
am
ple: area B3
)
i
n
a
sta
ti
on
ary c
ondi
ti
on
, as
sho
wn in F
i
gure
5
.
In
Fig
ure
5
is
a
m
easur
em
ent
schem
e carr
ie
d
ou
t i
n
the co
ndit
ion
of line of sigh
t (L
OS), N
ode X
in
a
sta
ti
on
ary
co
ndit
ion
in
A
rea
B3
gets
the
RSSI
val
ue
fro
m
the
ref
eren
c
e
Node
(A,
B,
and
C)
.
The
n
the
RSSI
value
is
ente
re
d
int
o
t
he
database
w
hich
will
be
us
e
d
as
a
r
efere
nce
value
.
The
dur
at
ion
of
the
t
rain
in
g
ph
a
se
with F
TP i
n
Area
B3 ta
kes 10
seco
nd
s
for ea
ch refe
ren
ce
Nod
e
.
In
Fig
ur
e
6
c
onta
ins
the
dura
ti
on
of
c
reati
ng
a
ref
e
ren
ce
da
ta
base
in
t
he
t
rainin
g
ph
a
se
with
FT
P
in
each
cal
culat
io
n
area
f
or
3
r
efere
nce
N
ode
s
(A
,
B
an
d
C).
The
X
-
a
xis
sh
ows
the
li
st
of
areas
that
RSSI
m
easur
em
ents
hav
e
bee
n
m
ade
by
N
ode
X
again
st
the
re
fer
e
nce
Node,
nam
ely
areas
A1
to
F5.
T
he
Y
-
axis
determ
ines
the
tim
e
(in
sec
onds
)
to
ta
ke
RS
SI
m
easur
em
e
nts
in
each
m
a
pp
i
ng
area
.
T
he
∆t
N
ode
A
ba
r
c
har
t
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng
&
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Impr
oved fi
ng
e
rp
rinti
ng
perf
orma
nce
in
i
ndoor
posit
ion
in
g by re
duci
ng
dura
ti
on
of the…
(
And
ik
a M
uharam
)
241
sh
ows
the
dura
ti
on
of
the
R
S
SI
m
easur
em
e
nt
unti
l
in
pu
t
i
nto
t
he
databas
e
by
Node
X
a
gainst
N
od
e
A
.
The
∆t
Node
B
ba
r
ch
art
shows
t
he
durati
on
of
the
RSSI
m
easur
em
ent
un
ti
l
the
input
into
the
database
by
N
od
e
X
against
N
ode
B.
The
∆t
No
de
C
bar
char
t
sh
ows
the
inc
re
ase
in
RSSI
to
input
i
nto
the
da
ta
base
by
Node
X
t
o
Node
C.
The
durati
on
of
the
trai
ning
phase
with
FTP
re
qu
ires
an
ave
rag
e
tim
e
of
30
sec
onds
in
eac
h
area
to
cal
culat
e
fo
r
the
us
e
of
3
r
efere
nce
N
od
e
s.
The
m
or
e
sens
or
N
odes
use
d
an
d
the
la
rg
e
num
ber
of
area
m
app
in
gs
,
the
m
or
e
tim
e i
t t
a
kes
t
o
c
ollec
t r
efere
nces.
Figure
5
.
RSS
I
m
easur
e
m
ent in a
rea
B3
s
c
he
m
e
Figure
6. G
raph
of trainin
g p
hase
process
dur
at
io
n wit
h F
TP
3.2.
Measure
ment o
f
t
he m
ov
in
g t
arg
e
t
p
arameter
met
ho
d
The
seco
nd
m
et
hod
in
the
tr
ai
nin
g
ph
a
se
is
us
in
g
MTP.
In
this
m
et
ho
d,
the
ref
e
ren
ce
Node
ta
kes
RSSI dat
a fro
m
the
m
ov
in
g
Node X
. In Fi
gure
7
s
hows
t
he
N
ode
X
m
oves f
r
om
the r
ef
eren
ce
N
od
e
(
Node A)
to
oth
e
r
re
fer
e
nce
N
ode
(
Node
B).
Wh
e
n
Node
X
sta
rts
to
m
ov
e,
each
ref
e
ren
ce
Nod
e
(A
,
B,
a
nd
C)
ta
ke
s
RSSI
data
fro
m
No
de
X.
I
n
Fig
ur
e
7
is
t
he
first
Node
X
m
ov
em
ent
m
od
el
on
MT
P,
the
re
a
re
2
othe
r
m
ov
e
m
ent
m
od
el
s,
nam
ely
from
area
A3
t
o
Ar
ea
F3
a
nd
f
r
om
No
de
C
t
o
area
A
4.
The
r
esults
of
the
R
SSI
i
n
area B3
can
be
seen
in
Ta
ble
1.
Table
2
s
hows
li
sts
the
durat
ion
of
the
trai
ning
ph
ase
wi
th
MTP
in
are
a
B3.
Of
the
t
hr
ee
MT
P
m
ov
e
m
ent
m
o
dels
,
3
ti
m
es
the
RSSI
m
easur
e
m
ents
wer
e
m
ade
f
or
Nodes
A,
B
an
d
C
in
a
rea
B3.
I
n
the
MTP
m
easur
em
ent
schem
e,
Node
X
m
ov
es
with
a
sp
ee
d
of
0.0
6
m
/s,
so
t
he
durati
on
nee
ded
f
or
t
he
trai
ning
ph
a
se
with MTP
is a
n
a
ver
a
ge
ti
m
e
of 25 sec
onds i
n
eac
h
m
app
in
g
a
rea
Table
1.
RS
SI
m
easur
em
ent r
esults i
n
a
rea B
3 wit
h
MTP
m
et
hod
No
d
e
X
In Are
a
B3
RSSI
No
d
e
A
(dB
m
)
RSSI
No
d
e
B
(dB
m
)
RSSI
No
d
e
C
(dB
m
)
Mod
el
-
1
-
5
5
/
-
64
-
5
5
/
-
62
-
5
2
/
-
57
Mod
el
-
2
-
5
5
/
-
65
-
5
4
/
-
61
-
5
2
/
-
62
Mod
el
-
3
-
5
5
/
-
62
-
5
5
/
-
62
-
5
2
/
-
55
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
2
4
, N
o.
1
,
Oct
o
ber
20
21
:
23
6
-
24
4
242
Figure
7. Illust
rati
on of the
m
ov
em
ent of
Node X
from
N
ode A
t
o Node B
Table
2.
List
of traini
ng phas
e durati
ons
with
MTP
in
a
rea
B3
No
Mov
e
m
en
t
Mod
el M
T
P
Towards
No
d
e
A,
B an
d
C in
Area
B3
(1)
∆t
ABC
-
1
(secon
d
)
Towards
No
d
e
A,
B an
d
C in
Area
B3
(2)
∆t
ABC
-
2
(secon
d
)
Towards
No
d
e
A,
B an
d
C in
Area
B3
(3)
∆t
ABC
-
3
(secon
d
)
Start
End
Start
End
Start
End
1
Mod
el 1 (
No
d
e
A
t
o
No
d
e
B)
1
7
.19
.4
6
1
7
.20
.1
3
0
0
.00
.27
1
7
.32
.5
6
1
7
.33
.2
0
0
0
.00
.24
1
7
.43
.4
0
1
7
.44
.0
7
0
0
.00
.27
2
Mod
el 2 (A
rea
F3
to
Area
A3
)
1
7
.35
.5
5
1
7
.36
.2
0
0
0
.00
.25
1
7
.36
.0
0
1
7
.36
.2
6
0
0
.00
.26
1
7
.47
.2
4
1
7
.47
.4
8
0
0
.00
.24
3
Mod
el 3 (
No
d
e
C t
o
Area
A4
)
1
7
.38
.5
4
1
7
.39
.1
8
0
0
.00
.24
1
7
.38
.4
5
1
7
.39
.0
9
0
0
.00
.24
1
7
.49
.2
5
1
7
.49
.4
8
0
0
.00
.23
Av
erage Duratio
n
s
of
eac
h
Area
pe
r
-
3
No
d
e
s
(secon
d
)
0
0
.00
.25
0
0
.00
.25
0
0
.00
.25
3.3.
Data pr
oc
essing
and
analysis
Fr
om
the
re
su
l
ts
of
RSS
I
m
e
asur
em
ents
wi
th
the
fixe
d
ta
rg
et
par
am
et
er
(F
T
P)
a
nd
m
ov
i
ng
ta
r
get
par
am
et
er
(MTP)
m
et
ho
ds
tha
t
hav
e
bee
n
car
ried
ou
t,
sam
ple
m
easur
em
ents
are
ta
ke
n
in
Ar
ea
B
3.
T
he
RSS
I
m
easur
em
ent
resu
lt
s
is
showe
d
Ta
ble
3.
Tab
le
3
sho
ws
that
the
RSSI
valu
e
m
easur
ed
by
FTP
is
cl
os
e
t
o
th
e
RSSI
value
m
easur
e
d
by
MTP.
Wh
e
n
m
e
asur
i
ng
us
i
ng
t
he
FT
P
m
et
ho
d,
a
n
a
ver
a
ge
tim
e
of
30
sec
onds
is
require
d
in
ea
ch
m
app
in
g
ar
ea,
w
hile
the
MTP
m
et
h
od
requires
a
n
a
ve
rag
e
ti
m
e
of
25
seco
nds
in
each
m
app
in
g
area
.
In
t
he
MTP
m
e
thod,
it
can
be
seen
f
r
om
sever
al
gr
a
phs
that
hav
e
bee
n
displ
ay
ed
that
the
r
ange
of
the
l
ow
e
r
a
nd
up
per
lim
its
of
the
RSS
I
ob
ta
ine
d
is
cl
ose
to
m
easur
e
m
ent
by
the
F
TP
m
et
ho
d.
S
o
t
hat
the
MTP m
et
ho
d
c
an
im
pr
ove the
eff
ic
ie
nc
y o
f
t
he fin
gerpr
i
nting m
et
ho
d
th
at
is ap
plied t
o
i
ndoor
posit
ion
i
ng.
Table
3.
C
om
par
iso
n o
f
RSS
I FTP a
nd MTP
m
easur
em
ent r
esults i
n
a
rea B
3
No
d
e X
In Are
a
B3
RSSI Nod
e A
(dB
m
)
RSSI Nod
e B
(dB
m
)
RSSI Nod
e C
(dB
m
)
FTP
-
5
6
/
-
62
-
5
2
/
-
58
-
5
0
/
-
60
MT
P
Mod
el
-
1
-
5
5
/
-
64
-
5
5
/
-
62
-
5
2
/
-
57
MT
P
Mod
el
-
2
-
5
5
/
-
65
-
5
4
/
-
61
-
5
2
/
-
62
MT
P
Mod
el
-
3
-
5
5
/
-
62
-
5
5
/
-
62
-
5
2
/
-
55
4.
CONCL
US
I
O
N
Ind
oor
posit
io
ning
with
the
f
ing
e
rprintin
g
m
et
ho
d
is
ve
ry
eff
ect
ive
a
nd
accurate
w
he
n
us
e
d
in
sm
all
indoor
ar
eas.
Eff
ic
ie
ncy
in
the
trai
ni
ng
ph
ase
will
i
m
pr
ove
the
perf
orm
ance
of
t
he
fin
gerpr
i
nting
m
et
ho
d.
Tw
o
m
et
ho
ds
hav
e
bee
n
us
e
d
du
rin
g
the
t
r
ai
nin
g
pha
se,
nam
ely
MTP
(
m
ov
ing
ta
r
get
par
am
et
er)
an
d
FTP
(f
ixe
d
ta
r
ger
pa
ram
et
er)
.
The
MTP
m
et
ho
d
o
n
t
he
trai
ning
ph
ase
of
fin
ge
rprintin
g
can
i
ncr
ease
t
he
ef
f
ic
ie
ncy
of
the
pr
ocess
du
rati
on
in
c
ollec
ti
ng
RSSI
data
fr
om
eac
h
sens
or
Node
in
the
m
app
in
g
area
by
16%
o
r
5
seco
nd
s
f
ast
e
r t
han
us
i
ng the
FTP
m
et
ho
d.
A
s
a
sug
gestio
n
f
or
furthe
r
re
search
,
f
or
e
xa
m
ple
the
Node
X
m
ov
em
ent
schem
e
can
be
adjusted
t
o
the
co
ndit
ion
s o
f
t
he
m
app
ing
area,
s
o
that
al
l
areas
can b
e
e
xceed
e
d
a
nd
al
l
RSSI
f
ro
m
each
se
ns
or N
od
e
ca
n
be
m
e
asur
ed
i
n
one
m
ov
em
e
nt
m
od
el
,
su
ch
as
zi
g
-
zag
or
li
ke
a
sine
wa
ve
.
F
or
t
he
pur
po
s
es
of
preci
s
ion
in
the posit
io
ning
stage,
m
or
e th
an 3
un
it
s
of se
ns
or
ref
e
re
nce
Nodes a
re
nee
ded
.
Evaluation Warning : The document was created with Spire.PDF for Python.
Ind
on
esi
a
n
J
E
le
c Eng
&
Co
m
p
Sci
IS
S
N:
25
02
-
4752
Impr
oved fi
ng
e
rp
rinti
ng
perf
orma
nce
in
i
ndoor
posit
ion
in
g by re
duci
ng
dura
ti
on
of the…
(
And
ik
a M
uharam
)
243
ACKN
OWLE
DGE
MENTS
The
a
uthors
w
ou
l
d
li
ke
to
t
ha
nk
U
niv
e
rsita
s
Me
rcu
B
uana,
a
nd
the
Ind
on
e
sia
n
Highe
r
Ed
ucati
on
(Rist
ekBRIN
)
for
s
upportin
g
this
re
searc
h.
W
e
a
ppreciat
e
your
s
uppo
r
t
for
c
omm
ent
s
or
s
uggestio
ns
t
o
i
m
pr
ove this
re
search
.
REFERE
NCE
S
[1]
W
.
W
ebb,
"
Wir
e
le
ss
Comm
unic
a
ti
ons: The
Fut
ur
e
,"
Chi
che
st
er,
UK
:
John W
il
e
y
&
Sons
,
L
td, 20
07.
[2]
X.
Fang,
Z.
Jia
ng,
L
.
Nan
,
and
L.
Chen
,
“
Opti
m
al
weighted
K
-
nea
rest
nei
ghbo
ur
al
gor
it
hm
for
wire
l
ess
sensor
net
work
fing
erp
rint
lo
ca
l
isat
ion
in
nois
y
env
iron
m
ent
,
”
I
ET
Comm
un.
,
vol
.
12,
n
o.
10,
pp
.
11
71
-
1177,
2018,
do
i:
10.
1049/iet
-
com
.
2017.
0515
.
[3]
A.
Adria
ns
y
ah, H.
Suw
o
y
o,
Y.
Ti
an
,
“
Im
proving
the
wa
ll
-
fol
lo
wing
robot
per
fo
rm
anc
e
using
PI
D
-
PS
O
Control
le
r,
"
Jurnal
Tekno
log
i
,
vo
l. 81,
no.
3,
pp.
119
-
126
,
20
19,
doi
:
10
.
1111
3/j
t.
v81
.
13098
.
[4]
Z.
Farid,
R.
Nor
din,
and
M.
Ismai
l
,
“
Rec
ent
adv
anc
es
in
wire
l
ess
indoor
loc
al
izati
on
techniqu
es
and
sy
st
em,”
J.
Comput.
Ne
tworks
Comm
un.
,
vo
l
.
2013
,
2013
,
doi
:
10.
1155
/2013/185138.
[5]
J.
La
rr
ana
g
a,
L.
Muguira,
J.
M.
Lope
z
-
Gard
e,
an
d
J.
I.
Va
zquez,
“
An
envi
ronm
ent
ada
p
ti
ve
Z
igB
ee
-
base
d
indoo
r
positi
oning
al
go
rit
hm
,
”
2010
Int
.
Conf
.
Indoor
Posit
ion.
Indoor
Navi
g
.
IPI
N
20
10
-
Conf
.
Proc
.
,
no.
Septe
m
ber
,
2010,
pp
.
15
-
17
,
doi: 10.
1109
/IPIN.2010.
5647828
.
[6]
R.
B.
Bah
awe
re
s,
S.
Mokogint
a,
and
M
.
Ala
y
dr
us,
“
Com
par
ison
of
m
et
hods
for
l
ocalizing
th
e
so
urc
e
posi
ti
on
o
f
dea
uth
ent
i
ca
t
ion
at
t
ac
ks
on
W
AP
802.
11n
using
Chana
l
y
z
er
and
W
i
-
Sp
y
2
.
4x,
”
Journal
of
Phy
s
ic
s:
Confe
r
enc
e
Serie
s
,
Int
ern
ati
onal
Confer
ence
on
Com
puti
ng
and
Applie
d
Inform
at
ic
s,
Me
dan,
Indone
si
a,
vol.
801,
201
6,
doi:
10.
1088
/
174
2
-
6596/801/
1/01
2056.
[7]
A.
Singh,
S.
Kum
ar,
and
O.
Kaiwa
rt
y
a
,
“
A
H
y
brid
Lo
ca
l
iza
ti
on
Algorit
hm
for
W
ire
le
ss
Sensor
Networks,”
Proce
dia
Comput.
S
ci.
,
vol. 57, p
p.
1432
-
1439
,
2
015,
doi
:
10
.
101
6/j
.
pro
cs.
2015.
0
7.
464.
[8]
A.
W
hee
ler,
“
C
om
m
erc
ia
l
app
lications
of
wire
les
s
sensor
net
works
using
Zi
gBe
e
,
”
I
EE
E
Comm
un.
Mag.
,
vol
.
45
,
no.
4
,
pp
.
70
-
77
,
2007,
doi: 10.
11
09/MCOM
.
2007.
343615.
[9]
R.
F.
Brena
,
J.
P.
Garc
ía
-
Vá
z
quez
,
C
.
E.
Ga
lvá
n
-
T
ej
ad
a,
D.
Muñoz
-
Rodrigue
z,
C
.
Varga
s
-
Rosale
s,
and
J
.
Fangm
e
y
er
,
“
E
volut
ion
of
Ind
oor
Pos
it
ioni
ng
Te
chno
logi
es:
A
Surve
y
,
”
J
.
Sensors
,
vol.
2017,
2017,
do
i:
10.
1155/2017/
26
30413.
[10]
R.
M.
Dell
osa
,
A.
C.
Faja
rdo,
and
R.
P.
Medina
,
“
Modifie
d
finge
rprinting
loc
a
li
z
at
ion
t
echnique
of
indo
or
positi
oning
s
y
ste
m
base
d
on
coor
dina
t
es,
”
Indon
e
s.
J.
Elec
tr.
Eng
.
Comput.
S
ci
.
(
IJE
ECS)
,
vol.
15,
no.
3,
pp.
1345
-
1355,
2019
,
doi
: 10.11591/
i
je
e
cs.
v15.
i3.
pp1345
-
1
355.
[11]
S.
Subedi
and
J.
Y.
P
y
un,
“
Prac
tical
Fing
erp
rintin
g
Loc
a
li
z
at
ion
f
or
Indoor
Pos
it
ioni
ng
S
y
stem
b
y
Us
ing
Bea
cons,
”
J.
S
ensors
,
vol
.
2017,
2017
,
doi
: 10.1155/
2017/97
42170.
[12]
X.
W
ang,
L.
Ga
o,
S.
Mao,
and
S.
Pande
y
,
“
Dee
pFi:
Dee
p
learni
ng
for
indoor
finge
rprint
ing
using
cha
nnel
st
ate
informati
on,
”
2
015
IEEE
Wire
l.
Comm
un.
N
e
tw.
Conf
.
WCN
C
2015
,
no
.
A
pril
,
pp.
1666
-
1671,
2015,
do
i:
10.
1109/WCN
C.
2015.
7127718.
[13]
C.
Kum
ar
and
K.
Raj
aw
at
,
“
Dicti
onar
y
le
a
rning
b
ase
d
finge
rpr
int
i
ng
for
indoor
localiza
t
ion,”
2018
24th
Na
tl
.
Conf
.
Comm
un.
NCC
2018
,
2019
,
doi
: 10.1109/
NCC.2
018.
8600195.
[14]
M.
Uradz
inski,
H.
Guo,
X.
Li
u,
and
M.
Yu,
“
A
dvanc
ed
Indoor
Pos
it
ioni
ng
Us
ing
Zi
gbee
W
irele
ss
Te
chnol
og
y
,
”
Wirel
.
P
ers.
Co
mm
un.
,
vol
.
97
,
no.
4
,
pp
.
6509
-
6518,
2017
,
doi
:
10.
1007/s11277
-
017
-
4852
-
5.
[15]
C.
W
.
Ou
et
a
l.
,
“
A
Zi
gBee
posi
t
ion
t
ec
hniqu
e
fo
r
indoor
loc
a
liza
ti
on
b
ase
d
on
pr
oximit
y
learni
ng
,
”
2017
IE
EE
In
t
.
Conf.
M
ec
ha
tronic
s A
u
tom.
ICM
A
2017
,
pp.
875
-
880,
2017
,
doi
:
1
0.
1109/ICMA.2
017.
8015931.
[16]
P.
Fons
eka
and
K.
Sandra
seg
ar
an,
“
Indoor
lo
calizat
ion
for
IoT
appl
i
ca
t
ions
using
finge
rpr
int
in
g,
”
I
EE
E
World
Forum
Inte
rnet
Things
,
WF
-
I
oT
2018
-
Proc.
,
vo
l.
2018
-
Janua
,
2018
,
p
p.
736
-
741,
do
i:
10.
1109
/WF
-
IoT.
2018.
83551
05.
[17]
D.
Alsham
aa
,
F
.
Mourad
-
Cheh
a
de,
and
P
.
Honeine,
“
Loc
a
li
z
ati
on
of
sensors
in
indoor
wire
l
e
ss
net
works
:
A
n
observa
ti
on
m
odel
using
W
iFi
R
SS
,
”
2018
9th
IFI
P
Int.
Conf.
Ne
w
Technol
.
Mobi
l.
Sec
ur.
NT
MS
2018
-
Proc.
,
vol
.
2018
-
Janua
,
201
8,
pp
.
1
-
5
,
doi
:
1
0.
1109/NTMS.2
018.
8328699.
[18]
I.
Nurhai
d
a
et
al.
,
“
Im
ple
m
e
nta
ti
on
o
f
Dee
p
Le
arn
ing
Pre
dic
tor
(
LSTM)
Algorit
hm
for
Hum
an
Mobilit
y
Predic
ti
on
,
”
Int
.
J.
In
te
rac
t. Mob. Tec
hnol
.
,
vol
.
1
4,
no
.
18
,
pp
.
13
2
-
144,
2020
,
doi
:
10.
3991
/i
j
im.v14i18.
16867.
[19]
Z.
He,
Y.
Ma,
and
R.
Ta
fa
zo
ll
i,
“
A
h
y
brid
dat
a
fusion
bas
ed
coope
ra
ti
v
e
loc
a
li
z
at
ion
app
roa
ch
for
cellular
net
works
,
”
W
irel
.
Comm
un.
Mob.
Comput.
Conf.
(
IWCMC)
,
2011
7th
Int.
,
2011,
p
p.
162
-
166,
d
o
i
:
10.
1109/IWCM
C.
2011.
5982409
.
[20]
T.
Alhm
ie
d
at a
n
d
A.
Abu Sa
le
m
,
“
A h
y
brid
ran
g
e
-
fre
e localization
al
gor
it
hm
for Zi
gBee
wir
eless
sensor ne
t
works
,
”
Int.
Arab
J
.
In
f.
Technol
.
,
vo
l. 14
,
no
.
4A Spe
ci
a
l Is
sue,
pp.
647
-
6
53,
2017
.
[21]
Sandee
p
Verm
a,
“
Network
Top
ologi
es
in
W
ire
l
ess
Sensor
Networks:
A
Revi
ew
1,
”
Int.
J.
E
l
ec
tron.
Comm
un.
Technol
.
,
vo
l. 4,
no.
3
,
pp
.
1
-
5
,
2
013,
doi
:
10
.
1.
1
.
308.
796.
[22]
S.
Šafa
rić
and
K.
Mala
r
ić,
“
Zi
gB
ee
wire
le
ss
stan
dar
d,
”
Proc.
Elm
ar
-
Int.
Symp.
El
ectron.
Mar.
,
no.
June,
pp.
25
9
-
262,
2006
,
doi
:
1
0.
1109/ELMA
R.
2006.
329562.
[23]
S.
S.
R.
Aham
ed,
“
The
role
of
zi
gbee
te
chno
l
og
y
in
futur
e
d
at
a
comm
unic
a
t
ion
s
y
stem,”
J.
Theor.
Ap
pl.
Inf
.
Technol
.
,
vo
l. 5,
pp.
129
-
135
,
20
09.
[24]
G.
Venka
t
ara
m
a
n,
Z
.
Feng
,
J.
Hu,
and
P.
Li,
“
Com
bina
tori
a
l
a
lg
orit
hm
s
for
fast
cl
ock
m
esh
optim
iz
at
ion
,
”
IEEE
Tr
ans.
Ve
ry
Lar
ge
Sca
le Int
egr
.
Syst.
,
vol
.
18
,
no
.
1
,
pp
.
131
-
141
,
2010,
doi: 10.
11
09/T
VLSI.2008.2
007737.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2502
-
4752
Ind
on
esi
a
n
J
E
le
c Eng &
Co
m
p
Sci,
Vo
l.
2
4
, N
o.
1
,
Oct
o
ber
20
21
:
23
6
-
24
4
244
[25]
E
.
Nav
arr
o,
B.
Peuker
,
M.
Quan
,
A.
C.
Cla
rk
,
an
d
J.
Jipson,
“
W
i
-
Fi
Local
i
za
t
ion
Us
ing
RS
SI
Fin
ger
printing,
”
Test
,
pp.
1
-
6
,
2010
.
[26]
L.
Pei
,
R.
Ch
e
n,
J.
L
iu,
H.
Kuus
nie
m
i,
T.
Te
nhunen
,
and
Y.
Chen,
“
Us
ing
inqui
r
y
-
base
d
Blue
toot
h
RS
SI
proba
bil
i
t
y
d
ist
r
ibut
ions
for
indoor
positi
oning,”
J.
Glob.
Position.
Syst.
,
vol.
9,
no.
2,
pp.
12
2
-
130,
2010,
doi:
10.
5081/j
gps.9
.
2
.
122.
[27]
T.
Dag
and
T.
Ars
an,
“
Rec
ei
v
e
d
signal
strengt
h
base
d
le
ast
squ
are
s
later
a
ti
on
a
l
gorit
hm
for
indoor
loc
aliz
at
ion
,
”
Comput.
E
lectr.
Eng
.
,
vol
.
66
,
pp
.
114
-
126
,
2018
,
doi: 10.
1016
/j
.
c
om
pel
ec
eng
.
201
7.
08.
014
.
BIOGR
AP
HI
ES OF
A
UTH
ORS
An
dik
a
Mu
haram
rec
e
ive
d
a
bac
he
lor
'
s
degr
e
e
in
telec
om
m
unic
a
ti
ons
engi
n
e
eri
ng
in
2011
from
the
Instit
u
t
Te
knolog
i
T
el
k
om
,
Bandung,
I
ndonesia
.
In
202
0
he
re
ce
iv
ed
a
m
aste
r'
s
degr
ee
in
el
e
ct
ri
cal
enginee
ring
from
Univer
sit
as
Merc
u
Buana
,
Jak
art
a
,
I
ndonesia
.
He
ha
s
an
int
ere
st
in
the
fi
el
d
of
embedde
d
s
y
s
te
m
s
t
hat
use
m
ic
ro
co
ntrol
lers
such
as
the
int
ern
et
of
t
hings
(IoT
)
or
aut
om
at
ed
s
y
st
e
m
s.
Ab
di
Waha
b
is
a
rese
ar
che
r
a
t
Univer
sita
s
Me
rcu
Buana
Jak
ar
ta
.
He
is
in
te
r
este
d
in
m
obil
e
computing,
emb
edde
d
s
y
s
te
m
,
and
m
obil
e
pro
gra
m
m
ing
rese
a
rh
are
a
.
For
no
w,
he
is
a
lso
int
er
esti
ng
in
m
a
chi
ne
learni
ng
in
fields of
f
ina
n
ce.
Mudrik
Alay
dr
us
is
profe
ss
or
at
Univer
si
ta
s
Merc
u
Buana,
J
aka
rt
a.
His
res
e
arc
h
in
te
r
ests
ar
e
Num
eri
ca
l
Elec
tromagnet
i
cs
appl
ie
d
in
an
te
n
na
design,
m
icrow
ave
devi
c
es
and
inve
rse
proble
m
s.
Evaluation Warning : The document was created with Spire.PDF for Python.