TELK
OMNIKA
T
elecommunication,
Computing,
Electr
onics
and
Contr
ol
V
ol.
19,
No.
1,
February
2021,
pp.
51
62
ISSN:
1693-6930,
accredited
First
Grade
by
K
emenristekdikti,
No:
21/E/KPT/2018
DOI:
10.12928/TELK
OMNIKA.v19i1.16275
r
51
Smartphone
indoor
positioning
based
on
enhanced
BLE
beacon
multi-lateration
Ngoc-Son
Duong,
Thai-Mai
Dinh
Thi
F
aculty
of
Electronics
and
T
elecommunications,
VNU
Uni
v
ersity
of
Engineering
and
T
echnology
,
V
ietnam
National
Uni
v
ersity
,
Hanoi,
V
ietnam
Article
Inf
o
Article
history:
Recei
v
ed
Apr
7,
2020
Re
vised
Aug
4,
2020
Accepted
Sep
5,
2020
K
eyw
ords:
BLE
iBeacon
Bluetooth
lo
w
ener
gy
Indoor
positioning
Least
square
estimation
T
rilateration
ABSTRA
CT
In
this
paper
,
we
introduce
a
smartphone
indoor
positioning
method
using
bluetooth
lo
w
ener
gy
(BLE)
beacon
multi
lateration.
At
first,
based
on
si
gnal
strength
analysis,
we
construct
a
distance
calculation
model
for
BLE
beacons.
Then,
with
the
aims
to
impro
v
e
positioning
accurac
y
,
we
propose
an
impro
v
ed
lateral
method
(range-based
method)
which
is
applied
for
4
nearby
beacons.
The
method
is
intended
to
design
a
real-time
system
for
some
services
such
as
emer
genc
y
assistance,
personal
localiza-
tion
and
tracking,
location-based
adv
ertising
and
mark
eting,
etc.
Experimental
results
sho
w
that
the
proposed
method
achie
v
es
high
accurac
y
when
compared
with
the
state
of
the
art
lateral
methods
such
as
geometry-based
(con
v
entional
trilateration),
least
square
estimation-based
(LSE-based)
and
weighted
LSE-based.
This
is
an
open
access
article
under
the
CC
BY
-SA
license
.
Corresponding
A
uthor:
Thai-Mai
Dinh
Thi
F
aculty
of
Electronics
and
T
elecommunications
VNU
Uni
v
ersity
of
Engineering
and
T
echnology
,
V
ietnam
National
Uni
v
ersity
G2
Building,
144
Xuan
Thuy
Str
.,
Cau
Giay
Dist.,
Hanoi,
V
ietnam
Email:
dttmai@vnu.edu.vn
1.
INTR
ODUCTION
No
w
adays,
in
lar
ge
cities,
human
acti
vi
ties
tend
to
shift
from
outdoor
to
indoor
en
vironments.
This
has
led
to
a
gro
wing
need
for
services
related
to
the
indoor
en
vironment,
such
as
location-based
services
(LBSs)
and
social
netw
orking
services
(SNSs).
Location
accurac
y
is
a
measure
of
service
quality
.
GPS
has
done
this
well
for
outdoor
en
vironments.
Ho
we
v
er
,
due
to
the
obstruction
of
b
uilding
materials,
GPS
signals
can
not
w
ork
well
in
indoor
en
vironments.
Therefore,
man
y
technologies
are
utilized
to
deplo
y
indoor
positioning
sys-
tems
(IPS)
such
as
W
iFi
[1,
2],
radio
frequenc
y
identification
(RFID)
[3],
Zigbee
[4],
ultra-wideband
(UWB)
[5]
and
camera-based
(photo-based)
[6].
T
o
o
v
ercome
the
limitations
of
pre
vious
technologies,
a
bluetooth
lo
w
ener
gy
(BLE)
based
technology
called
iBeacon
w
as
introduced
as
an
appropriate
solution
for
IPS
requirements
due
to
the
adv
antages
such
as
lo
w
ener
gy
consumption,
wide-co
v
erage,
easy
deplo
yment,
and
potential
high
ac-
curac
y
.
W
e
can
use
iBeacon
technology
to
b
uild
an
indoor
positioning
system
that
uses
recei
v
ed
signal
strength
(RSS)
to
estimate
user
location.
There
are
tw
o
kinds
of
recei
v
ed
signal
strength
inde
x
(RSSI)
based
technique:
fingerprinting
[7]
and
range-based
method
[8]
(as
kno
wn
as
lateral
or
lateration
method).
In
the
localization
problem,
the
range-based
method
utilizes
an
estimated
distance
from
the
path-loss
model
to
estimate
the
user’
s
position.
Meanwhile,
Fingerprinting
relies
on
map
surv
e
y
steps
to
b
uild
an
radio
signal
strength
(RSS)
database
of
an
interested
area.
Then,
the
position
decision
is
made
based
on
online
signals
and
of
fline
database
using
a
matching
algorithm.
Some
recent
studies
ha
v
e
chosen
fingerprint
as
the
main
approach
[9–12].
Others
choose
J
ournal
homepage:
http://journal.uad.ac.id/inde
x.php/TELK
OMNIKA
Evaluation Warning : The document was created with Spire.PDF for Python.
52
r
ISSN:
1693-6930
range-based
as
their
main
approach
[13–15].
Looking
at
the
number
of
current
research,
we
can
see
that
fingerpr
inting
seems
to
be
more
popular
than
range-based
methods.
Ho
we
v
er
,
to
achie
v
e
high
accurac
y
,
the
fingerprinting
method
requires
data
collec-
tion
for
man
y
reference
points.
This
task
tak
es
much
time
and
is
unfeasible
to
practical
implementation.
In
deplo
yment,
it
seems
f
air
to
say
that
range-based
methods
are
more
feasible
than
fingerprinting.
Range-based
methods
can
help
implement
an
indoor
positioning
system
without
additional
requirements
such
as
map
surv
e
y
or
reb
uild
the
database.
Ho
we
v
er
,
range-methods
ha
v
e
their
disadv
antages
as
well.
Their
problem
is
distance
estimation,
which
directly
af
fects
the
accurac
y
of
the
estimat
ed
position.
Theoretically
,
we
all
kno
w
the
dis-
tance
v
aries
according
to
the
log
arithmic
function.
But,
it
is
not
easy
to
calculate
the
distance
correctly
due
to
the
ef
fects
of
f
ading,
small-scale,
and
human
absorption.
T
o
cope
with
thi
s
problem,
we
propose
an
impro
v
ed
me
thod
for
a
range-based
method
applied
to
BLE
signals
and
indoor
en
vironments.
W
e
consider
that,
at
a
certain
RSS
le
v
el,
the
estimation
of
the
distance
between
the
phone
and
the
BLE
beacon
is
relati
v
ely
accurate.
So,
such
beacons
are
called
reliable
beacons.
Therefore,
we
obtain
some
reliable
circles
with
the
center
of
the
reliable
beacons
and
the
radius
of
the
estimated
distance
from
the
mobile
de
vice.
Then
the
returned
position
must
be
in
such
circles.
Interv
ention
by
geometric
method,
we
mo
v
e
the
estimated
position
of
con
v
entional
trilateration
to
a
position
that
belongs
t
o
the
circles.
In
order
to
get
higher
accurac
y
,
we
e
xploit
the
information
of
a
lar
ge
number
of
beacons.
In
this
study
,
we
applied
the
proposed
method
for
four
beacons
simultaneously
.
Each
cluster
of
three
beacons
will
estimate
the
phone’
s
position
based
on
T
rilateration
combined
with
reliable
circles.
Then
the
final
position
will
be
determined
by
the
a
v
erage
of
possible
positions.
Experiments
were
carried
out
in
the
real
w
orld,
and
the
res
ults
sho
wed
that
the
proposed
method
outperformed
e
xisting
con
v
entional
methods.
In
the
ne
xt
part,
the
T
rilateration
method
and
its
e
n
c
ou
nt
ered
problem,
are
briefly
presented
in
section
2.
Section
3
describes
our
proposed
method,
including
an
impro
v
ed
geometry-based
method
for
T
rilateration
and
multiple
BLE
beacon
usage.
Section
4
pro
vides
system
parameters
and
e
xperimental
results.
Ultimately
,
section
5
concludes
this
paper
.
2.
RESEARCH
METHOD
2.1.
T
rilateration
In
the
global
positioning
system,
trilateration
[16,
17]
(short
for
con
v
entional
trilateration)
is
a
tradi-
tional
method
for
determining
the
location
of
recei
v
er
equipment
on
earth.
The
position
of
the
object
can
be
obtained
by
calculating
the
distance
from
the
satellites.
W
e
can
e
xploit
this
concept
for
indoor
localization
by
scaling
do
wn
the
trilateration
concept
used
for
global
positioning
system
(GPS).
In
our
study
,
trilateration,
as
illustrated
in
Figure
1,
is
defined
as
a
method
to
obtain
the
position
of
an
object
or
people
under
the
influence
of
the
indoor
en
vironment
based
on
RSS
information
of
three
beacons.
Recei
v
ed
signal
strengths
from
these
beacons
are
calculated
via
the
follo
wing
formula
[18]:
(
d
)
=
(
d
0
)
10
log
d
d
0
(1)
whereas,
(
d
)
and
(
d
0
)
are
RSSIs
at
Euclidean
distance
d
and
reference
distance
d
0
,
respecti
v
ely
(in
dBm).
d
0
is
usually
chosen
equal
to
one
meter
for
the
indoor
en
vironment
and
represents
the
path
loss
e
xponent.
The
distance
from
the
smart
phone
to
the
i
-th
beacon
can
be
e
xpressed
as:
d
i
=
d
0
10
(
d
0
)
(
d
i
)
10
(2)
In
order
to
calculate
the
smart
phone
position,
the
coordinates
of
BLE
beacons
on
the
map
must
be
kno
wn
in
adv
ance.
Assume
that
(
x
i
;
y
i
)
is
the
coordinates
of
i
th
beacon
on
the
map.
The
equation
for
each
beacon
re
gion
is
represented
by
(
z
=
0
):
(
x
x
i
)
2
+
(
y
y
i
)
2
=
d
2
i
;
i
=
f
1
;
2
;
3
g
(3)
(3)
is
equi
v
alent
to:
x
2
2
xx
i
+
x
2
i
+
y
2
2
y
y
i
+
y
2
i
d
2
i
=
0
(4)
TELK
OMNIKA
T
elecommun
Comput
El
Control,
V
ol.
19,
No.
1,
February
2021
:
51
–
62
Evaluation Warning : The document was created with Spire.PDF for Python.
TELK
OMNIKA
T
elecommun
Comput
El
Control
r
53
Let
a
i
=
2
x
i
,
b
i
=
2
y
i
,
c
i
=
x
2
i
+
y
2
i
d
2
i
,
then
(4)
is
re
written
as:
x
2
+
y
2
+
a
i
x
+
b
i
y
+
c
i
=
0
(5)
The
position
of
a
track
ed
object
can
be
estimated
by:
^
x
=
(
c
2
c
1
)(
b
2
b
3
)
(
c
3
c
2
)(
b
1
b
2
)
(
a
1
a
2
)(
b
2
b
3
)
(
a
2
a
3
)(
b
1
b
2
)
(6)
^
y
=
(
a
2
a
1
)(
c
3
c
2
)
(
c
2
c
1
)(
a
2
a
3
)
(
a
1
a
2
)(
b
2
b
3
)
(
a
2
a
3
)(
b
1
b
2
)
(7)
B1
B
2
B
3
B
1
B2
B
3
(
b)
(
c)
B1
B
2
B
3
(
a)
Figure
1.
T
rilateration
method;
(a)
ideal
condition,
3
circles
intersect
at
one
point,
(b)
and
(c)
imperfect
condition,
3
circles
do
not
intersect
at
one
point
2.2.
Distance
estimation
Distance
estimation
plays
a
vital
role
in
internet
of
things
(IoT)
applications.
In
this
study
,
the
accurac
y
of
distance
estimation
has
a
high
impact
on
reck
oning
trilateration
position.
The
more
accurate
the
con
v
ersion
from
RSSI
to
distance
is,
the
more
precise
trilateration’
s
estimated
position
is.
Unfortunately
,
for
BLE
signals,
it
is
not
easy
to
calculate
true
distance
through
the
log-distance
path
loss
model.
Man
y
f
actors
af
fect
the
BLE
signal
in
the
indoor
en
vironment,
such
as
the
material
of
indoor
structures,
body-blocking.
Intending
to
construct
a
real-time
indoor
positioning
system,
we
consider
humans’
presence
as
a
high-impact
f
actor
.
The
human
body
consists
of
70%
w
ater;
therefore,
this
is
a
strong
attenuation
f
actor
at
a
frequenc
y
of
2.4
GHz.
The
attenuation
of
the
BLE
signal
caused
by
humans
body
(in
dBm)
is
estimated
by
[19,
20]:
PL
H
=
30
+
10
log
(2
w
n
)
(8)
where
w
and
n
are
body
weight
and
number
of
detected
person
in
the
area.
As
sho
wn
in
Figure
2,
we
can
see
a
significant
dif
ference
when
the
phone
is
block
ed
by
the
user
body
.
Thus,
use
(2)
with
a
fix
ed
to
estimate
the
distance
no
longer
matches
the
BLE
signal.
T
o
look
closely
at
the
problem,
we
conducted
some
e
xperiments
to
analyze
the
RSSI
atte
nu
a
tion
of
the
BLE
signal
in
2
cases:
line-of-sight
(LOS)
and
non-LOS
(NLOS)
(body-blocking).
Figure
3
sho
ws
that
whene
v
er
we
recei
v
e
the
signal
that
has
an
RSS
v
alue
lar
ger
than
-70
dBm,
we
can
use
LOS
curv
e
line
confidently
to
calculate
the
distance
from
the
smartphone
to
the
beacon.
F
or
our
empirical
data,
LOS
curv
e
line
is
represented
by
a
log
arithmic
trendline
[21],
i.e.:
=
10
:
18
ln(
d
)
59
:
533
(dBm)
(9)
where,
is
RSSI
at
distance
d
.
At
lo
wer
RSSI
le
v
els,
a
RSS
v
alue
can
correspond
to
man
y
distances.
F
or
e
xample,
in
our
e
xperimental
en
vironment,
-75
dBm
can
corresponds
to
either
3
m
(LOS)
or
4.5
m
(NLOS)
while
the
smartphone
does
not
kno
w
what
type
of
condition
it
is
confronting.
So,
it
is
considered
as
an
unre-
liable
situation.
In
this
case,
the
distance
is
estimated
by
an
a
v
erage
fitted
curv
e
between
LOS
and
NLOS,
i.e:
=
8
:
621
ln(
d
)
64
:
321
.
Smartphone
indoor
positioning
based
on
enhanced
BLE
beacon...
(Ngoc-Son
Duong)
Evaluation Warning : The document was created with Spire.PDF for Python.
54
r
ISSN:
1693-6930
2.3.
RSSI
filtering
As
precise
distance
estimation
results
in
proper
T
rilateration
localization,
it
w
ould
be
better
to
use
some
RSS
filtration.
F
or
RSS,
noise
is
often
kno
wn
for
short-te
rm
f
ading
[22],
which
is
caused
by
surrounding
pedestrians.
RSS
can
fluctuate
sharply
in
a
short
period.
T
o
reduce
its
impact,
we
apply
Kalman
filter
[23]
for
RSS
model:
(
k
=
k
1
z
k
=
k
+
v
k
(10)
herein,
we
consider
there
is
no
noise
in
the
process
and
v
k
denotes
measurement
noise
of
RSS
observ
ation
z
k
which
is
introduced
as
pedestrian’
s
mo
v
ement.
The
detailed
process
of
the
Kalman
filter
algorithm
is
sho
wn
in
Algorithm
1.
Figure
2.
Attenuation
of
recei
v
ed
signal
strength
at
2
m
of
distance.
0
corresponds
to
the
situation
that
the
user
stands
and
holds
the
phone
the
opposite
side
to
the
beacon
(LOS).
180
corresponds
to
the
case
that
the
user
turns
a
w
ay
from
the
beacon
(non-LOS)
1
2
3
4
5
6
7
8
9
10
Distance (m)
-85
-80
-75
-70
-65
-60
RSSI (dBm)
LOS condition
NLOS condition
Fitted curve of LOS condition
Fitted curve of NLOS condition
Figure
3.
RSSI
v
aries
by
distance
in
2
cases:
LOS
and
NLOS
at
transmitting
po
wer
of
0
dBm.
Note
that,
this
is
the
empirical
data
and
needs
to
be
calibrated
with
another
en
vironment
TELK
OMNIKA
T
elecommun
Comput
El
Control,
V
ol.
19,
No.
1,
February
2021
:
51
–
62
Evaluation Warning : The document was created with Spire.PDF for Python.
TELK
OMNIKA
T
elecommun
Comput
El
Control
r
55
Algorithm
1:
Kalman
Filter
Algorithm
Initialize:
State
mean:
0
=
z
0
State
co
v
ariance:
0
=
1
1.
Predict
state:
^
k
=
^
k
1
2.
Predict
state
co
v
ariance:
^
k
=
k
1
3.
Calculate
Kalman
filter
g
ain:
K
k
=
k
(
k
+
R
)
1
4.
Update
state:
^
k
=
^
k
+
K
k
(
z
k
^
k
)
5.
Update
state
co
v
ariance:
k
=
(1
K
k
)
k
3.
ENHANCED
BLE
BEA
CON
MUL
TI-LA
TERA
TION
3.1.
Enhanced
trilateration
f
or
BLE
signal
Based
on
the
abo
v
e
signal
analysis,
whene
v
er
the
signal
strength
recei
v
ed
from
the
beacon
is
h
i
gher
than
-70
dBm,
the
relati
v
e
position
of
the
phone
to
the
beacon
is
lik
ely
to
be
LOS.
F
or
simplicity
,
we
call
the
beacon
that
satisfies
the
condition
with
the
RSSI
returned
on
the
phone
higher
than
-70
dBm
as
rLOS
beacon
and
i
s
denoted
by
B
.
(abbre
viation
for
reliable
LOS
beacon).
Then,
the
estimated
distance
of
rLOS
beacon
from
(9)
is
considered
the
most
reliable.
Moreo
v
er
,
the
position
is
estimated
by
trilater
ation
completely
depends
on
the
radius
of
three
circles.
In
theory
,
if
the
distance
from
the
beacons
to
the
smartphone
is
absolutely
correct,
the
position
returned
from
t
rilateration
must
be
the
intersection
of
the
three
circles.
Let
B
.
be
the
coordinates
of
rLOS
beacon,
T
be
the
coordinates
returned
by
trilateration
and
P
is
the
position
calculated
by
the
proposed
method.
Case
1:
There
is
no
presence
of
rLOS
beacon.
In
this
case,
there
will
not
be
an
y
impro
v
ement,
we
simply
apply
(6),
(7)
to
estimate
the
user
position.
Case
2:
Only
one
rLOS
beacon
is
a
v
ailable.
S
ince
there
is
only
one
rLOS
beacon,
that
means
we
ha
v
e
only
one
trusted
circle
in
total
of
three,
called
rLOS
circle.
Therefore,
the
estimated
T
rilateration-based
position
should
be
returned
on
this
circle.
Then,
P
is
defined
as
the
intersection
of
infinite
line
!
B
.
T
and
a
rLOS
circle
which
has
radius
of
d
.
,
centered
at
B
.
and
is
denoted
by
(
B
.
;
d
.
)
.
Figure
4
is
the
visual
vie
w
of
the
proposed
method.
P
must
satisfy
the
follo
wing
conditions:
(
P
=
!
B
.
T
\
(
B
.
;
d
.
)
PT
is
minimum
(11)
Estimated position of
the
proposed method
Estimated position of
Trilateration
rLOS beacon
(a)
(b)
rLOS circle
B
1
B
2
B
3
P
T
B
1
T
P
B
2
B
3
.
.
Figure
4.
Position
estimation
in
Case
2
Case
3:
T
w
o
rLOS
beacons
e
xist.
Let
tw
o
circles
of
radii
d
.
1
and
d
.
2
and
centered
at
B
.
1
and
B
.
2
intersect
at
one
or
tw
o
point.
Figure
5
illustrates
ho
w
we
estimate
the
user
position.
In
this
case,
P
must
satisfy
the
follo
wing
conditions:
(
P
=
T
2
i
=1
(
B
.
i
;
d
.
i
)
PT
is
minimum
(12)
In
addition,
we
consider
the
case
of
tw
o
rLOS
circle
does
not
intersect
as
a
case
is
described
in
Figure
4
where
the
position
is
decided
by
a
smaller
rLOS
circle.
Because
iBeacon
is
uniformly
distrib
uted,
the
smartphone
rarely
gets
the
RSS
abo
v
e
-70
dBm
simultaneously
from
three
dif
ferent
beacons
due
to
the
body’
s
obstruction;
therefore,
there
is
no
Case
4
for
three
rLOS
beacons.
Smartphone
indoor
positioning
based
on
enhanced
BLE
beacon...
(Ngoc-Son
Duong)
Evaluation Warning : The document was created with Spire.PDF for Python.
56
r
ISSN:
1693-6930
Estimated position of
the
proposed method
Estimated position of
Trilateration
(a)
rLOS
beacon
(b)
rLOS
circle
B
1
B
2
B
3
P
T
P
T
B
1
B
2
B
3
.
.
.
.
Figure
5.
Position
estimation
in
Case
3
3.2.
Multi
BLE
iBeacon
lateration
F
or
achie
ving
high
accurac
y
,
we
use
information
from
four
beacons
simultaneously
.
Accordingly
,
when
applying
T
rilateration
for
four
beacons,
we
ha
v
e
C
(4
;
3)
w
ays
to
choose
a
cluster
of
three
beacons
from
a
set
of
four
ones
where
commutation
is
not
allo
wed.
In
detail,
at
first,
we
collect
signals
from
all
beacons
on
the
map.
Then
arrange
them
in
an
array
of
four
components:
B
=
[
B
1
;
B
2
;
B
3
;
B
4
]
(13)
in
which,
B
i
represents
i
th
beacon
object
in
descending
order
of
RSSI.
In
programming,
B
i
is
a
tuple,
which
is
represented
in
the
form:
B
i
=
[(
x
i
;
y
i
);
i
;
d
i
]
,
where,
(
x
i
;
y
i
)
is
the
coordinates
in
tw
o
dimensions
space,
i
and
d
i
are
RSSI
and
distance
from
the
smartphone
to
that
beacon,
respecti
v
ely
.
Let
T
(
)
be
the
con
v
entional
T
rilateration
function
and
P
(
)
be
the
function
used
to
estimate
the
position
according
to
the
proposed
method.
If
we
ha
v
e
tw
o
rLOS
beacons,
four
possible
positions
are
calculated
by:
8
>
>
>
<
>
>
>
:
^
P
1
=
P
(
B
1
;
B
2
;
B
3
)
^
P
2
=
P
(
B
1
;
B
2
;
B
4
)
^
P
3
=
P
(
B
1
;
B
3
;
B
4
)
^
P
4
=
P
(
B
2
;
B
3
;
B
4
)
(14)
In
case
we
onl
y
ha
v
e
a
rLOS
beacon
(
B
1
is
the
rLOS
beacon
object),
four
possible
positions
are
calculated
by:
8
>
>
>
<
>
>
>
:
^
P
1
=
P
(
B
1
;
B
2
;
B
3
)
^
P
2
=
P
(
B
1
;
B
2
;
B
4
)
^
P
3
=
P
(
B
1
;
B
3
;
B
4
)
^
P
4
=
T
(
B
2
;
B
3
;
B
4
)
(15)
In
the
absence
of
an
y
rLOS
beacon,
the
entire
P
(
)
function
in
(15)
is
replaced
by
the
T
(
)
function.
The
final
position
is
estimated
by:
P
=
(
^
P
1
+
^
P
2
+
^
P
3
+
^
P
4
4
for
Case
1,
Case
3
3
^
P
1
+3
^
P
2
+3
^
P
3
+
^
P
4
10
for
Case
2
(16)
Herein,
the
P
(
)
function
i
s
considered
to
be
more
reliable
than
the
T
(
)
function.
Hence,
the
position
that
returned
by
the
P
(
)
function
should
be
assigned
with
a
higher
weight.
It
is
also
possible
if
we
u
s
e
T
rilateration
for
a
lar
ger
number
of
beacons.
F
or
e
xample,
with
fi
v
e
beacons,
we
ha
v
e
to
calculate
C
(5
;
3)
=
10
operations
instead
of
C
(4
;
3)
=
4
.
Ho
we
v
er
,
this
increases
the
computational
cost.
Besides,
it
requires
a
bigger
number
of
beacons,
which
leads
to
an
increase
in
deplo
yment
costs.
Therefore,
choosing
to
use
the
four
beacons
is
reasonable.
4.
EV
ALU
A
TION
4.1.
Experiment
setup
The
e
xperiment
w
as
e
x
ecuted
on
the
1st
floor
of
G2
b
uilding,
Uni
v
ersity
of
Engineering
and
T
echnol-
ogy
,
VNU.
The
testbed
is
typically
open
with
some
decorati
v
e
trees,
tw
os
big
columns,
and
occasionally
has
the
TELK
OMNIKA
T
elecommun
Comput
El
Control,
V
ol.
19,
No.
1,
February
2021
:
51
–
62
Evaluation Warning : The document was created with Spire.PDF for Python.
TELK
OMNIKA
T
elecommun
Comput
El
Control
r
57
appearance
of
pedestrians.
W
e
deplo
y
four
beacons
in
an
area
of
90
m
2
,
and
the
shortest
distance
between
tw
o
beacons
is
6
meters.
The
layout
of
the
floor
plan
and
the
interested
area
are
sho
wn
in
Figure
6.
Beacons
are
set
according
to
the
strate
gy
outlined
in
[24]
at
the
height
of
150
cm
(equi
v
alent
to
the
height
of
user
equipment)
with
the
same
technical
configuration.
Detailed
s
p
e
cifications
of
the
system
are
gi
v
en
in
T
able
1.
The
phone
is
held
close
to
the
body
and
in
a
parallel
position
to
the
horizontal
plane.
The
proposed
method
is
designed
for
discrete
positioning
and
applied
to
e
x
ecute
multiple
measurements.
W
e
wrote
a
measurement
softw
are
with
the
help
of
se
v
eral
a
v
ailable
frame
w
orks,
i.e:
Cor
eLocation
[25],
Cor
eGr
aphic
[26].
The
collected
data
is
sent
via
email
and
we
then
use
it
to
plot
the
figures
using
MA
TLAB.
T
able
1.
System
parameters
De
vice
iPhone
SE
Operation
System
iOS
12.2
Beacon
4
Proximity
Estimote
Beacon
Bluetooh
Interf
ace
BLE
v5.0/2.4
GHz
Adv
ertising
Interv
al
100
ms
Broadcasting
Po
wer
0
dBm
Broadcasting
Range
50
m
P. 1
0
3
NHÀ G
2
B
P. 1
0
5
8.1
m
P. 1
0
7
P. 1
0
2
P
H
ÒN
G T
H
Í
N
GH
I
Ệ
M
NG
UY
Ễ
N V
Ă
N
ĐẠ
O
45
m
2
6.0
m
P
H
ÒN
G H
Ọ
C NV
CL
90
m
2
12
m
7.5
m
P
H
ÒN
G
T
N
18
m
2
P
H
ÒN
G
MU
L
T
I
MED
I
A
T
RUN
G
T
ÂM
MÁ
Y T
Í
N
H
PH
Ò
N
G
MÁ
Y
CH
Ủ
P
H
ÒN
G
CH
Ờ
GI
Ả
NG
18
m
2
P. 1
0
4
P
H
ÒN
G
K
H
O
10
m
2
TR
ƯỜ
NG
ĐẠ
I H
Ọ
C
KI
NH
T
Ế
Up
Up
P. 1
0
6
P
H
ÒN
G H
Ọ
C NV
CL
78
m
2
Up
K
K
K
K
KI
NH
KI
NH
KI
NH
IN
H
T
Ế
T
Ế
T
Ế
T
Ế
TR
Ư
TR
Ư
TR
Ư
TR
ƯỜ
ƯỜ
ƯỜ
ƯỜ
N
N
N
N
NG
Đ
NG
Đ
NG
Đ
NG
Đ
Đ
Đ
Đ
Đ
Ạ
I
Ạ
I
Ạ
I
Ạ
I
Ế
Ế
Ế
H
H
H
H
Ọ
C
Ọ
C
Ọ
C
Ọ
C
C
C
C
C
WC
Na
m
WC
N
ữ
Up
O
x
y
N
E
S
W
T
e
s
t
p
o
in
t
(1
4
;
9.
6
)
St
ar
t
poi
n
t
(6;8)
(8
;
1
1
.6)
(1
4
;
1
1
.6)
(1
1
;
6)
(1
5
.
6;
6)
Figure
6.
Layout
of
the
e
xperiment
area
4.2.
Experiment
r
esults
4.2.1.
Distance
estimation
Firstly
,
we
w
ant
to
e
v
aluate
the
accurac
y
in
estimating
the
distance
from
the
beacon
to
the
user
de
vice
.
As
mentioned
abo
v
e,
the
distance
estimation
strate
gy
is
applied
to
tw
o
dif
ferent
RSSI
ranges.
If
RSSI
is
higher
than
-70
dBm,
the
(9)
will
be
used.
In
contrast,
if
RSSI
is
less
than
or
equal
to
-70
dBm,
the
distance
is
calculated
by
the
a
v
erage
between
LOS
and
NLOS.
W
e
measure
RSS
v
alue
at
dif
ferent
distances
and
directions
that
change
from
1
to
10
m.
Each
position
is
1
m
apart,
and
we
collect
100
RSSI
samples
for
each
one.
The
result,
including
the
a
v
erage
estimated
distance
and
v
ariance,
are
gi
v
en
in
T
able
2.
As
the
results
are
sho
wn
in
T
able
2,
at
distances
of
1
m,
2
m,
and
3
m,
the
estimated
dis
tances
are
entirely
accurate.
The
cause
of
this
result
is,
at
those
distances,
the
RSS
le
v
el
i
s
almost
higher
than
-70
dBm,
then
the
distance
is
estimated
by
(9).
Since
the
equation
for
(9)
changes
slo
wly
in
this
RSSI
range,
the
v
ariance
is
not
too
much.
F
or
distances
greater
than
3
m,
the
RSS
v
aries
considerably
between
LOS
and
NLOS
and
is
usually
less
than
-70
dBm.
Thus,
the
distance
is
estimated
by
the
a
v
erage
model.
Consequently
,
the
errors,
as
well
as
the
v
ariance
in
these
cases,
are
high.
Smartphone
indoor
positioning
based
on
enhanced
BLE
beacon...
(Ngoc-Son
Duong)
Evaluation Warning : The document was created with Spire.PDF for Python.
58
r
ISSN:
1693-6930
T
able
2.
Distance
estimation
T
rue
Distance
(m)
1
2
3
4
5
6
7
8
9
10
A
vg.
Estimated
Distance
(m)
1.08
1.91
3.12
3.76
5.56
7.25
8.3
8.88
10.82
12.23
V
ariance
0.12
0.21
0.3
0.74
1.23
1.73
1.96
2.08
2.25
4.23
4.2.2.
Impact
of
Kalman
filter
In
Figure
7,
the
red
line
represents
nature
RSS
at
a
distance
of
2.5
m
under
LOS
condition,
and
the
blue
line
re
presents
filtered
RSS
using
the
Kalman
filter
.
At
some
time
steps,
the
red
line
sharply
attenuated
due
to
the
presence
of
pedestrians
.
W
e
easily
see
the
Kalman
filt
er
some
what
reduced
the
impact
of
the
RSS
fluctuations
in
the
blue
line.
Figure
8
depicts
the
estimated
position
in
the
case
abo
v
e.
When
not
using
the
Kalman
filter
,
the
estimated
positions
sho
w
a
lar
ge
dispersion.
After
filtering,
estimated
positions
sho
w
an
opposite
trend.
The
cause
of
this
result
is
the
determination
of
whether
a
beacon
is
under
rLOS
condition
or
not.
In
sensiti
v
e
cases
(RSS
70
dBm),
Kalman
filter
helps
to
reduce
about
37%
of
the
v
ariance.
Figure
7.
Comparison
of
ra
w
RSS
and
KF-filtered
RSS
10
12
14
16
9
9.5
10
10.5
11
Nature Position
True Position
10
12
14
16
9
9.5
10
10.5
11
Filtered Position
True Position
Figure
8.
The
estimated
position
is
thanks
to
using
Kalman
filter
4.2.3.
Ov
erall
perf
ormance
The
proposed
method
is
v
erified
via
the
e
xperiment
to
see
ho
w
ef
fecti
v
e
it
is.
Distance
errors
are
used
to
e
v
aluate
the
accurac
y
of
the
system.
W
e
define
location
error
e
as
the
distance
between
the
estimated
position
(
x
est
;
y
est
)
and
the
actual
position
(
x;
y
)
,
i.e:
e
=
p
(
x
x
est
)
2
+
(
y
y
est
)
2
(17)
TELK
OMNIKA
T
elecommun
Comput
El
Control,
V
ol.
19,
No.
1,
February
2021
:
51
–
62
Evaluation Warning : The document was created with Spire.PDF for Python.
TELK
OMNIKA
T
elecommun
Comput
El
Control
r
59
W
e
c
h
oos
e
three
other
methods
to
compare
with
proposed
method,
i.e:
con
v
entional
trilateration,
least
square
estimation
and
weighted
least
squares
estimation
[27].
T
able
3
specifies
the
parameters
used
in
the
e
xperiments.
T
able
3.
Experiment
parameter
Apply
for
...
beacons
Distance
estimati
on
Kalman
filter
T
rilateration
3
A
vg.
Model
Y
es
LS
4
A
vg.
Model
Y
es
WLS
4
A
vg.
Model
Y
es
Proposed
Method
4
LOS
and
A
vg.
Model
Y
es
a.
Accurac
y
of
the
proposed
method
First
of
all,
we
w
ant
to
in
v
estig
ate
the
a
v
erage
error
for
each
site
in
the
re
gion
of
interest.
The
e
x-
periment
w
as
carried
out
on
the
zigzag
route
(as
sho
wn
in
Figure
6)
at
80
dif
ferent
points
on
the
map.
Error
distrib
ution
is
sho
wn
in
Figure
9.
W
e
can
recognize
that
errors
of
positions
that
lie
in
the
parallelogram
of
four
beacons
are
lo
wer
than
others
that
li
e
outside
this
re
gion.
The
a
v
erage
error
in
thi
s
area
is
about
1.7
m.
Especially
,
the
positions
with
coordinates
(8.4;
11.2),
(14;
11.2),
(11;
8)
and
(15.6,
18)
ha
v
e
ne
gligible
errors.
The
reason
for
this
result
is
that
these
positions
are
located
in
front
of
beacons
where
the
distance
is
v
ery
close.
Consequently
,
when
the
user
stands
at
these
points,
the
estimated
position
will
be
treated
with
another
polic
y
,
as
desc
ribed
in
section
3.1.
In
contrast,
edge
areas
ha
v
e
poor
accurac
y
because
phones
and
iBeacon
are
often
under
an
unreliable
situation.
At
both
ends
of
the
e
xperiment
area,
the
error
may
reach
more
than
4
m.
18
17.2
16.4
15.6
14.8
14
13.2
0
1
X
12.4
2
11.6
3
Avg. Localization Error (m)
8
4
10.8
5
8.8
10
Y
9.2
9.6
8.4
10.4
7.6
11.2
6.8
6
0
0.5
1
1.5
2
2.5
3
3.5
4
Figure
9.
Error
distrib
ution
of
proposed
method
on
the
testbed
b
.
Positioning
accurac
y
in
dif
ferent
directions
Figure
10
describes
the
accurac
y
of
the
algorithms
in
four
directions
at
a
fix
ed
position,
say
,
(14;
9.6).
When
the
heading
of
the
user’
s
de
vice
is
the
North,
since
none
of
the
beacon
w
as
disco
v
ered
is
under
rLOS
condition,
the
accurac
y
of
the
proposed
method
is
not
much
better
than
the
other
algorithms.
When
users
point
their
phone
to
w
ards
the
East,
South,
or
W
est,
a
beacon
at
the
coordinate
of
(14;
11.6)
is
detected
as
a
reliable
beacon.
As
a
result,
our
proposed
method
sho
ws
a
mark
ed
impro
v
ement
when
compared
to
other
methods.
c.
Ov
erall
accurac
y
comparison
Figure
11
depicts
the
accumulati
v
e
error
of
four
approaches.
As
we
can
see,
our
proposed
method
has
the
best
performance
in
total
of
four
.
The
a
v
erage
error
of
the
method
is
1.9
m.
Our
proposed
method
helps
increase
the
accurac
y
of
indoor
positioning
by
35.15%
o
v
er
the
con
v
entional
trilateration
method,
23.22%
o
v
er
the
least
square
method,
and
15.55%
o
v
er
weighted
least
square
method.
Smartphone
indoor
positioning
based
on
enhanced
BLE
beacon...
(Ngoc-Son
Duong)
Evaluation Warning : The document was created with Spire.PDF for Python.
60
r
ISSN:
1693-6930
10
11
12
13
14
15
16
17
18
7
8
9
10
11
12
North
Proposed Method
Trilateration position
LS position
WLS position
True Position
10
11
12
13
14
15
16
17
18
7
8
9
10
11
12
East
10
11
12
13
14
15
16
17
18
7
8
9
10
11
12
South
10
11
12
13
14
15
16
17
18
7
8
9
10
11
12
West
Figure
10.
Actual
and
estimated
position
of
4
methods
at
4-orientations
0
1
2
3
4
5
6
7
Localization Error (m)
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
CDF
Conventional Line Intersection-Based Trilateration
Least Square Estimation
Proposed Method
Weighted Least Square Estimation
Figure
11.
Comparison
of
distance
error
distrib
ution
for
dif
ferent
methods
5.
CONCLUSION
In
this
paper
,
we
introduced
a
plain
and
rob
ust
method
to
impro
v
e
the
accurac
y
of
the
tri
lateration-
based
indoor
positioning
system
using
BLE
beacon.
The
proposed
method
mak
es
use
of
an
RSSI
range
(greater
than
-70
dBm
equi
v
alent
to
a
distance
of
less
than
3
m)
to
estimate
the
distance
accurately
.
This
increases
the
positioning
accurac
y
by
mo
ving
the
estimated
position
of
trilateration
to
reliable
circles.
In
addition,
the
po
wer
of
four
beacons
is
utilized
at
the
same
time
for
more
accurate
positioning.
Experimental
results
sho
w
that
this
is
an
ef
fecti
v
e
and
rob
ust
proposed
scheme.
As
we
ha
v
e
seen
the
impact
of
rLOS
beacon,
in
the
future,
we
will
study
the
optimal
method
of
beacon
placement,
in
which,
in
an
y
position,
users
al
w
ays
be
able
to
observ
e
an
rLOS
beacon.
A
CKNO
WLEDGMENT
This
w
ork
has
been
supported/partly
supported
by
V
ietnam
National
Uni
v
ersity
,
Hanoi
(VNU),
under
Project
No.
QG.19.25
REFERENCES
[1]
H.
Liu,
J
.
Y
ang,
S.
Sidhom,
Y
.
W
ang,
Y
.
Chen
and
F
.
Y
e,
“
Accurate
W
iFi
Based
Localization
for
Smartphones
Using
Peer
Assistance,
”
IEEE
T
r
ansactions
on
Mobile
Computing
,
v
ol.
13,
no.
10,
pp.
2199-2214,
2014.
TELK
OMNIKA
T
elecommun
Comput
El
Control,
V
ol.
19,
No.
1,
February
2021
:
51
–
62
Evaluation Warning : The document was created with Spire.PDF for Python.