Indonesian J
ournal of Ele
c
trical Engin
eering and
Computer Sci
e
nce
Vol. 2, No. 2,
May 2016, pp
. 426 ~ 430
DOI: 10.115
9
1
/ijeecs.v2.i2.pp42
6-4
3
0
426
Re
cei
v
ed
Jan
uary 16, 201
5
;
Revi
sed Ap
ril 12, 2016; Accepted Ap
ril 26, 2016
Weed Detection Using Fractal-Based Low Cost
Commodity Hardware Raspberry Pi
Mohamad Iq
bal Surians
y
ah*
1
, Heru S
u
koco
2
, Moh
a
mad Solah
udin
3
1
Departme
n
t of Computer Sci
ence F
a
cu
lt
y
of
Natu
ral Sci
enc
e and Math
em
atics, Pakuan
Univers
i
t
y
,
Indon
esi
a
,
2
Departme
n
t of Computer Sci
ence F
a
cu
lt
y
of
Natu
ral Sci
enc
e and Math
em
atics, Bogor Ag
ricultura
l
Univers
i
t
y
, Ind
ones
ia,
3
Department o
f
Mechanic
a
l a
nd Bios
ystem
Engi
neer
in
g, F
a
cult
y of Agric
u
ltura
l
Engi
ne
e
r
ing a
nd
T
e
chnolog
y, B
ogor Agr
i
cultur
al Univ
ersit
y
, Indo
nesi
a
,
addr
es, telp/fa
x of instituti
on/
affiliati
on
*Corres
p
o
ndi
n
g
author, e-ma
i
l
: mohama
d
.iq
bal@
u
n
pak.ac.
id
1
, hsrkom@ip
b
.ac.id
2
, msoul
9@
ya
hoo.com
3,
A
b
st
r
a
ct
Conv
entio
na
l
w
eed contro
l s
ystem is
usu
a
l
l
y use
d
by spr
a
yin
g
her
bici
d
e
s un
ifor
mly th
roug
hout
the lan
d
. Excessive use of h
e
rbici
des o
n
a
n
ong
oin
g
basi
s
can produc
e che
m
ic
al w
a
ste that is harmf
ul to
pla
n
ts and s
o
il.
T
he ap
plic
atio
n of prec
isio
n
agric
ulture
far
m
i
ng i
n
the
det
ection
process
in ord
e
r to co
ntro
l
w
eeds us
ing
Co
mp
uter Vis
i
on On
Farm
beco
m
es i
n
ter
e
sting,
but it
still h
a
s s
o
me
pro
b
le
ms
du
e to
computer
si
z
e
and
pow
er
con
s
umptio
n. Ras
pberry
Pi
is o
n
e
of th
e
mi
nic
o
mputer
w
i
th l
o
w
price
an
d
l
o
w
pow
er co
nsu
m
ption. H
a
vi
ng
comput
i
ng
like
a d
e
sktop c
o
mp
uter w
i
th th
e op
en s
ourc
e
Lin
u
x o
per
ati
n
g
system c
an
be
used
for i
m
ag
e proc
essi
ng
a
nd w
e
e
d
fracta
l di
mensi
on
pr
ocessi
ng
usin
g
OpenCV
li
brar
y
and
C pr
ogr
a
m
mi
ng. T
h
is
re
search r
e
su
lts the b
e
st
fractal
co
mput
ation
ti
me
w
hen
perfo
rmi
ng th
e i
m
ag
e
w
i
th dimens
ion
si
z
e
of 1
28 x
128
pixe
ls. It is abo
ut
7
mil
l
i
s
econ
ds. F
u
rthermore, the
av
erag
e spe
ed r
a
ti
o
between personal comput
er
and Raspberry
Pi is 0.04 times faster
. The use of Ras
p
berry Pi is cost and
pow
er consu
m
ption effici
ent c
o
mpar
ed to per
sona
l co
mput
e
r
.
Ke
y
w
ords
:
W
eeds D
e
tectio
n
,
Comp
uter Vis
i
on, F
r
actal, Ra
spberry Pi
Copy
right
©
2016 In
stitu
t
e o
f
Ad
van
ced
En
g
i
n
eerin
g and
Scien
ce. All
rig
h
t
s reser
ve
d
.
1. Introduc
tion
Farm ma
nag
ement syste
m
s ba
sed o
n
informat
ion t
e
ch
nolo
g
y has bee
n wid
e
ly used to
obtain o
p
timum be
nefit, increa
se the
efficien
cy of agri
c
ultu
ral
manag
eme
n
t [1], protectin
g
the
environ
ment
[2] and incre
a
sin
g
ag
ricult
ural p
r
od
ucti
vity [3]. Farmers
nee
d info
rmation from
a
wide
ran
ge o
f
ICT tools to
identify, analyze, and ma
nage info
rma
t
ion in sp
atial
and tempo
r
al
diversity [4] a
s
well as th
e
spe
c
ific
cha
r
acteri
st
ics of
the land [5], so that the d
e
ci
sion
-ma
k
in
g
process to
be more preci
s
e during soil preparat
ion, seed
selection,
fertilizer regulation,
manag
eme
n
t pesti
cide
s, watering
sched
ules
water a
n
d
wee
d
mana
gement [6].
The p
r
o
c
e
ss
of identificati
on of weed
s in t
he field i
s
very imp
o
rt
ant to dete
r
mine the
effective cont
rol of this du
e
to lack of p
r
o
per
weed
co
n
t
rol will cau
s
e
improp
er u
s
e of herbi
cid
e
s
,
inefficien
cie
s
co
st, time and energy [7]. Convent
io
na
l weed
control system i
s
usu
a
lly done
by
spraying
herbicid
e
s
uniformly throug
ho
ut the lan
d
[8
], it results in
exce
ssive
u
s
e of
herbici
des
will pote
n
tially generate waste in th
e fo
rm of c
hemi
c
al re
sidu
es,
emission
s to
air an
d soil [9].
Dep
end
ence
on ch
emicals
also h
a
rm h
u
m
an
health [1
0] and the en
vironme
n
t [11].
The he
rbici
d
es can be re
duced by the applicat
io
n of preci
s
io
n farmin
g appli
c
ation by
spraying on right land by detectin
g
we
eds on la
nd.
Therefore, preci
s
ion farmi
ng is nee
ded
to
determi
ne th
e level of
we
ed vegetatio
n
in orde
r to
control th
e co
ndition
s an
d
need
s of th
e
plant
based on the
spe
c
ific ch
a
r
acte
ri
stics of the land
[12]. Precisio
n farmin
g is the
applicatio
n of
informatio
n tech
nolo
g
y in agri
c
ultural
manag
em
e
n
t system
s that allow
rig
o
rou
s
treatm
ent
(preci
se tre
a
tment) ag
ribu
siness chain from upst
r
eam
(on farm) to downs
t
ream (off farm) [13].
Comp
uter vision a
s
one
o
f
the p
r
e
c
isi
o
n farm
i
n
g
ap
plicatio
ns is
very promi
s
i
ng [14]
whi
c
h
ca
n be
used fo
r the
identificatio
n
and
cla
s
sifi
ca
tion of pl
ants.
Ope
n
CV i
s
a library Pu
bl
ic
License can
be used to detect the
im
age of
weeds.
Weed detecti
on in
realtime is
still difficul
t
to
impleme
n
t in the field due
to need a l
a
rge pla
c
e a
n
d
the use
of la
rge el
ectri
c
p
o
we
r. The n
e
ed
for sp
ecifi
c
ation of minico
mputer a
nd small pow
er consumption has
attr
acte
d
the attention of
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 2, May 2016 : 426 –
430
427
some
re
sea
r
che
r
s to crea
te a single b
oard
comp
ut
er and a
cre
d
it card
size
d using the
open
sou
r
ce Li
nux
operating
system whi
c
h
is
calle
d t
he
Ra
spb
e
rry Pi [1
5]. The
Ra
sp
berry Fou
nda
tion
laun
che
d
the
latest
Ra
sp
b
e
rry Pi
produ
ct in
t
he fo
rm
of Singl
e Bo
ard
Compute
r
, a
sm
all-sized
comp
uter wit
h
low po
we
r
consumpt
ion,
3.5 W
(5 V a
n
d
0.75 A) [16]
. The o
r
iginal
Ra
spb
e
rry Pi is
based on A
R
M1176
JZF
-
S
700 M
H
z p
r
o
c
e
s
sor, Video
Core IV GPU, and wa
s o
r
i
g
inally shi
ppe
d
with 25
6 me
gabytes
of RAM, later
upgrade
d
(m
odel
s B and
B+) to
512
MB [17]. Th
e
developm
ent
of a mini compute
r
Ra
spberry Pi has ope
n
ed u
p
great opp
ortunities in th
e
comp
uting
system to be
a
pplied in
a n
u
mbe
r
of
re
sea
r
ch a
r
ea
s [18] and
ca
n be o
ne of
the
solutions to be implemented
easily as functionally has th
e ability like a desktop
computer.
2. Rese
arch
Metho
d
2.1. Image Acquisition
At this sta
ge,
we
ed
s ima
g
e
is a
c
qui
red
usi
ng a
digit
a
l came
ra. T
he d
a
ta u
s
ed
in thi
s
study i
s
a
collectio
n of i
m
age
s o
b
tai
ned
we
ed
pl
ants f
r
om th
e lab
o
rato
ry
of Me
ch
ani
c
al
Enginee
ring
and Biosy
s
te
ms IPB, Facu
lty
of Agricultural Te
ch
nolo
gy IPB.
In this study, the image of wee
d
s
whi
c
h
is
used is a
wide vari
ety of dimensi
o
n
a
l image
as can be
see
n
in Tab
l
e 1. The maximum si
ze
of 0.3 MB image is a
s
sumed a
s
the
con
s
cientio
us size imag
e t
o
perfo
rm filt
ering
proc
es
s
.
Test
s
ca
rri
e
d
out
mainly
on lan
d
t
h
at
has
not
bee
n sprayed
by
h
e
rb
icide befo
r
e planting pe
riod (P
re Eme
r
ge
), imag
e
data captu
r
e
d
i
s
planting p
e
rio
d
1-4 wee
ks t
hat is don
e b
e
ca
use in tha
t
span it is the right time a
s
a critical pe
riod
of weed
com
petition with the main cro
p
.
Table 1. Imag
e Dimen
s
io
n
Image
Dimension
(Pixel)
1
128 x 1
28
2
256 x 2
56
3
380 x 3
80
4
480 x 4
80
5
512 x 5
12
2.2. Weed
s Image Filtrati
on
The ima
ge
wa
s taken
and a
nalyze
d
to det
erm
i
ne the
colo
r of its con
s
tituent
comp
one
nts.
Base
d on
the colo
r co
mpone
nts
are
then determined param
et
ers filtratio
n
to
sep
a
rate th
e
backg
rou
nd i
m
age of the
staple
crop
s i
n
bina
ry (bla
ck
and
white
)
. Data array
o
f
pixels that sto
r
e bina
ry valu
es processe
d
image
usi
ng
fractal dim
e
n
s
ion a
nalysi
s
,
can be
see
n
in
figure 1. Assessment of the huma
n
eye is use
d
as
a ben
chma
rk to determine
the accura
cy
of
the perfo
rma
n
ce
system
which i
s
built.
Figure 1. Image filtering p
r
oce
s
s we
ed
s into binary da
ta
2.3. Fractal
Dimension Analy
s
is
Fra
c
tal di
me
nsio
n a
nalysi
s
i
s
p
e
rfo
r
me
d by
fragme
n
t
ation of the
i
m
age
that h
a
s
b
een
difilterisa
s
i int
o
a re
ctang
ul
ar shape m
e
asu
r
ing
s. Th
en cal
c
ul
ate the numb
e
r of
squa
re
s N
(s)
that contain
s
the white co
lor (re
s
ults fil
t
ration pl
a
n
t). This calculat
ion is repe
ated with different
values of s a
s
much as 1
0
intervals. Th
e next step is
to plot the va
lue of log N (s) to the valu
e of
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Wee
d
Dete
cti
on Using F
r
a
c
tal-Ba
se
d Lo
w Co
st Com
m
odity Hardware …
(M. Iqbal Suriansy
ah)
428
log (1 / s) a
nd dete
r
min
e
the sh
ape
of the li
nea
r reg
r
e
ssi
on
equatio
n y = ax + b. F
r
a
c
tal
dimen
s
ion i
s
a value on th
e linear
reg
r
e
ssi
on eq
uatio
n.
3. Results a
nd Analy
s
is
3.1. Image Filtering
Comp
utation
a
l image of p
r
ep
ro
ce
ssi
ng
filter is
obtain
ed by comp
a
r
ing an
d testi
ng as
N
= 10 trial
s
for
each image o
n
the Ra
spb
e
rry Pi and PC, can be seen
in Table 2 an
d Table 3.
Table 2. Imag
e filtering on
Ra
spb
e
rry Pi
Image Dimension (Pixel)
N Testing (Seco
nd)
1 2
3
4 5
6 7
8
9
10
128
x
1
2
8
0.1
0.1
0.13
0.11 0.11
0.11 0.12
0.12
0.1 0.11
256
x
2
5
6
0.48 0.49
0.45
0.45 0.43
0.44 0.48
0.45
0.47 0.44
380
x
3
8
0
0.89 1.09
1.04
0.93 1.18
1.12 1.04
0.98
0.92 0.89
480
x
4
8
0
1.47 1.79
1.53
1.63 1.78
1.52 1.89
1.56
1.41 1.47
512
x
5
1
2
1.72 1.98
1.8
2.21 1.65
1.8 1.69
1.81
1.78
1.8
Table 3. Imag
e filtering on
PC
Image Dimension (Pixel)
N Testing (Seco
nd)
1 2
3
4 5
6 7
8
9
10
128
x
1
2
8
0.01 0.02
0.01
0.01 0.01
0.01 0.01
0.01
0.02 0.01
256
x
2
5
6
0.02 0.02
0.02
0.02 0.02
0.01 0.02
0.03
0.02 0.02
380
x
3
8
0
0.04 0.04
0.04
0.05 0.04
0.04 0.04
0.05
0.04 0.04
480
x
4
8
0
0.06 0.06
0.05
0.06 0.05
0.05 0.05
0.05
0.06 0.05
512
x
5
1
2
0.07 0.07
0.06
0.06 0.06
0.07 0.06
0.06
0.06 0.06
From th
e
compa
r
ison o
f
the prep
ro
ce
ssi
ng com
putation of
i
m
age proce
s
s
with
Ra
spb
e
rry Pi
and the PC can be ge
nera
t
ed averag
e value which ca
n be se
en in
Table 4.
Table 4. The
averag
e time of image filtering on Raspb
e
rry Pi and P
C
Image Dimension (Pixel)
Computation (Se
c
ond)
Raspberry PC
128 x 1
2
8
0.111
0.012
256 x 2
5
6
0.458
0.020
380 x 3
8
0
1.008
0.042
480 x 4
8
0
1.605
0.054
512 x 5
1
2
1.824
0.063
Figure 2. Co
mpari
s
o
n
of image filterin
g
on Ra
spb
e
rry Pi and PC
Evaluation Warning : The document was created with Spire.PDF for Python.
ISSN: 25
02-4
752
IJEECS
Vol.
2, No. 2, May 2016 : 426 –
430
429
The
re
sults
of the ab
ove expe
rimen
t
s u
s
in
g
R
a
s
p
be
rr
y a
n
d PC
s
h
ow
th
a
t
the
comp
utationa
l time of image prepr
ocessing filter i
s
worth
polyno
m
ial that is close to 1
whi
c
h
mean
s that the com
puting
time propo
rtional to t
he si
ze of the ima
ge, the large
r
image is u
s
ed
more a
nd m
o
re
comp
utin
g time requi
red. In gen
e
r
al, PC co
m
p
utation time
is faster th
an
Ra
spb
e
r
r
y
Pi.
3.2. Fractal
Dimension Analy
s
is
In the process of comp
utin
g the fra
c
tal d
i
mens
i
on a
n
a
l
ysis
ca
rrie
d
o
u
t by the processing
results of
ima
ge
filterin
g weed
s su
ch
as bina
ry data
i
n
ra
sp
be
rry p
i
, can
be
see
n
in T
able
5
by
usin
g fractal
algorith
m
s u
s
ing C.
Table 5. Co
m
putational results usi
ng fra
c
tal on Raspb
e
rry Pi and P
C
Image Dimension
(Pi
x
el
)
Raspberry PC
Computation
(ms
)
Fractal
Computation
(ms
)
Fractal
128 x 1
2
8
7
0,9
0.1
0,9
256 x 2
5
6
12
1,1
0.5
1,1
380 x 3
8
0
37
1,3
2.1
1,3
480 x 4
8
0
74
1,7
4.2
1,7
512 x 5
1
2
93
1,8
5
1,8
The re
se
arch
result
s in T
able 5 shows be
st
fractal
computatio
n
time in Raspbery Pi
whe
n
perfo
rm
ing the image
with dimen
s
i
on si
ze of 12
8 x 128 pixels. It is about 7 millise
c
on
ds.
Figure 3. Co
mpari
s
o
n
of computation ti
me fractal im
age on
Ra
sp
berry Pi and PC
The re
sult
s of the above experime
n
ts show th
at the fractal comp
u
t
ational time on a PC
is fast
er tha
n
the
Ra
spb
e
rry Pi. Fu
rt
herm
o
re, th
e
avera
g
e
sp
eed
ratio b
e
t
ween
pe
rso
nal
comp
uter
and
Ra
spb
e
rry Pi is 0.0
4
time
s faster.
T
he
compa
r
ison of
the sp
ecifi
c
ati
o
ns i
n
Ta
ble
8
is 4 : 1 and the ratio of the power con
s
u
m
ption of
34 : 1 is more effi
cient to use Rasp
berry Pi.
Table 6. Gen
e
ral comp
ari
s
on between
Ra
spb
e
rry Pi
and PC
Specification
Raspberry
Pi
PC
Comparison
Processor speed
700 MHz x 1 co
re
1,5 GHz
x 2 core
4 : 1
RAM Size
Watts
512 MB
3.5
2 GB
65
4 : 1
16 : 1
Price
Rp 500.000
Rp 4.000.00
0
4 : 1
The averag
e rati
o
4 : 1
Evaluation Warning : The document was created with Spire.PDF for Python.
IJEECS
ISSN:
2502-4
752
Wee
d
Dete
cti
on Using F
r
a
c
tal-Ba
se
d Lo
w Co
st Com
m
odity Hardware …
(M. Iqbal Suriansy
ah)
430
4. Conclusio
n
The
avera
g
e
ratio
bet
wee
n
the
sp
eed
of t
he P
C
Rasp
berry Pi
and f
r
actal
p
r
oce
s
s i
s
0,04 time
s fa
ster. T
he b
e
st filtering com
putation
that
can
be
done
by Ra
spb
e
rry Pi is 5
12 x 5
12
Pixels. The u
s
e of Ra
sp
be
rry Pi is co
st and po
we
r co
nsum
ption eff
i
cient compa
r
ed to perso
n
a
l
comp
uter
.
Referen
ces
[1]
Blackmor
e S. Precisio
n F
a
rm
ing: an ov
ervi
e
w
.
Agric
u
ltur
al Engi
neer
. 1
994
; 49(3): 86-88.
[2]
Kuhar JE. T
he
Precisio
n-F
a
rm
ing: Guid
e for Agricult
urist.
Illi
nois: Joh
n
De
e
r
Publis
hin
g
.
1997.
[3]
Shibusa
w
a S,
Anom IM, Sas
ao A, S
a
kai K, Hach
e C.
Sa
mplin
g strate
gi
esin s
o
i
l
ma
ppi
ng
w
i
t
h
r
eal
-
time soi
l
sp
ectroph
otometer.
Di d
a
lam: Inte
ll
ige
n
t Co
ntrol f
o
r Agric
u
ltur
al
Appl
icatio
n.
Pr
ocee
din
g
o
f
2nd IF
AC-CIGR
W
o
rkshop o
n
, Bali Indo
nes
ia 22-
24 Au
gus
t 2001. Bali: IF
AC-CIGR. 200
1: 40-43.
[4]
Sola
hud
in
M. Peng
emb
ang
a
n
meto
de
pe
n
gen
dal
ia
n g
u
l
m
a p
ada
p
e
rtani
an
presis
i
berb
a
sis m
u
lt
i
age
n komp
uta
s
ion
a
l. Diserta
si In
stitut Pertania
n
Bog
o
r. 20
13.
[5]
Mc Bratne
y,
AB, Prongl
e,
MJ.
Spatial V
a
ria
b
il
ity in S
o
il I
m
pl
ic
ati
o
n
for Precisi
o
n
Agricu
lture
.
Procee
din
g
Precisio
n Agric
u
lt
ure, BIOS
Scientific Publ
iser L
t
d, Oxford. 199
7.
[6]
F
ahad S
h
a
hba
z Khan, Sa
ad
Razza
q, Kashi
f
Irf
an, F
ahad
Maqb
ool, Ahm
ad F
a
ri
d, Inam
Illahi
an
d
T
auqeer ul am
in. Dr.
W
heat: A W
eb-base
d
Expert System for
Diagn
osis
of Diseas
es a
nd Pests i
n
Pakistani Wheat
. Proceedi
ng
s of the W
o
rld Con
g
ress on E
ngi
neer
in
g W
C
E. 2008; 1.
[7]
Solahudin,
M
Seminar,
K Astika, W Buono
A.
W
eeds an
d
Plants R
e
cog
n
i
t
ion us
ing F
u
zzy Clusteri
n
g
and
F
r
actal D
i
m
e
n
si
on M
e
th
ods for A
u
to
matic W
eed
Co
n
t
rol
. Internati
o
n
a
l C
onfer
ence,
the Qu
alit
y
Information for
Comp
etitive Ag
ricultura
l
Ba
se
d Producti
on S
y
stem a
nd Co
mmerce. 201
0.
[8]
Kargar BAH, Shirzadifar AM.
Autom
a
tic Weed Detection System
an
d Sma
r
t Herbicid
e Sp
rayer Rob
o
t
for Corn Fields.
Proceeding of the 2013 RSI/ISM Inte
rnational Conferenc
e on
Robotics and
Mechatro
nics, T
ehran, Iran. 2013.
[9]
Hon
g
, Sung M
i
nza
n
, Li Z
han
g, Qin.
Detecti
on Syste
m
of Smart Spr
a
yer
s
: Status, Chal
len
ges a
n
d
Perspectiv
e
s
. Int J Agric & Biol Eng. 20
12; 5
(
3).
[10]
Pimente
l
D, McLau
ghl
in
L, Z
epp A, L
a
kita
n B, Kraus T
,
Kleinm
an P,
et al.
En
vi
ronm
en
ta
l
an
d
econ
o
m
ic effec
t
s of reducing
pestici
de us
e
. Bio Scie
nce. 1
991; 41(
6): 402
-409.
[11]
Leac
h AW
, Mu
mford JD.
Pes
t
icide envir
on
me
ntal acco
un
ting:
A meth
od
for
assessi
ng the
extern
a
l
costs of indivi
d
ual p
e
sticid
e a
pplic
atio
ns.
En
vironme
n
tal Po
llutio
n
. 20
08; 1
51(1): 13
9-1
4
7
.
[12]
McBratne
y AB, Prongl
e MJ.
Spatia
l Varia
b
i
lit
y in Soil I
m
pl
ic
ation for Prec
is
ion Agr
i
cultur
e
.
Proceed
in
g
Precisio
n Agric
u
lture, BIOS Scient
ific Pu
blis
er Ltd, Oxford. 199
7.
[13]
Semin
a
r KB. Parad
i
gma P
e
nda
ya
g
una
an
T
e
knol
ogi Inf
o
r
m
asi u
n
tuk Pe
rtania
n
. Prosid
ing S
e
min
a
r
Nasio
n
a
l
Semi
nar Informatika
Pertania
n
Indo
nesi
a
. 201
1: 34-42.
[14]
Ste
w
a
r
d, B a
n
d
L T
i
an.
Re
al
-time machi
ne vision
w
eed d
e
tection.
ASA
E pap
er No.
9
830
33 (UIL
U-
ENG-98-7
006).
1998.
[15]
M Richa
dson,
S W
a
llace. Getting Start
ed
w
i
t
h
Rasp
berr
y
Pi
. US: O’Reill
y
.
201
2.
[16]
Rahm
an A, M
a
rdh
ani
R. Hi
gh Perform
a
n
c
e
Com
putin
g
on C
l
uster
a
nd Mu
lticore
Architecture.
T
E
LKOMNIKA Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
n
g
.
2015; 1
3
(4): 1
408-
141
3.
[17]
Albert Sa
gal
a, Deni
Lumb
a
n
t
oruan, Ep
eli
n
M
anuru
ng, Iroma Situmor
a
ng, Adi Gu
na
w
a
n. Sec
u
r
e
d
Commun
i
cati
o
n
amo
n
g
HMI
and
Co
ntroll
er
usin
g RC-
4 A
l
gorithm
an
d R
a
spb
e
rr
y Pi.
T
E
LKOMNIKA
Indon
esi
an Jou
r
nal of Electric
al Eng
i
ne
eri
n
g
.
2015; 1
5
(3): 5
26-5
33.
[18]
Ali, Murat et al
.
T
e
chnical
De
velo
p
m
ent a
n
d
Socioec
on
o
m
i
c
im
plic
ations
of the Rasp
be
rry Pi as
a
Lear
nin
g
T
o
o
l
in
Deve
lo
pin
g
co
untries
.
5
t
h Proce
edi
ng
s of the IEE
E
Comp
uter
Scienc
e a
n
d
Electron
ic Engi
neer
ing C
o
mfe
r
ence (CEE
C). 201
3: 103-
108.
Evaluation Warning : The document was created with Spire.PDF for Python.