series b / 2012 number 1 volume 4 internetworking ... · series b of the internetworking indonesia...
TRANSCRIPT
InternetworkingIndonesiaJournal
Volum
e 4N
umber 1
Series B / 2012
ISSN: 1942-9703IIJ ©
2012w
ww
.InternetworkingIndonesia.org
Editors’ Introduction 1
Artificial Intelligence Elman Backpropagation Computing Models 3for Predicting Shelf Life of Processed CheeseSumit Goyal and Gyanendra Kumar Goyal
Analysis of Level and Barriers of E-Commerce Adoption by Indonesian 9Small, Medium and Micro Enterprises (SMMEs)Rajesri Govindaraju and Dissa R. Chandra
Android-Based License Plate Recognition using Pre-trained 15Neural NetworkRizqi K. Romadhon, Muhammad Ilham, Nofan I. Munawar, Sofyan Tan and Rinda Hedwig
Manually Interchangeable Heads of Homemade Computer 19Numerical Control (CNC) MachineJimmy Linggarjati & Rinda Hedwig
The International Journal of ICT and Internet Development
Series B: Computing, Communications, Engineering and Internet Technologies
Internetworking Indonesia JournalThe International Journal of ICT and Internet Development
ISSN: 1942-9703www.InternetworkingIndonesia.org
The Internetworking Indonesia Journal (IIJ) is a peer reviewed international journal devoted to the timely study of Information and Communication Technology (ICT) and Internet development. The journal seeks high-quality manuscripts on the challenges and opportunities presented by information technology and the Internet, and their impact to society. The IIJ is published electronically based on the Open Access Publication Policy, and the journal is indexed by SCOPUS and Google Scholar. Papers in the IIJ appear in one of two series:
Journal mailing address: Internetworking Indonesia Journal, PO Box 397110 MIT Station, Cambridge, MA 02139, USA.
Series B of the Internetworking Indonesia Journal covers the broad technical areas of computing, communications, engineering and Internet Technologies.
Possible topics for papers include, but are not limited to the following:
• Information technology and information systems• Communications technology• Software and hardware engineering• Algorithms and computation• Applications and services• Broadband and telecommunications technologies• Mobile and wireless networks
• Internet infrastructure systems, protocols, standards and technologies
• Multimedia and content development• Education and distant learning• Open source software development and deployment• Cloud Computing, SaaS, and Grid Computing
Series B: Computing, Communications, Engineering & Internet Technologies
• Series A: Community Informatics & Society • Series B: Computing, Communications, Engineering & Internet Technologies
Prof. Edy Tri Baskoro, PhD (ITB, Indonesia)Mark Baugher, MA (Cisco Systems, USA) Lakshminath Dondeti, PhD (Qualcomm, USA)Paul England, PhD (Microsoft Research, USA)Brian Haberman, PhD (Johns Hopkins University, USA)Prof. Svein Knapskog, PhD (NTNU, Norway)
Prof. Bambang Parmanto, PhD (University of Pittsburgh, USA) Prof. Wishnu Prasetya, PhD (Utrecht University, The Netherlands)Graeme Proudler, PhD (HP Laboratories, UK) Prof. Jennifer Seberry, PhD (University of Wollongong, Australia) Prof. Willy Susilo, PhD (University of Wollongong, Australia)Prof. David Taniar, PhD (Monash University, Australia)
Thomas Hardjono, PhD (MIT, USA)
Budi Rahardjo, PhD (ITB, Indonesia)
Kuncoro Wastuwibowo, MSc(PT. Telkom, Indonesia)
International Advisory Board
Series B:Co-Editors
Technical Review Board
Moch Arif Bijaksana, MSc (IT Telkom, Indonesia)Teddy Surya Gunawan, PhD (IIUM, Malaysia)Brian Haberman, PhD (Johns Hopkins University, USA)Dwi Handoko, PhD (BPPT, Indonesia)Martin Gilje Jaatun, PhD (SINTEF, Norway)Mira Kartiwi, PhD (IIUM, Malaysia)Maciej Machulak, PhD (Google, USA)Jesus Molina, PhD (Fujitsu Laboratories of America)
Bobby Nazief, PhD (UI, Indonesia) Anto Satriyo Nugroho, PhD (BPPT, Indonesia) Bernardi Pranggono, PhD (University of Leeds, UK)Bambang Prastowo, PhD (UGM, Indonesia)Bambang Riyanto, PhD (ITB, Indonesia)Andriyan Bayu Suksmono, PhD (ITB, Indonesia)Setiadi Yazid, PhD (UI, Indonesia)
Henri Uranus, PhD(UPH, Indonesia)
Vol. 4/No. 1 B (2012) INTERNETWORKING INDONESIA JOURNAL 1
ISSN: 1942-9703 / © 2012 IIJ
ELCOME to the Spring 2012 issue of the Internetworking Indonesia Journal. This issue sees a number of positive
developments in the Internetworking Indonesia Journal. One of the most significant developments in 2012 is the introduction of a second stream (or series) of the IIJ focusing broadly on ICT and society. As such, going forward the existing IIJ journal will be denoted as Series B and will focus on the broad areas of Computing, Communications, Engineering & Internet Technologies. The new second series (denoted as Series A) will focus on the broad field of Community Informatics & Society. Together with this development is the introduction of an additional co-editor for IIJ Series B, namely Dr Henri Uranus from the Universitas Pelita Harapan (UPH), in Jakarta, Indonesia. The IIJ welcomes his strong leadership and vision.
This regular issue of the journal brings four (4) papers from differing areas of technology. The first paper concerns the use of artificial neural networks to predict the shelf life of processed cheese. Solving this problem is relevant to the dairy industry in India, as the authors are from National Dairy Research Institute, in Karnal, India.
The editors can be reached individually at the following email addresses.
Thomas Hardjono is at [email protected], Budi Rahardjo is at [email protected], Henri Uranus is at [email protected], while Kuncoro Wastuwibowo is at [email protected].
The second paper addresses e-commerce developments in
Indonesia, notably among small, medium and micro Enterprises (SMMEs). The paper seeks to discover the barriers to the adoption of e-commerce among SMEEs in Indonesia, paying particular attention to the non-oil-and-gas processing sector and part of the services sector.
Character recognition is the topic of the third paper. The paper reports on work being done to use open-source Android software on a mobile device to recognize vehicle plate-numbers. The approach is based on the use of artificial neural networks with back-propagation, where training of the neural network is carried out in a desktop computer instead of mobile device to speed up the training. The work has promise to become the basis for low-cost plate number recognition usable for the government sector.
The fourth paper reports on the early development of a low-cost Computer Numerical Control (CNC) machine. Such a machine would be attractive to many small businesses in Indonesia, which so far have had to rely on imported machines. The resulting “home-made” CNC machine allows three different kinds of tools to be manually interchanged.
Thomas Hardjono
Budi Rahardjo Henri Uranus
Kuncoro Wastuwibowo
Editors’ Introduction
W
2 INTERNETWORKING INDONESIA JOURNAL Vol. 4/No. 1 B (2012)
ISSN: 1942-9703 / © 2012 IIJ
Vol.4/No.1 B (2012) INTERNETWORKING INDONESIA JOURNAL 3
ISSN: 1942-9703 / © 2012 IIJ
Abstract— This paper presents the latency of Artificial
Neural Network based Elman models for predicting the shelf life
of processed stored at 30oC. Soluble nitrogen, pH, standard
plate count, yeast & mould count, and spore count were taken as
input parameters, and sensory score as output parameter. Mean
square error, root mean square error, coefficient of
determination and nash - sutcliffo coefficient performance
measures were used for testing prediction potential of the
developed models. In this study, Elman models predicted the
shelf life of processed cheese very close to the experimentally
determined shelf life.
Index Terms— ANN, Shelf Life, Elman, Backpropagation,
Processed Cheese
I. INTRODUCTION
ROCESSED cheese is very nutritious and generally
manufactured from ripened Cheddar cheese, but
sometimes less ripened Cheddar cheese is also added in lesser
proportion. Its manufacturing technique includes addition of
emulsifier, salt, water and sometimes selected spices. The
mixture is heated in jacketed vessel with continuous stirring
in order to get homogeneous paste. This variety of cheese has
several advantages over raw and ripened Cheddar cheese,
such as tastier and longer shelf life. It is a protein rich food,
and is a comparable supplement to meat protein. Neural
Networks have become very famous topic of interest since last
few years and are being implemented in almost every
technological field to solve wide range of problems in an
easier and convenient way. Such a great success of neural
networks has been possible due to their sophisticated nature
as they can be used with ease to model many complicated
functions. In human body, neural networks are the building
blocks of the nervous system which controls and coordinates
the different human activities. Neural network consists of a
Sumit Goyal is Senior Research Fellow at the National Dairy Research
Institute, Karnal, India. He can be contacted at [email protected].
Gyanendra Kumar Goyal is Emeritus Scientist at the National Dairy
Research Institute, Karnal, India.
group of neurons (nerve cells) interconnected with each other
to carry out a specific function. Each neuron or a nerve cell is
constituted by a cell body call cyton and a fiber called axon.
The neurons are interconnected by the fibrous structures
called dendrites by the help of special gapped connections
called synapses. The electric impulses (called Action
Potentials) are used to transmit information from neuron to
neuron throughout the network. Artificial neural networks
(ANNs) have been developed based on the similar working
principle of human neural networks. Artificial Neurons are
similar to their biological counterparts. The input connections
of the artificial neurons are summed up to determine the
strength of their output, which is the result of the sum being
fed into an activation function, the most common being the
Sigmoid Activation Function which gives output varying
between 0 (for low input values) and 1 (for high input
values).
Fig.1. Input and output parameters for Elman backpropagation models
The resultant of this function is then passed as the input to
other neurons through more connections, each of which are
weighted and these weights determine the behavior of the
network. An ANN is devised for specific applications, such as
Artificial Intelligence Elman Backpropagation
Computing Models for Predicting Shelf Life of
Processed Cheese
Sumit Goyal and Gyanendra Kumar Goyal
P Soluble
nitrogen
pH
Standard
plate count
Yeast &
mould
count
Spore
count
Sensory
Score
4 INTERNETWORKING INDONESIA JOURNAL GOYAL & GOYAL
pattern recognition or data classification, through a learning
process [1]. ANNs predicted shelf life of Kalakand [2], milky
white dessert jeweled with pistachio [3], and instant coffee
flavoured sterilized drink [4, 5]. Time-Delay and Linear
Layer ANN models were developed for predicting shelf life of
soft mouth melting milk cakes [6], and soft cakes [7]. Radial
Basis models were successfully applied for predicting shelf
life of brown milk cakes [8]. The objective of this research is
to develop Elman single and multilayer backpropagation
computing models for predicting the shelf life of processed
cheese stored at 30o C.
II. MATERIAL AND METHODS
A. Data Modeling
The data consisted of 36 samples, which were divided into
two subsets, i.e., 30 were used to train and 6 to validate the
Elman backpropagation models. Soluble nitrogen, pH,
standard plate count, yeast & mould count, and spore count
were taken as input variables, and sensory score as output
variable for developing Elman single and multilayer models
(Fig.1).
Several combinations were tried and tested, as there is no
defined rule of getting good results rather than hit and trial
method. As the number of neurons increased, the training
time also increased. Algorithms like Levenberg Marquardt
algorithm, Gradient Descent algorithm with adaptive
learning rate, Bayesian regularization, Powell Beale restarts
conjugate gradient algorithm and BFG quasi-Newton
algorithms were tried. Backpropagation algorithm based on
Bayesian regularization was finally selected for training the
networks, as it gave most promising results. ANN was trained
upto100 epochs with single as well as double hidden layers
and transfer function for hidden layer was tangent sigmoid
while for the output layer, it was pure linear function.
The Neural Network Toolbox under MATLAB software was
used for developing the models. Training pattern of Elman
models is presented in Fig.2.
Fig .2. Training pattern of Elman models
Weights Selection
Error Evaluation
Weights Adjustment
Training Elman models
Training
Dataset
Error Calculation
Minimum
Error
End
No
Yes
Vol.4/No.1 B (2012) INTERNETWORKING INDONESIA JOURNAL 5
ISSN: 1942-9703 / © 2012 IIJ
B. Measures for Prediction Performance
Mean Square Error (MSE) (1), Root Mean Square Error
RMSE (2), Coefficient of Determination: R2 (3), and Nash -
Sutcliffo Coefficient: E2 (4) were used in order to compare the
prediction ability of the developed models.
2
1
expN
cal
n
QQMSE
(1)
2
1 exp
exp1 Ncal
Q
nRMSE
(2)
2
12
exp
exp2 1N
cal
Q
QQR
(3)
2
1 expexp
exp2 1N
cal
QQE
(4)
Where,
expQ = Observed value;
calQ = Predicted value;
expQ =Mean predicted value;
n = Number of observations in dataset.
III. RESULTS AND DISCUSSION
Elman model’s performance matrices concerning the single
layer and multilayer models for predicting sensory scores
based on equations 1, 2, 3 and 4 are presented in Table 1 and
Table 2, respectively.
Table 1: Performance of single layer for predicting sensory score
Neurons MSE RMSE R2 E
2
3 7.04669E-05 0.008394454 0.991605546 0.999929533
5 0.000176202 0.013274109 0.986725891 0.999823798
7 0.001911122 0.043716386 0.956283614 0.998088878
9 0.000191461 0.01383696 0.98616304 0.999808539
11 0.000227001 0.015066536 0.984933464 0.999772999
13 0.00022446 0.014981988 0.985018012 0.99977554
15 0.000428985 0.020711958 0.979288042 0.999571015
17 0.001079962 0.03286278 0.96713722 0.998920038
20 0.000437028 0.020905211 0.979094789 0.999562972
25 0.000491983 0.022180685 0.977819315 0.999508017
30 1.21846E-05 0.003490644 0.996509356 0.999987815
Table 2: Performance of multilayer for predicting sensory score
Neurons MSE RMSE R2 E
2
3:3 8.77114E-06 0.002961612 0.997038388 0.999991229
4:4 0.006907863 0.083113554 0.916886446 0.993092137
5:5 2.93847E-05 0.005420764 0.994579236 0.999970615
6:6 0.007876012 0.088746898 0.911253102 0.992123988
7:7 0.000225112 0.015003729 0.984996271 0.999774888
8:8 0.007324755 0.085584785 0.914415215 0.992675245
9:9 0.008067535 0.089819455 0.910180545 0.991932465
10:10 0.000224894 0.014996482 0.985003518 0.999775106
11:11 7.57117E-05 0.008701245 0.991298755 0.999924288
12:12 0.000252731 0.015897527 0.984102473 0.999747269
13:13 2.35759E-05 0.004855497 0.995144503 0.999976424
14:14 0.01241181 0.111408301 0.888591699 0.98758819
15:15 0.000110574 0.010515413 0.989484587 0.999889426
16:16 0.000117277 0.01082945 0.98917055 0.999882723
The developed Elman ANN single and multilayer models
showed that single layer model with 30 neurons performed
the best (MSE: 1.21846E-05, RMSE: 0.003490644, R2:
0.996509356, E2: 0.999987815); while multilayer with 3:3
neurons (MSE: 8.77114E-06, RMSE: 0.002961612, R2:
0.997038388, E2: 0.999991229) performed the best. On
comparing them with each other, it was observed that
multilayer model performed better. Therefore, it was selected
6 INTERNETWORKING INDONESIA JOURNAL GOYAL & GOYAL
for predicting the shelf life. The comparison of Actual
Sensory Score (ASS) and Predicted Sensory Score (PSS) for
single layer and multilayer models are illustrated in Fig.3 and
Fig.4, respectively.
Fig. 3. Comparison of ASS and PSS single layer model
Fig. 4. Comparison of ASS and PSS for multilayer model
Coefficient of determination (R2) was calculated based on the
total variation as explained by sensory scores. Period of
storage (days) for which the processed cheese has been in the
shelf is illustrated in Fig.5.
Fig.5. Sensory score and period of storage for processed cheese
The shelf life is calculated by subtracting the obtained value
of days from experimentally determined shelf life, which was
found to be 29.31 days. The predicted value is quiet close to
the experimentally obtained shelf life of 30 days suggesting
that the developed model is quite effective in detecting the
shelf life of processed cheese.
IV. CONCLUSION
Soluble nitrogen, pH, standard plate count, yeast & mould
count, and spore count were taken as input variables, and
sensory score as output variable for developing Elman models
for predicting the shelf life of processed cheese stored at 30o
C. Mean square error, root mean square error), coefficient of
determination and nash - sutcliffo coefficient were used in
order to compare the prediction ability of the developed
models. Bayesian regularization was selected for training
Elman ANN models. Regression equations were developed
for predicting the shelf life of processed cheese which gave
29.31 days shelf life vis-à-vis 30 days experimentally obtained
shelf life. From the study it is concluded that the developed
Elman models are very efficient for predicting the shelf life of
processed cheese.
REFERENCES
[1] http://www.engineersgarage.com/articles/artificial-neural-networks
(accessed on 11.1.2011).
[2] Sumit Goyal and G.K. Goyal “Advanced computing research on cascade
single and double hidden layers for detecting shelf life of kalakand: an
artificial neural network approach”. International Journal of
Computer Science & Emerging Technologies, vol. 2, no.5, pp. 292- 295,
2011.
Vol.4/No.1 B (2012) INTERNETWORKING INDONESIA JOURNAL 7
ISSN: 1942-9703 / © 2012 IIJ
[3] Sumit Goyal and G.K. Goyal “A new scientific approach of intelligent
artificial neural network engineering for predicting shelf life of milky white
dessert jeweled with pistachio”. International Journal of Scientific and
Engineering Research, vol. 2, no.9, pp. 1-4, 2011.
[4] Sumit Goyal and G.K. Goyal “Cascade and feedforward backpropagation
artificial neural networks models for prediction of sensory quality of instant
coffee flavoured sterilized drink”. Canadian Journal on Artificial
Intelligence, Machine Learning and Pattern Recognition, vol. 2, no.6,
pp.78-82, 2011.
[5] Sumit Goyal and G.K. Goyal “Application of artificial neural engineering
and regression models for forecasting shelf life of instant coffee drink”.
International Journal of Computer Science Issues, vol.8, no.4, pp.320-
324, 2011.
[6] Sumit Goyal and G.K. Goyal “Development of intelligent computing
expert system models for shelf life prediction of soft mouth melting milk
cakes”. International Journal of Computer Applications, vol.25, no.9, pp.
41-44, 2011.
[7] Sumit Goyal and G.K. Goyal “Simulated neural network intelligent
computing models for predicting shelf life of soft cakes”. Global Journal
of Computer Science and Technology, vol.11, no.14, Version1.0, pp. 29-
33, 2011.
[8] Sumit Goyal and G.K. Goyal ” Radial basis artificial neural network
computer engineering approach for predicting shelf life of brown milk
cakes decorated with almonds”. International Journal of Latest Trends in
Computing, vol.2, no.3, pp.434-438, 2011.
Sumit Goyal is Senior Research Fellow at National Dairy
Research Institute, Karnal (Haryana) India. He received
Bachelor of Information Technology degree and Masters in
Computer Applications from a central university of
Government of India, New Delhi, India. He received his
Master of Philosophy degree in computer science from
Vinayak Missions University, Tamil Nadu, India. His
research interests have been in the area of artificial neural
networks and prediction of shelf life of food products. His
research has appeared in Canadian Journal on Artificial
Intelligence, Machine Learning and Pattern Recognition,
International Journal of Computer Applications,
International Journal of Computational Intelligence and
Information Security, International Journal of Latest
Trends in Computing, International Journal of Scientific
and Engineering Research, International Journal of
Computer Science Issues, International Journal of
Computer Science & Emerging Technologies, Global
Journal of Computer Science and Technology, amongst
others. He is member of IDA.
Gyanendra Kumar Goyal is Emeritus Scientist at National
Dairy Research Institute, Karnal, India. He received the B.S.
degree and Ph.D. degree from Panjab University,
Chandigarh, India. In the year 1985-86, he did specialization
in Dairy and Food Packaging from School of Packaging,
Michigan State University, USA; and in the year 1999 he
specialized in Education Technology at Cornell University,
New York, USA. His research interests include dairy & food
packaging and shelf life determination of food products. He
has published more than 150 research papers in national and
international journals. His research work has been published
in Int. J. of Food Sci. Technol. and Nutrition, Nutrition and
Food Science, Milchwissenschaft, American Journal of
Food Technology, British Food Journal, Canadian Journal
on Artificial Intelligence, Machine Learning and Pattern
Recognition, International Journal of Computer
Applications, International Journal of Computational
Intelligence and Information Security, International
Journal of Latest Trends in Computing, International
Journal of Scientific and Engineering Research,
International Journal of Computer Science Issues,
International Journal of Computer Science & Emerging
Technologies, Global Journal of Computer Science and
Technology, amongst others. He is life member of IDA and
AFST (I).
8 INTERNETWORKING INDONESIA JOURNAL Vol. 4/No. 1 B (2012)
ISSN: 1942-9703 / © 2012 IIJ
Vol.4/No.1 B (2012) INTERNETWORKING INDONESIA JOURNAL 9
ISSN: 1942-9703 / © 2012 IIJ
Abstract—The contribution of Indonesian small, medium, and
micro enterprises (SMMEs) to the national economy growth has
been proven. Today one form of SMMEs empowerments is the
adoption of e-commerce. Although e-commerce adoption by
SMMEs has many advantages, a number of barriers for the
adoption exist. By understanding the current level and the
barriers of e-commerce adoption by SMMEs, it is expected that
the empowerment of SMMEs can be improved. This paper
reports a study on e-commerce adoption by Indonesian SMMEs
in the non-oil-and-gas processing and the trading, hotels, and
restaurants sectors by using business environment framework.
This study found that most of surveyed Indonesian SMMEs have
strategic plans to adopt higher level of e-commerce. Further, it is
found that human resources, sources of information, and push
forces factors are the significant barriers for e-commerce
adoption by Indonesian SMMEs.
Index Terms—electronic commerce, information technology
adoption, adoption barrier, SMMEs, business environment
framework
I. INTRODUCTION
mall and medium enterprises (SMEs) play a major role in
countries at all levels of economic development [1]. In
addition to the contribution from small and medium
enterprises, Indonesia also recognizes the contribution of
micro enterprises to the growth of the national economy
through employment provision and gross domestic product
(GDP) value which reach 91,03% and 33,08% of national
achievement, respectively [2]. Therefore, the empowerment of
small, medium, and micro enterprises (SMMEs) is one of
Indonesian government’s programs to develop the national
economy [3].
One form of recent SMMEs empowerment is the adoption
of e-commerce. There are some different definitions of e-
commerce from literatures. This study use a definition of e-
commerce as “a process of information exchange and
transaction, which involving products and services, using
Manuscript received June 30, 2012. This work was supported in part by the
Institut Teknologi Bandung under International Research Grant.
Rajesri Govindaraju is with the Faculty of Industrial Technology, Institut
Teknologi Bandung, Bandung, Indonesia (corresponding author; e-mail:
Dissa R. Chandra is with the Faculty of Industrial Technology, Institut
Teknologi Bandung, Bandung, Indonesia (e-mail: [email protected]).
information technology, such as networks, softwares, non-
wireless electronic equipments, and wireless electronic
equipments” [4].
In order to enhance the e-commerce adoption by Indonesian
SMMEs, Indonesian government has to support the SMMEs.
In this situation, it is crucial for Indonesian government to
understand the current adoption level and the adoption
barriers. Based on the understandings, the government’s
programs for Indonesian SMMEs will be focused on the real
problems effectively.
In the previous studies on e-commerce adoption, Diffusion
of Innovation theory and Technology Acceptance Model
dominated the researches framework [5]. The focus of the
models is the upstream issues of e-commerce, so the detail
aspects in the firms and other external aspects that influence
the firms were not identified. Furthermore, several studies also
used other frameworks, such as external and internal
framework [6, 7] and stakeholder theory [8].
Some of the previous researches, such as [6, 8], studied e-
commerce adoption by Indonesian SMEs. In [6] the barriers of
e-commerce adoption are classified into “too difficult”,
“unsuitable”, and “time/choice” factors. Meanwhile, [8]
identifies “Government”, “Employees”, “Owners”,
“Competitors”, “Customers” and “Managers” as the influential
stakeholders in e-commerce adoption by Indonesian SMEs.
However, in the studies the external barriers of e-commerce
adoption were not identified as detail as the internal barriers.
Therefore, this study aims to give a comprehensive and
detail understanding in e-commerce adoption by Indonesian
SMMEs. Therefore, the questions in this study are: (1) “What
is the current majority level of e-commerce adoption by
SMMEs being surveyed in Indonesia?” and (2) “What are the
barriers of e-commerce adoption by SMMEs in Indonesia?”
II. METHODOLOGY
Many studies identified and used classification models of e-
commerce adoption stages. One of them is a simple and
widely used model in [9]. There is also a model in [10] that is
based on enterprises characteristics clustering. Recently, a
standard model of e-government adoption was released by the
UN/ASPA. Although the model was set up to examine the
stage of e-government adoption, the model also can be used to
assess e-commerce adoption in developing countries [11].
Analysis of Level and Barriers of E-Commerce
Adoption by Indonesian Small, Medium, and
Micro Enterprises (SMMEs)
Rajesri Govindaraju and Dissa R. Chandra
S
10 INTERNETWORKING INDONESIA JOURNAL GOVINDARAJU & CHANDRA
TABLE I
E-COMMERCE ADOPTION STAGE
E-Commerce
Adoption Stage Description
Non-adopter · No website
Level 1 : Presence
[16]
· Use websites (including Facebook, Multiply, Indonetwork domain, etc.) only to display information about products and
services (provide a one-way communication on the website)
Level 2 : Portals
[16]
· Use websites (including Facebook, Multiply, Indonetwork domain, etc.) for two-way communication with customers and
suppliers
· Provide services such as product and customization order, product feedback, and also surveys
Level 3 :
Transaction
integration
[16]
· Use websites (including Facebook, Multiply, Indonetwork domain, etc.) for two-way communication with customers and
suppliers
· Provide services such as product and customization order, product feedback, and also surveys
· Provide online payment and / or online order fulfilment
Level 4 :
Enterprise
integration
[16]
· Use websites (including Facebook, Multiply, Indonetwork domain, etc.) for two-way communication with customers and
suppliers
· Provide services such as product and customization order, product feedback, and also surveys
· Provide online payment and / or online order fulfilment
· Integrate internal processes with online booking
· Implement Supplier Relationship Management (SRM) and / or Customer Relationship Management (CRM)
In addition, a model of e-commerce adoption stage is
presented in Table I [12]. By adding a non-adopter stage, the
model was used in this study to categorize the “level of e-
commerce adoption by SMMEs”.
To analyze the barriers of e-commerce adoption, Business
Environment framework [13] is used in this study. The
framework is presented in Figure 1. Business Environment
framework is an enhancement of Marketing Environment
framework [14] that has been widely used as a theoretical base
of marketing studies. Business Environment framework was
chosen because it can provide a comprehensive approach to
analyze the aspects of a company and its business
environment. The theory gives us a way to view a company as
a dynamic system.
Four components of Business Environment framework are
eliminated from the initial model. The elimination process was
done based on its effect to e-commerce business operation by
SMMEs. They are “machine” and “material” component in the
internal environment and “natural” and “demography”
component in the macro external environment.
Based on previous studies [6, 7, 15-22], various variables
influencing e-commerce adoption by SMMEs were identified.
Those variables were grouped into the components of the
Business Environment framework. Selection processes of the
variables were based on the real situation in Indonesia. To get
a better understanding about real situation in Indonesia, a pilot
study were done by involving 3 (three) SMMEs. The latent
and the measured variables that were used in the initial model
are presented in Table II.
Business Environment
Component
Internal
Environment
External
Environment
Material
Management
Machine
Money
Man
Micro
Environment
Macro
Environment
Suppliers
Customers
Market
Intermediaries
Competitor
Public
Economic
Political
Socio-cultural
Technology
Natural
Demography
International
Figure 1. Business Environment Component (Jain, 2009)
The e-commerce adoption barriers for SMMEs influence
their e-commerce adoption. In this regards, to establish a
structural model, we added a latent variable "e-commerce
adoption”. The variable represents the concept of SMMEs’ e-
commerce adoption level which is presented in Table I.
For further data processing, the latent variable “e-commerce
adoption” and its’ measured variable were transformed into
the “obstructed e-commerce adoption” (H) latent variable and
“TA” measured variable.
TA = 4 - "level of e-commerce adoption by SMMEs"
Indonesian government suggests that enterprises grouping
in Indonesia can be done based on nine economic sectors [23].
The sectors with high employment levels, gross domestic
product value, and export values are: (1) trading, hotels, and
restaurants sector, and (2) processing industry sector. In
addition, the processing industry sector is divided into oil-
and-gas industry and non-oil-and-gas industry. Oil-and-gas
industry sector only contains large-scale enterprises, while
non-oil-and-gas sector is an industry sector with a large
number of SMMEs. Therefore, SMMEs in non-oil-and-gas
processing sector and trading, hotels, and restaurants sector
were chosen as the focus of this study. The classification of
SMMEs in this study is based on assets value and gross
income value criteria as stated in [24] and shown in Table III.
A questionnaire set was developed using 4 point Likert like
scaling, from strongly agree until strongly disagree.
Questionnaires were given to 400 SMMEs using email and
direct meeting. Follow-up emails and phone calls were made
to encourage respondents. The response rate was 21.75% with
87 returned questionnaires, but there are only 85 complete
questionnaires that can be used for further analysis.
Respondents are non-oil-and-gas processing, trading, and
restaurants SMMEs in 8 provinces. There is no sample from
hotel business. The majority is the trading business. Majority
of the sample is the micro scale enterprises. The SMMEs’ age
ranged between below 1 year until 61 years and their e-
commerce experience ranged between below 1 year until 7
years. The respondents’ profiles are shown in Table IV.
Vol.4/No.1 B (2012) INTERNETWORKING INDONESIA JOURNAL 11
ISSN: 1942-9703 / © 2012 IIJ
TABLE II
INITIAL LATENT AND MEASURED VARIABLES
TABLE III
CLASSIFICATION OF INDONESIAN SMMES [24]
Enterprises Assets (excluding land
and buildings) Gross Income
Micro ≤ Rp.50,000,000.00 ≤ Rp. 300,000,000.00
Small > Rp. 50,000,000.00
- Rp. 500,000,000.00
> Rp. 300,000,000.00
- Rp. 2,500,000,000.00
Medium > Rp. 500,000,000.00
- Rp. 10,000,000,000.00
> Rp. 2,500,000,000.00
- Rp. 50,000,000,000.00
TABLE IV
RESPONDENTS’ PROFILES
Profile Category % Profile Category %
SMMEs' Age
≤ 1 year 49.4% Desktop
Yes 63.5%
> 1 year - ≤ 5 years 30.6% No 36.5%
> 5 years - ≤ 10 years 8.2% Laptop
Yes 69.4%
> 10 years 11.8% No 30.6%
Economic
Sector
Processing Industry 20.0% LAN
Yes 24.7%
Trading 68.2% No 75.3%
Restaurant 11.8% Internet
Yes 77.7%
SMMEs'
Classification
Micro 76.5% No 22.3%
Small 16.5% Camera
Yes 67.1%
Medium 7.1% No 32.9%
E-Commerce
Adoption
Stage
Non-adopter 30.6% Scanner
Yes 32.9%
Presence 21.2% No 67.1%
Portals 38.8%
Transactions
Integration
8.2%
Enterprises
Integration
1.2%
III. RESULT
Construct validity of each variable was assessed by
calculating the correlation between the value of each variable
and the sum value of each respondent answer using software
SPSS 17.0. Variables with insignificant correlation were
eliminated from the model. Then, reliability was assessed by
calculating the Cronbach alpha coefficient. The calculation
showed that the Cronbach alpha coefficient was above 0.7. In
the next step, we found 7 variables with non-normal
distribution from normality test by calculating z score of
data’s skewness and curtosis. After transforming them using
LISREL 8.7, we eliminated 2 variables. Then, sample size
adequacy and multicolinearity were calculated using KMO
and Bartlett’s Test respectively. The KMO value was 0.780. It
means that the sample size was enough for statistical analysis.
Lastly, based on the value of each variable’s MSA, we had to
eliminate 1 variable. The transformed and eliminated variables
are shown in Table V.
TABLE V
DELETED AND TRANSFORMED VARIABLES
Test Transformed Eliminated
Construct
Validity
PO3
PO4
T2
SE2
Reliability -
Normality
ME2 -
ME4 ME4
ME5 -
P5 -
PU2 PU2
PU3 -
PO2 -
Multicolinearity SE3
Exploratory Factor Analysis (EFA) was conducted to
validate the initial model. There were 10 factors extracted
from 37 measured variables in the EFA’s result. In the initial
model, there were 9 components from 44 measured variables.
The difference of measured variables was caused by the
elimination of 7 measured variables in the data preparation
process.
Although the number of latent variables and factors were
different, there were 2 factors from EFA that were exactly the
same with 2 latent variables defined in the initial model. They
are shown in the Table VI. In addition, high similarity is
indicated by factor 1 with "man" latent variable, factor 2 with
"customers" latent variable, factor 4 with "partners" latent
variable, and factor 5 with "technology" latent variable. It is
shown in the Table VII.
To obtain appropriate names for the factors, generalizations
were done. It could be observed that all of the measured
variables in each factor expressed a general idea. Then, the
general idea was adjusted and established as the factors’ name.
As the result, factor 2 expresses "market" which are products,
customers, and service delivery providers, factor 4 expresses
"internet services" as the SMMEs partners, factor 5 represents
"security" of information technology, factor 7 expresses "push
forces" which is encouragements from the internal and
external environment, factor 8 is "sources of information"
about e-commerce for SMMEs, factor 9 measures supports
and inputs from "enterprises association", and factor 10
associates with the “e-commerce popularity” in general.
Environment Latent Variabel Measured Variabel Code Environment Latent Variabel Measured Variabel Code
Lack of educated human resources M1 Partners have not implemented e-commerce yet P1
Lack of resources with IT skills M2 High cost of internet access P2
Employee resistance M3 Bad service from internet providers P3
Lack of knowledge about the benefits of e-commerce M4 Low internet speed P4
Lack of knowledge about the implementation of e-commerce M5 Lack of application developer P5
Lack of experience using information technology / electronic marketing media M6 Low quality of service delivery P6
Lack of time to learn about e-commerce M7 The absence of the threat from competitors who are using e-commerce PU1
Lack of time to operate e-commerce M8 Lack of support and input from SMMEs associations PU2
Financial investment for the implementation of e-commerce is too high F1 Lack of support and input from import export companies association PU3
Term of return on investment and capital is too long F2 Limited number of other organizations that have used e-commerce PU4
Lack of financial resources F3 Lack of information about e-commerce from media PU5
Insuitability with the characteristics of e-commerce products ME1 Ambiguity of government regulation PO1
Incompatibility of e-commerce with the way organizations conduct business process ME2 Unclear rules of international cooperation in trading PO2
Absence of standards of quality between sellers and buyers ME3 High intensity of changes in government regulations PO3
Lack of support for corporate strategic objectives ME4 Lack of support from the government PO4
Lack of support from management ME5 Limited support from online payment system that is feasible and effective T1
High level of risk decision to use e-commerce ME6 Potential security risks in internet network T2
Incompatibility between customers and e-commerce K1 Information privacy issues T3
Low level of trust from customers about the product quality K2 Low on-line payment security T4
Low level of trust from customers regarding delivery time K3 Low popularity of on-line sales and marketing SE1
Low level of trust from customers about the added-value of buying the products K4 Instability of national economic climate SE2
Limited numbers of customers who use internet K5 Poor condition of transportation infrastructure in Indonesia SE3
Internal
Environment
Micro External
Environment
Partners
Publics
Macro External
Environment
Political
Technology
Socio-economy
Man
Financial
Management
Customers
Micro External
Environment
12 INTERNETWORKING INDONESIA JOURNAL GOVINDARAJU & CHANDRA
TABLE VI IDENTICAL FACTORS
TABLE VII
OTHER FACTORS
Furthermore, there were K5 and P1 in the factor 1.
Conceptually, these two measured variables are very different
from the other internal environment measured variables in
factor 1. The sources of P1 and K5 are external factors, which
can be affected, but can not be controlled by the SMMEs.
Therefore, P1 and K5 were taken out from factor 1.
Both, P1 and K5, state the use of internet by customers and
the use of e-commerce by supplier. Internet is the basic
information technology for e-commerce. Furthermore,
customers and suppliers are the elements of SMMEs’ supply
chain. Thus, both of the measured variables represent the use
of information technology in SMMEs supply chain. The new
latent variable for P1 and K5 is called as "supply chain
information technology". The revised model is shown in the
Table VIII.
TABLE VIII
REVISED COMPONENT
The data processing was continued using Partial Least
Square (PLS). PLS was used to calculate each latent variable’s
significance in influencing the overall e-commerce adoption
process by Indonesian SMMEs. The data processing result is
shown in the Figure 2. Three latent variables influence
significantly the “obstructed e-commerce adoption” in
Indonesian SMMEs with T-value above 1.67. They are “Push
Forces”, “Man” and “Sources of Information”.
IV. DISCUSSION
From the data, it is known that 69.4% respondents are e-
commerce adopters. Further, majority of the adopters are at
the second level of e- commerce adoption stage. Although
only a small number of SMMEs adopted e-commerce on a
higher level, but most of the SMMEs have desire to move on
and increase their level of e-commerce adoption. It can be
viewed based on 77.97% of adopter SMMEs and 57.69% of
non-adopter SMMEs who stated their desire to increase their
level of e-commerce adoption in the survey. It reveals that
most of the surveyed SMMEs have low-level of e-commerce
adoption, but they have strategic plans to increase the level.
Based on the result, “Push Forces”, which is reflected by
support from management and threat from competitors, is still
not enough to push e-commerce adoption by Indonesian
SMMEs. For the aspect of competition threat, the result
indicates that there are not many e-commerce adopter
enterprises in Indonesian market. Therefore, SMMEs believe
that there is an absence of threat from e-commerce adopter
enterprises. This result is consistent with the result found in
[8] about the influence of “Owners”, who in general manage
the SMMEs directly, and “Competitors”.
The result also indicates that the quality of Indonesian
SMMEs human resources leads to the low e-commerce
adoption in their companies. It is also consistent with the
result in [8]. This quality is measured by the number of
educated employees, the number of employees with sufficient
Latent
VariabelCode Factor Code
F1 F1
F2 F2
F3 F3
PO1 PO1
PO2 PO2
Initial Model EFA
Financial 3
Political 6
Latent
VariabelCode Factor Code
M1 M1
M2 M2
M3 M3
M4 M4
M5 M5
M6 M6
M7 M7
M8 M8
ME1 ME2
ME2 K5
ME3 P1
ME5 ME1
ME6 ME3
K1 K1
K2 K2
K3 K3
K4 K4
K5 P6
P1 P2
P2 P3
P3 P4
P4 ME6
P5 T1
P6 T3
PU1 T4
PU3 ME5
PU4 PU1
PU5 P5
T1 PU5
T3 9 PU3
T4 PU4
Socio-
economySE1 SE1
Partner
4
5
Publics 7
8
Technology
10
Initial Model EFA
Man
1
Management
2
Consumers
Code Code
M1 ME1
M2 ME3
M3 K1
M4 K2
M5 K3
M6 K4
M7 P6
M8 P5
ME2 PU5
F1Enterprises
AssociationPU3
F2 PU4
F3 SE1
ME5 ME6
PU1 T1
K5 T3
T4
PO1
P2 PO2
P3
P4
Supply Chain
Information
TechologyP1Micro External
Environment
Component
Internal
Environment
Man
Financial
Push Forces
Component
Micro External
Environment
Market
Internet Services
Sources of
Information
Macro External
Environment
E-commerce
Popularity
Security
Political
Vol.4/No.1 B (2012) INTERNETWORKING INDONESIA JOURNAL 13
ISSN: 1942-9703 / © 2012 IIJ
\
Figure 2. SmartPLS Result
IT skills, employees’ resistance, knowledge about the benefit
and implementation of e-commerce, time to learn about e-
commerce, and the organization way of doing business. It
indicates that in Indonesian SMMEs the organization way of
doing business is strongly bounded with the human resources.
Therefore, it is possible that the question about the
organization way of doing business had been translated by
respondents into the human resources way of doing their job.
The other result indicates that Indonesian SMMEs need
more sources of information about e-commerce. The sources
can be e-commerce application vendors, IT consultants and
public information media. In some cases, e-commerce
application vendors or consultants are also direct sources of e-
commerce information.
Eight latent variables have no significant influence on e-
commerce adoption by Indonesian SMMEs. It is good for
further studies to test again the influence of those variables in
a better setting, such as better variable definition, better
operationalization, and better data sample.
This study did not use a proportional sample because we did
not know about e-commerce adoption level proportion of the
population. Until now, it has been no census data about e-
commerce adoption by SMMEs in Indonesia. However, we
realize that we could use SMMEs’ geographic location or
government classification to define the SMMEs groups and
their proportion. Future studies could consider using a
proportional sample.
V. CONCLUSION
From the study, it is found that most of Indonesian SMMEs
participating in the study have strategic plans to adopt higher
level of e-commerce, though majority of the firms currently
adopt e-commerce at the lower level. Furthermore, there are
three significant barriers of e-commerce adoption by
Indonesian SMMEs. They are “Push Force”, “Man”, and
“Sources of Information”.
As “Man” is an internal barrier, it can be solved by the
SMMEs itself and by the help of government. Examples of the
solutions for this problem are recruitment of IT employees,
improvement of employees’ skill by SMMEs, and e-
commerce workshop for SMMEs employees held by
government. In relation to “Push Forces” barrier., to solve the
lack of management support in SMMES, the role of SMMEs
owner as a part of management is very important. Another
aspect of “Push Force” is the threat from enterprises using e-
commerce. The result indicates that e-commerce usage by
other companies has not threatened Indonesian SMMEs yet.
Consequently, it can be expected that government’s effort to
support Indonesian SMMEs to adopt ecommerce which can be
done through socialization, education and training may
increase the adoption rate exponentially.
The “Sources of Information” problems need to be solved
by external entities including government and other
associations outside the SMMEs. As an example, public
information media can be arranged by Indonesian government.
If the number and the quality of these sources are increased,
the barrier of e-commerce adoption in Indonesian SMMEs can
be reduced. In this case the role of government as the initiator
and facilitator is very important.
14 INTERNETWORKING INDONESIA JOURNAL GOVINDARAJU & CHANDRA
REFERENCES
[1] M. C. J. Caniëls and H. A. Romijn, "What works, and why, in
business services provision for SME: insights from evolutionary
theory," Managing Service Quality, vol. 15, 2005.
[2] Kementerian Koperasi Dan Usaha Kecil Dan Menengah Republik
Indonesia, "Perkembangan Data Usaha Mikro, Kecil, Menengah
(UMKM) Dan Usaha Besar (UB) Tahun 2005 - 2009," Jakarta,
2010.
[3] Kementerian Koperasi Dan Usaha Kecil Dan Menengah Republik
Indonesia, "Rencana Strategis Kementerian Koperasi Dan Usaha
Kecil Dan Menengah Republik Indonesia Tahun 2010 - 2014,"
Jakarta, 2010.
[4] R. Govindaraju and D. R. Chandra, "E-commerce Adoption
Barrier in Indonesian Micro, Small and Medium Enterprises
(SMEs)," in The International Conference on Industrial
Engineering and Business Management (ICIEBM) 2010
Yogyakarta, Indonesia, 2010.
[5] R. Mohamad and N. A. Ismail, "Electronic Commerce Adoption in
SME: The Trend of Prior Studies," Journal of Internet Banking
and Commerce, vol. 14, 2009.
[6] M. Kartiwi and R. C. MacGregor, "Electronic Commerce
Adoption Barriers in Small to Medium-Sized Enterprises (SMEs)
in Developed and Developing Countries: A Cross-Country
Comparison," Journal of Electronic Commerce in Organizations,
vol. 5, pp. 35-51, 2007.
[7] M. Kapurubandara and R. Lawson, "Barriers to Adopting ICT and
e-commerce with SMEs in Developing Countries: An Exploratory
study in Sri Lanka," in Conference of CollecteR Adelaide, 2006,
pp. 1-13.
[8] R. Govindaraju, D. R. Chandra, and Z. A. Siregar, "Stakeholder
Role in e-Commerce Adoption by Small and Medium Enterprises,"
in IEEE International Conference on Management of Innovation &
Technology Bali, Indonesia, 2012.
[9] R. Kalakota and M. Robinson, E-business 2.0 : Roadmap for
Success. New York: Addisom-Wesley, 2001.
[10] E. Daniel, H. Wilson, and A. Myers, "Adoption of E-Commerce by
SMEs in the UK: Towards a Stage Model," International Small
Business Journal, vol. 20, pp. 253–270, 2002.
[11] J. Toland, "E-commerce in Developing Countries," in
Encyclopedia of Ecommerce, E-government, and Mobile
Commerce Hershey: Idea Group Reference, 2006, pp. 308-313.
[12] S. S. Rao, G. Metts, and C. A. M. Monge, "Electronic Commerce
Development in Small and Medium Sized Enterprises : A Stage
Model and Its Implications," Business Process Management
Journal, vol. 9, pp. 11-32, 2003.
[13] R. Jain, Business Environment. New Delhi: V.K. Enterprises,
2009.
[14] P. Kotler and G. Armstrong, Principles of Marketing. Englewood
Cliffs: Prentice-Hall, 1991.
[15] M. A. Sarlak and A. A. Hastiani, "E-Business Barriers in Iran’s
Free Trade Zones," Journal of Social Sciences, vol. 4, pp. 329-333,
2008.
[16] M. Kartiwi, R. MacGregor, and D. Bunker, "Electronic Commerce
Adoption Barriers in Indonesian Small Medium-sized Enterprises
(SMEs): An Exploratory Study," in Americas Conference on
Information Systems Acapulco, Mexico, 2006.
[17] R. C. MacGregor, L. Vrazalic, S. Carlsson, D. Bunker, and M.
Magnusson, "The impact of business size and business type on
small business investment in electronic commerce: a study of
swedish small businesses," Australian Journal of Information
Systems, vol. 9, pp. 31-39, 2002.
[18] M. A. Sarlak, A. A. Hastiani, L. F. Dehkordi, and A. Ghorbani,
"Investigating on Electronic Commerce Acceptance Barriers in
Dried Fruits Exporting Companies of Iran," World Applied
Sciences Journal, vol. 6, pp. 818-824, 2009.
[19] O. Sorensen and R. Hinson, "E-Business Triggers: Further Insights
into Barriers and Facilitators amongst Ghanaian Non-Traditional
Exporters (NTEs)," in Consumer Behavior, Organizational
Development, and Electronic Commerce: Emerging Issues for
Advancing Modern Socioeconomies Hershey: Information Science
Reference, 2009, pp. 294-310.
[20] M. Stanfield and K. Grant, "Barriers to the take up of electronic
commerce among small-medium sized enterprises," Informing
Science, pp. 737-745, 2003.
[21] L. F. Sugianto and S. Jantavongso, "eBusiness Adoption Studies
Focusing on Thai SMEs," in International Conference on
Computational Intelligence for Modelling Control and Automation
and International Conference on Intelligent Agents,Web
Technologies and Internet Commerce Sydney, 2006.
[22] J. Sutanonpaiboon and A. M. Pearson, "E-Commerce Adoption:
Perceptions of Managers/Owners of Small- and Medium-Sized
Enterprises (SMEs) in Thailand," Journal of Internet Commerce,
vol. 5, pp. 53-82, 2006.
[23] Bagian Data - Biro Perencanaan Kementerian Koperasi Dan Usaha
Kecil Dan Menengah Republik Indonesia, "Statistik Usaha Kecil
dan Menengah Tahun 2007-2008." vol. 2010: DEPKOP, 2009.
[24] "Undang-Undang Republik Indonesia Nomor 20 Tahun 2008
Tentang Usaha Mikro, Kecil dan Menengah," P. R. Indonesia, Ed.,
2008.
Rajesri Govindaraju is an Associate Professor at the Faculty of Industrial
Technology, Institut Teknologi Bandung (ITB), Indonesia. She completed her
bachelor in informatics from ITB and PhD in enterprise information systems
implementation from University of Twente, The Netherlands. Her current
research areas include information systems, ERP, e-business systems
implementation and information system planning.
Dissa R. Chandra is an assistant professor in the Faculty of Industrial
Technology, Institut Teknologi Bandung (ITB), Indonesia. She received a
Bachelor Degree and Master Degree from Industrial Engineering Department
at ITB in 2010 and 2011 respectively. Her research interest is in the area of
Enterprise Information System and information system adoption.
Vol.4/No.1 B (2012) INTERNETWORKING INDONESIA JOURNAL 15
ISSN: 1942-9703 / © 2012 IIJ
Abstract— In this research an automobile license plate
recognizer is developed in a mobile device based on the android
platform to be used as an alternative solution to record road-side
transactions, such as parking and fine. The license plate
recognition is implemented using artificial neural network, where
training of the neural network is carried out in a desktop
computer instead of mobile device to speed up the training using
back propagation algorithm. Result of the training is the artificial
neural network along with their weight is implemented in the
mobile device to recognize characters in license plate. The
training result showed that the neural network is able to
recognize 88.2% characters in sample images outside the training
set. The image processing and neural network implemented in
the mobile device managed to recognize an average of 71%
characters in sample license plate images which are categorized
as having good image quality. Automobile license plate
information is also saved in database inside the mobile device.
Index Terms— android-base, back-propagation algorithm,
license plate recognition, neural network, pre-trained
I. INTRODUCTION
NDROID-based image recognition is one of the
applications that are used in many mobile devices that are
widely available nowadays. As an open source, android is
favorable and is applicable in image recognition as well. In
many parts of Jakarta city, the parking lots are along the street
and monitoring cars come and go is not easy as well as for
security purposes. A mobile device will be a good choice as
one of the media for the parking staff to capture license plate;
later on the image will be recognized and saved in the device.
License plate recognition is widely used in many automatic
parking systems in many countries and widely reported in
series of papers [1-10]. While in this paper we use a simple
back-propagation algorithm for license plate recognition. This
Manuscript received July 27, 2012. This work was supported in part by the
Computer Engineering Laboratory.
R. K. Romadhon was with the Computer Engineering Department, Binus
University, Jakarta, Indonesia (e-mail: rizqi_batavia @yahoo.com).
M. Ilham was with the Computer Engineering Department, Binus
University, Jakarta, Indonesia (e-mail: [email protected]).
N. I. Munawar was with the Computer Engineering Department, Binus
University, Jakarta, Indonesia (e-mail: [email protected]).
S. Tan is with the Computer Engineering Department, Binus University,
Jakarta, Indonesia (phone: +6221-534-5830 ext. 2205; e-mail:
R. Hedwig is with the Computer Engineering Department, Binus
University, Jakarta, Indonesia (phone: +6221-534-5830 ext. 2205; e-mail:
simple android-based application will benefit parking staff in
monitoring the cars along the street of Jakarta.
II. LITERATURE REVIEW
License Plate Recognition (LPR) has been actively studied
for couple of decades and still is an active subject of research
because of its wide use and potential application [1]. One of
the most popular approaches for License Plate recognition is
using artificial neural networks (ANN). Various approaches
using neural network can be seen in [1-4]. Variation of the
ANN used for LPR can been in for example in [1-2]. In [1] a
Radial Basis Function Neural Network (RBFNN) is used
instead of the more popular Back-Propagation (BP) neural
network. The evaluation showed better performance over BP
neural network. Other researcher [2] improved the BP learning
algorithm to increase recognition rate. In general there are
many possible algorithms using ANN to recognize characters
in license plate. This shows that the ANN is still a popular
choice to implement LPR. The benefit of using ANN for real-
time applications is because the training and classification
stage can be separated, where the classification time is much
faster than the training time. The classification of image
feature can thus be implemented in mobile devices with low
computing power.
Some researchers have even implemented the LPR
algorithm into Android device, such as in [3-4]. They include
image processing to detect the position of the license plate in
the capture image. Such approach can be time consuming in
low computing power devices such as Android devices. In this
paper the license plate detection problem is omitted simply by
having the user to point the camera directly in front of the
license plate with visual guide in the graphical user interface.
The mobile device then just have to perform light image
processing to remove noise and segmentation of the characters
before feeding them into the feed-forward neural network.
III. PRE-TRAINING THE NEURAL NETWORK
The neural network is trained to recognize a character in a
14x10 pixels-image containing a single character scaled to fill
the whole image area. Training is carried out in a desktop
computer to speed-up the process. There are 4 image samples
for each numeric and upper-case alphabet characters, hence
144 image samples are used as the training set for the neural
network. Each sample image is preprocessed manually by
cropping a character from a larger image, converting to B&W,
and resizing it to 14x10 pixels. Each larger image is a
Android-Based License Plate Recognition using
Pre-trained Neural Network
Rizqi K. Romadhon, Muhammad Ilham, Nofan I. Munawar, Sofyan Tan, and Rinda Hedwig
A
16 INTERNETWORKING INDONESIA JOURNAL ROMADHON ET AL.
photograph of a real automobile license plate. The authors
chose 14x10 pixel input image to minimize the calculation cost
when the neural network is implemented in an android mobile
device.
MATLAB® is used in the desktop computer to preprocess
image samples, construct and train the neural network [11].
After a character image is manually cropped from a larger
license plate image, it is converted from RGB format to
grayscale and then converted to black and white binary data,
where the threshold is calculated using Otsu algorithm [12-
13]. Each sample image is further scaled down to 14 pixels tall
and 10 pixels wide, suitable for the input layer of the neural
network. The whole 144 binary image samples are combined
into a training set.
The architecture of the neural network is a multilayer feed
forward neural network, where two hidden layers are used to
model the character recognition behavior as shown in figure 1.
The input layer is consisted of 140 nodes, each connected to
one pixel of the 14x10 input character images. The output
layer contains 36 nodes, each for one of 36 numerical and
alphabet characters (0 to 9 and A to Z). The first hidden layer
contains 44 nodes, while the second contains 36 nodes. Output
of all nodes in the hidden and output layers are calculated
using the sigmoid activation function (1) to produce
continuous output function with an output range between 0
and 1.
xexf
1
1)(
(1)
After training the neural network using back-propagation
algorithm in MATLAB®, the resulting weights matrix is
exported to be used in the mobile device as a fix weights
matrix implemented in the same neural network architecture,
but without the capability to retrain the weights.
IV. IMPLEMENTATION IN ANDROID-BASED MOBILE DEVICE
Most android device includes camera that can be used to
capture a license plate as a source image, however the source
image must be automatically preprocessed into 14x10-pixels
character images for recognition by the neural network. The
manual cropping used during training is not viable in the
implementation. Therefore a light image preprocessing
algorithm is developed in the mobile device to resize, convert,
and segment the source image into 14x10-pixels character
images.
The first step in the image preprocessing is pre-resizing the
source image into lower resolution to improve performance
and simply cropping out part of the source image that is not a
license plate number. Assuming that the aspect ratio of any
standard license plate is constant, the user can be guided to aim
the camera at a certain distance so that the license plate will
completely fill a certain area in the electronic view finder on
the mobile device. The application can then easily discard all
pixels outside the predefined area and size down the image
into 512x150 pixels for faster processing, while maintaining
most of the image details. Registration date information on the
plate is discarded in this pre-resizing step because of their
lower position in the plate.
Color information is not used in the image processing,
therefore the pre-resized source image is converted to
grayscale and then into black and white binary source image
for faster image processing.
The next step is to segment the binary source image into
multiple images each containing a single character. It is
assumed that all characters in the license plate is separated
from each other, therefore segmentation can be done by
uniquely labeling each island of connected pixels as a single
character. Connected pixels are search in the 8-neighbor of
each pixel as seen in figure 2. Figure 3 shows islands that
contain pixels less than a particular threshold number are
considered as noise and thus neglected. The big islands are
cropped out by finding the left, right, upper, and lower edge
coordinates of the islands in the source image. These cropped
out islands are then scaled and resized to 14x10 character
images.
Input
layer2 Hidden layers
Output
layer
Fig. 1. The 144 input nodes are directly fed from the 14x10 pixels of the
input character image, and then fully connected to the two hidden layers
containing 44 and 36 nodes respectively and finally to the 36 nodes of the
output layer.
N1 N2 N3
N4 N5
N6 N7 N8
Fig 2. The 8 neighbor pixels are searched for any connected pixels with binary
intensity of one.
L2 L5 L5
L5 L5 L5
L5 L5 L6
L3 L3 L5 L5 L6
L3 L3 L5 L5
L3 L5 L5
L5 L5
L5 L5 L5 L5
L4
Fig 3. In a low resolution example, five islands of connected pixels are found
in the above image, where one biggest island (labeled L5) is the character
image, while the other islands are considered as noise.
Vol.4/No.1 B (2012) INTERNETWORKING INDONESIA JOURNAL 17
ISSN: 1942-9703 / © 2012 IIJ
The same neural network architecture as those used in the
pre-training is implemented in the mobile device. Weights
matrix from the pre-training process is imported into the neural
network in the mobile device, resulting in exactly the same
network behavior as the pre-trained neural network. Character
images resulted from image preprocessing in the mobile device
are then fed into the neural network to be recognized as
number or alphabet characters.
For each license plate source image, there will be 3 to 9
characters depending on the plate. These characters are saved
in the local standard SQLite database in the android-based
mobile device. Information saved in the database is type of
automobile (car or motorcycle), license plate characters, and
the time of record. Receipt of the transaction record can be
printed using Bluetooth or Wifi interface to a remote printer.
V. RESULTS AND DISCUSSION
Even though the training result shows that the accuracy is as
high as 88.2%, it is in the contrary with the test result that was
performed in android-based mobile device. The result of image
recognition accuracy that is tested is shown in table 1 where all
license plates that were captured were in good condition. It is
seen that although the same license plate was captured two
times, but the reading result is different from one image to
another. Lightning position as well as camera position played
important role with the image recognition that should be
controlled since the beginning of capturing the image.
In the table 2, we did experiment by capturing the image of
license plate directly from the android-based mobile device
(dual core 1GHz) where some of the images were not well-
captured. It needed 15 to 20 seconds for the device to
recognize each image and it is seen that many of the
recognitions are failed. The image segmentation could not be
performed properly due to the poor images that were manually
captured by mobile device. This poor performance also found
when the license plate was not in the good condition such as
there was smooth crack in between the letters or pointing bolts
under the letters.
After image recognition process was done, the reading then
saved in database and in the form of file.txt format. The
database then is used as parking report which is contained of
type of vehicle, license plate number, and time. The file.txt
will be printed out by using PrintShare application and can be
used as parking receipt for the driver or the owner of the
vehicle.
VI. CONCLUSION
This simple experiment is done in order to build a license
place recognition application running on mobile device with
android operating system, which is an open source operating
system. Even though the result is still far from what is
expected for practical use, this small step can enhance
advanced development of this application in the near future for
road side automobile transaction system, such as parking and
ticketing.
TABLE 1
TEST RESULTS FOR LICENSE PLATE RECOGNITION IN GOOD CONDITION
No Image Capture Recognition
Result
Recognition
Percentage
1
B3289BJH 87.5 %
2
D328SHJH 50 %
3
B8692LH 71.43 %
4
B85926N 71.43 %
05
HS929SIF 75 %
Average: 71%
TABLE 2
TEST RESULTS FOR LICENSE PLATE RECOGNITION WHEN TAKING DIRECTLY
FROM ANDROID-BASED DEVICE
No Image Capture Recognition
Result
Recognition
Percentage
1
J157GPPT 63%
2
H66S7OKI 50%
3
H81S4DCK 38%
4
B6598I 75%
5
B6J08D9 43%
Average: 54%
REFERENCES
[1] Weihua Wang. License Plate Recognition Algorithm Based on Radial
Basis Function Neural Networks. Proceeding of International
Symposium on Intelligent Ubiquitous Computing and Education; 2009,
p. 38-41.
[2] Ning Chen, Li Xing. Research of License Plate Recognition based on
Improved BP Neural Network. Proceeding of International Conference
on Computer Application and System Modeling; 2010, p. V11-482 -
V11-485.
[3] Hsin-Fu Chen, Chang-Yun Chiang, Shih-Jui Yang, Ho, C.C. Android-
Based Patrol Robot Featuring Automatic License Plate Recognition.
18 INTERNETWORKING INDONESIA JOURNAL ROMADHON ET AL.
Proceeding of Computing, Communications and Applications
Conference; 2012, p. 117-122.
[4] Mutholib, A.; Gunawan, T.S.; Kartiwi, M. Design and implementation
of automatic number plate recognition on android platform. Proceeding
of International Conference on Computer and Communication
Engineering; 2012, p. 540-543.
[5] Zheng L., Samali B., He X., and Yang L.T. Accuracy Enhancement for
License Plate Recognition. Proceeding of 10th IEEE International
Conference on Computer and Information Technology (CIT); 2010, p.
511-516.
[6] Kasaei, S.H.M., and Kasaei S.M.M. Extractation and Recognition of the
Vehicle License Plate for Passing Under Outside Environment.
Proceeding of European Intelligence and Security Informatics
Conference; 2011, p. 234-237.
[7] Megalingam R.K., Krishna P., Somarajan P., Pillai V.A., and Hakkim
R.U.I. Extraction of License Plate Region in Automatic License Plate
Recognition. Proceeding of International Conference on Mechanical and
Electrical Technology (ICMET); 2010, p. 496-501.
[8] Feng Z., and Fang K. Research and Implementation of an Improved
License Plate Recognition Algorithm. Proceeding of 4th International
Conference on Biomedical Engineering and Informatics (BMEI); 2011,
p. 2300-2305.
[9] Vesnin E.N. and Tsarev V.A. Segmentation of Images of License Plates.
Pattern Recognition and Image Analysis 16 (2006) 1, pp. 108–110.
[10] Sirithinaphong T. and Chamnongthai K. The Recognition of Car License
Plate for Automatic Parking System. Proceeding of 5th
International
Symposium on Signal Processing and Its Application, Brisbane; 1999, p.
455-457.
[11] Beale M.H., Hagan M.T., and Demuth H.B. Neural Network ToolboxTM
– User’s Guide. The MathWorks, Inc.; 2012.
[12] Greensted A. Otsu Thresholding. Retrieved on Nov 1st, 2011, from The
Lab Book Pages: http://www.labbookpages.co.uk/software/imgProc/otsu
Threshold.html.
[13] Anonymous. Global Image Threshold Using Otsu’s Method – MATLAB.
Retrieved on Nov 1st, 2011, from MathWorks: http://www.mathworks.
com / help/toolbox/images/ref/graythresh.html.
Vol.4/No.1 B (2012) INTERNETWORKING INDONESIA JOURNAL 19
ISSN: 1942-9703 / © 2012 IIJ
Abstract—The aim of this research is to design and produce a
low cost Computer Numerical Control (CNC) machine which is
equipped with interchangeable head for many applications
especially for small-medium creative industry. The CNC machine
has good flexibility where its head can be changed manually and
easily between pen (impact-engraver) mounted and CO2 laser
mounted. The choice will depend on the application that is chosen
by the user. The software that is used is basically from open
source software and its precision is “what you see is what you
get” (WYSIWYG). Some creative applications are successfully
performed as shown in this paper, and its lower-production cost
is favorable for small-medium industry. By the year 2014, this
machine will be produced in a mass production and will be
marketable in Indonesia.
Index Terms—CNC machine, interchangeable head, pen head,
impact-engraver, CO2 laser head, creative industry applications
I. INTRODUCTION
ince the first CNC patent was published in 1995[1], many
industries start to use this machine for many applications especially for shaping process which is performed on a work
piece. This patent is also referenced by more than 100 patents
that produce similar machine but for special purposes. When it
was first developed, specialized hardware including
application specific integrated circuits were required for any
practical machines. With the performance of modern Personal
Computers (PCs), it is entirely practical to use a standard PC
to make a useful machine. In fact the majority of commercial
controllers are now simply ruggedized PC architectures which
include some form of isolated high speed digital inputs and
outputs[2]. A laser-based CNC machine from China that is sold in
Indonesia has price range from US$10,000 up to US$15,000
depending on the dimension and purposes. Nevertheless,
usually this typical CNC machine is only for one mounting
device since each application will need one head either for
cutting or welding as well as for drilling or engraving. We
studied that one company in India offered different kind of
machine for different applications such as laser cutting, laser
marking, laser engraver, laser 3D subsurface, and laser
This work is partially supported by “Hibah Bersaing KOPERTIS Wilayah
III” under Grant number 102/DRIC/IV/2012.
J. Linggarjati and R. Hedwig are with the Computer Engineering
Department, Bina Nusantara University, Jakarta, Indonesia (phone: +6221-
534-5830 ext. 2144; e-mail: [email protected] and [email protected],
respectively).
welding machine. It means that no single CNC machine can
do the several tasks simultaneously.
Understanding the needs for lower cost CNC machine as
well as multi purposes function to be used in small-medium
enterprise in Indonesia, we did a small research about these
needs, especially for creative industry. It took one year to
complete this homemade CNC machine construction with tool
path feedback control[3], which is equipped with one servo
motors and two stepper motors. Since the first design was for
writing purpose, an inked-pen head was mounted and the
machine was functioned as if it was a plotter. The drawing
result on a paperboard can be seen in figure 1(a). Having
succeeded with the first step, we then mounted an impact-
engraver head and used it for metal which was successful as
well and the result can be seen in figure 1(b). Since the
machine was developed in Computer Engineering Laboratory
at Bina Nusantara University, the main purpose of this
machine was for Printed Circuit Board (PCB) drawing,
drilling or technical drawing. In year 2012, we got a research
grant that supported the continuum of this research which
enhanced the CNC machine design by manually mounting it
with CO2 laser head. Actually CO2 laser can be used to cut
metal target[4], however the energy of the laser for this
machine is only 40Watt, and the application merely only for
paper, plastic, wood, and acrylic but it could cut, mark, and
engrave with WYSIWYG precision.
(a) (b)
Fig. 1 Engraving result on (a) paperboard target using inked-pen with
rectangle 10x10mm2 and (b) metal target, i.e. iron, using impact-engraver-pen.
II. DESIGN AND EXPERIMENTAL SETUP
The homemade CNC machine’s dimension is 800mm x
700mm x 1200mm as shown in figure 2 and the block diagram
can be seen in figure 3 with its working area’s dimension is
500mm x 450mm. It has three axes (X, Y, and Z) with two
racks to put the CPU inside. Most of previous CNC machine is
using table moving [5,6], while for this design the table is not
Manually Interchangeable Heads of Homemade
Computer Numerical Control (CNC) Machine
Jimmy Linggarjati and Rinda Hedwig
Computer Engineering Department, Bina Nusantara University, Jakarta, Indonesia
S
20 INTERNETWORKING INDONESIA JOURNAL LINGGARJATI & HEDWIG
ISSN: 1942-9703 / © 2012 IIJ
moving to the axes (X and Y), instead the two axes are driven
by a DC servo motors (TAMAgawa 200W) and a stepper
motor (VEXTA ASM 66A – 1.7A – 100W), respectively,
which in turn drive two precision ball-bearing screws, in order
to provide highly accurate movements. The end-tool, which
holds the pen, is placed along the Z axis. It is also driven by a
stepper motor (VEXTA ASD 13AA – 0.9A), equipped also
with a precision ball-bearing screw. Nevertheless, during the
CO2 laser operation the Z axis was fixed, since the focus
length of the laser is determined only by the focus-lens.
Fig. 2 Manually interchangeable cutting head of homemade CNC.
Fig. 3 Block diagram of the CNC system.
Four (4) wheels are attached to the four-legs of the CNC
machine so that we can move the position of the entire system
smoothly from one place to another. There is a 20Watt water
circulation pump for water cooling system during CO2 laser
operation, as well as vacuum filtration system with its energy
consumption as high as 750Watt. The total power
consumption for CO2 laser operation can vary up to 1500Watt.
This energy consumption can vary depends on the technology
that we choose. We can reduce the energy consumption by
changing the cathode ray tube (CRT) to liquid crystal display
(LCD) monitor. During the pen head operation, the vacuum
filtration system and the water circulation pump are not used.
The operating system that is used to control is based on Linux
Ubuntu 10.04 with open source program such as Linux
CNC[7-10] and Inkscape[11].
A comprehensive study[12] was done in order to select
appropriate laser for building a low-cost CNC machine and we
finally chose CO2 laser for its lower cost per watt along with
good beam quality as well as its wide range of average output
power ranging from a few watts up to over 60kW. The CO2
laser that is used is a continuum type with wavelength of
10.6m (power 40Watt; triggering voltage 20kV; current 16-
18mA; length 700mm; 50mm; water cooling system;
lifetime 1500-1800 hours). The setup of this system can be
seen in figure 4 where three sets of 20mm gold plated mirror
and a 18mm ZnSe lens (f = 50.88mm) are attached to the tools
of X, Y and Z axes. Figure 5 shows the Z axis which mounted
with impact-engraver pen, and “mirror-lens” for CO2 laser.
For reminder, the Z axis was fixed during “mirror-lens”
mounting and the distance between working area and the lens
are accordingly to the lens focal length, i.e. 50.88mm.
The materials that were used for this experiment are
plywood, acrylic, and cardboard, PCB with maximum 5mm,
3mm, 5mm, and 1mm thickness, respectively. The total
production cost that is needed to construct the whole system
was US$6,250 (US$1 equal to 9,600 rupiahs).
40
Wa
tt C
O2 L
ase
r H
ea
d
36mm
moves along X axes
mo
ve
s a
lon
g Y
axe
s
f = 50.88mm
working area of 500mm x 450mm
mirror
mirror mirror
Fig. 4 Setup of CO2 laser based system.
III. APPLICATIONS AND DISCUSSION
At the beginning, the whole system was dedicated for PCB
production where a piece of paper was set on the working area
and the drawing was done accurately with an inked-pen – only
used as a prove test that the precision of this machine is
enough for PCB engraving application. Later on the PCB was
drilled by the use of automated hand drill that was controlled
with computer. For further application, a CO2 laser head was
mounted in the CNC and functioned to cut cardboard,
plywood, and acrylic. The alignment for laser mounted head
was no easy at the beginning, we were using HeNe laser as
guidance along with a white thin string. The alignment took
several days before finally we fixed the position of the mirrors
and lens. Once this alignment had successfully done, the
manual interchange among pen head, spindle head, and laser
This area is interchangeable
Vol.4/No.1 B (2012) INTERNETWORKING INDONESIA JOURNAL 21
ISSN: 1942-9703 / © 2012 IIJ
head was easy. We just replaced the Z axis manually and once
we changed the head back to laser, we just had to do minor
adjustment with the guidance of laser beam on the paper.
Several results are shown in figure 6, such as the traditional
shadow puppet made of plywood, the picture engraved on
acrylic, and the 3D object made of plastic board.
Fig. 5 The head was mounted along the Z axis. From left to right: impact-
engraver-pen and mirror-lens.
(a) (b)
(c)
Fig. 6 Some results produced by the use of CO2 laser based CNC machine,
they are (a) traditional shadow puppet made of plywood, (b) cat’s picture
engraved on the acrylic, and (c) 3D object made of plastic board.
The accuracy of material produced compared to what was
displayed in the monitor was precisely the same and with
100% reproducibility as shown in figure 1a (using inked-pen)
and figure 7 (using laser-cut). The time consumption needed
for each process was depended on the size of the object and its
thickness. The thicker the object, the slower the laser will
move. The time consumption for 350mm x 250mm plywood
shadow puppet with lots of detailed, will take around 2 hours.
As for the engraved cat picture on the 105mm x 130mm
acrylic and the 3D dragon object made of plastic board will
take 3 hours and 1 hour, respectively.
When we tried to draw an electronic circuit on the PCB by
using the CO2 laser, we had to apply paint stripping
technique[13-15]. It has been studied previously that the
threshold of copper for the plasma generation in TEA CO2
laser was considerably high because of its high reflectivity
(more than 95%) at around 10.6m[16]. The process of paint
stripping as shown in figure 8 is explained as follows, the
paint on top of PCB will absorb the CO2 laser ablation since
PCB (made of copper) reflects most of the laser energy
leaving a clean stripped surface. Due to most metal’s high
reflection factor, metallic surfaces are suitable for laser
cleaning. Typically, the substrate is not mechanically or
thermally stressed or stained by the CO2 laser stripping
process. The process converts the stripped paint to particles
and vapor which are automatically removed by a vacuum
filtration system.
Fig. 7 Reproducibility and accuracy of 10x10mm2 squares using laser cut. We
got WYSIWYG result.
We tried to apply several kinds of paints in order to get the
most effective PCB that would be successfully etched. Even
though the paint stripping process finally was successful, we
failed to get a good etched-PCB during the etching process for
several times until we found that varnish is the most effective
paint to be layered on top of PCB before paint stripping
process as shown in figure 9. Nevertheless, we did not realize
that when the CO2 laser attacked the opened surface of the
PCB (the paint already removed from the surface on the first
bombardment) for the second time, the laser beam was
MIRROR
LENS
22 INTERNETWORKING INDONESIA JOURNAL LINGGARJATI & HEDWIG
ISSN: 1942-9703 / © 2012 IIJ
reflected back to the laser head and it removed the reflective
layer of the back mirror of the laser head. It happened since
we programmed the laser to do paint stripping twice for thick
path and once for thin path. Therefore, the laser head was
failed to produce 40Watt energy but produce two separated
beams instead, with a very low energy that could not even cut
a 5mm plastic board.
PCB
reflecting
absorbingthickness
paint layer
target plasma
laser beam
Fig. 8 Laser stripping process.
Fig 9 Example of the etched PCB after paint stripping process.
This result also proved that when the distances of laser
pathway closer to the laser head, the laser beam reflection
would come back to its path and attack the laser head. This is
also due to copper’s high reflectivity for laser wavelength at
10.6m. Therefore we need to put anti back reflection tool just
in front of the laser head.
This machine had been demonstrated to other Departments
(such as Architecture Department) and they showed their
interest to utilize this CNC machine for any design purposes.
In the mid of December 2012, we already offered this machine
to a small homemade CNC industry which also showed his
great interest in the machine. The further collaboration will be
discussed in order to make this machine marketable in 2014.
However, we still want to do further development research in
3D cutting or sculpturing or engraving on the woods.
IV. CONCLUSION
Building a home-made CNC machine with a low cost
budget gives advantage for small-medium industry, especially
when this CNC machine can be manually interchanged with 3
different kinds of tools according to its needs. This CNC
machine can be attached with pen head, spindle head, and
laser head for different purposes but the most useful part
happened when we attach with laser head. The open source
program also support this machine to perform with 100%
reproducibility and WYSIWYG accuracy.
REFERENCES
[1] M.R. Wright, D.E. Platts, D.B. French, G. Traicoff, M.A. Dupont, and
G.A. Head,“CNC Control System Patents,” US patent 545393, Sep 26,
1995.
[2] McKilliam G., “CNC Overview White Paper,” Industrial Automation
Series Ltd., Oct 2012.
[3] Gong Zhiming, Huang Sheng, ZengHaoand Chen Xiaoqi, “CNC Laser
Marking on Freeform Surfaces without Prior Geometrical Information,”
in Proc. 2002 IEEE/RSJ Intl. Conf. Intelligent Robots and Systems,
Lausanne - Switzerland, 2002, pp. 1862-1867.
[4] J. Wang, “An Experimental Analysis and Optimisation of the CO2 Laser
Cutting Process for Metallic Coated Sheet Steels,” Int. J. Adv. Manuf.
Technol., vol. 16, pp. 334-340, 2000.
[5] A. Hace, K. Jezernik, B. Curk, and M. Terbuc, “Robust Motion Control
of XY Table for Laser Cutting Machine,” in Proc. 24th
Annu. Conf. IEEE
Industrial Electronics Society, vol. 2, 31 Aug-4 Sept 1998, pp. 1097-
1102.
[6] Cao Shukun, Zhang Heng, ZeXiangbo, Yang Qiujuan, and Ai
Changsheng, “Software and Hardware Platform Design for Open-CNC
System,” in IEEE Int. Symposium on Knowledge Acquisition and
Modeling Workshop, 21-22 Dec 2008, pp. 139-142.
[7] Wang Changlei, Cao Shukun, Zhang Qianqian, Sui Zhiming, and Yang
Shangwei, “Design of Five-Axis CNC Based on Linux,” in Int. Conf.
Computer Science and Network Technology, vol. 2, pp. 1336-1339, 24-
26 Dec 2011
[8] Huang Hai-peng, Chi Guan-xin, and Wang Zhen-long, “Development of
a CNC System for Multi-Axis EDM Based on RT-Linux,” in World
Cong. Software Engineering, vol. 3, pp. 211-216, 19-21 May 2009.
[9] Wang Zi-niu, Li Song, and Wang Yan, “Researching of Real-Time
Version and Testing Its Performance of CNC System Based on RT-
Linux,” in Int. Conf. Networking and Digital Society, vol. 2, pp. 174-
177, 30-31 May 2009.
[10] TianrongGao, Dong Yu, DongfengYue, and Yi Hu, “Design and
Implementation of Communication Platform in CNC System,” in Int.
Conf. Mechatronics and Embedded Systems and Applications, pp. 355-
360, 15-17 July 2010.
[11] D. Spinellis, “Drawing Tools,” IEEE Software, vol. 25, issue , pp. 12-
13, May-June 2009.
[12] A.J. DeMaria, and T.V. Hennessey, Jr., “The CO2 Laser: The Workhorse
of the Laser Material Processing Industry,” SPIE Professional
Magazine, submitted for publication.
[13] R.J. Ranalli, “Laser-Based System and Method for Stripping Coatings
from Substrates,” US patent: 5662762, Sep 2, 1997.
[14] G. Schweizer, and L. Werner, “Industrial 2-kW TEA CO2 Laser for
Paint Stripping of Aircraft,” in Proc. 10th Int. Sym. Gas Flow and
Chemical Lasers, vol. 2502, pp. 57-818, March 31, 1995.
[15] Sp.G. Pantelakis, Th.B. Kermanidis, and G.N. Haidemenopoulos,
“Mechanical Behavior of 2024 Al Alloy Specimen Subjected to Paint
Stripping by Laser Radiation and Plasma Etching,” Theo. and Appl.
Fracture Mech., vol. 25, issue 2, pp. 139-146, Sept-Oct 1996.
[16] MM. Suliyanti, R. Hedwig, H. Kurniawan, and K. Kagawa, “The Role
of Sub-Target in the Transversely Excited Atmospheric Pressure CO2
Laser-Induced Shock-Wave Plasma,” Jpn. J. Appl. Phys., vol. 37, pp.
6628-6632, 1998.
Internetworking Indonesia JournalThe International Journal of ICT and Internet Development
ISSN: 1942-9703www.InternetworkingIndonesia.org
The Internetworking Indonesia Journal (IIJ) is a peer reviewed international journal devoted to the timely study of Information and Communication Technology (ICT) and Internet development. The journal seeks high-quality manuscripts on the challenges and opportunities presented by information technology and the Internet, and their impact to society. The IIJ is published electronically based on the Open Access Publication Policy, and the journal is indexed by SCOPUS and Google Scholar. Papers in the IIJ appear in one of two series:
Journal mailing address: Internetworking Indonesia Journal, PO Box 397110 MIT Station, Cambridge, MA 02139, USA.
Open Access Publication Policy: The Internetworking Indonesia Journal (IIJ) provides open access to all of its content on the prin-ciple that making research freely available to the public supports a greater global exchange of knowledge. This follows the philoso-phy of the Open Journal Systems (see the Public Knowledge Project at pkp.sfu.ca). The IIJ issues are published electronically (PDF) and there are no subscription fees. Such access is associated with increased readership and increased citation of an author’s work.
No publication fees: There is no publication fees to authors for publishing in the IIJ. This is consistent with the open access phi-losophy adopted by the IIJ.
Copyright Policy: The IIJ adopts the Creative Commons Attribution Non-Commercial No Derivatives (by-nc-nd) license. The IIJ is an academic journal dedicated to the open exchange of information. For this reason, the IIJ is freely available to individuals and insti-tutions. Copies of this journal or articles in this journal may be distributed for research or educational purposes free of charge and without permission. However, commercial use of the IIJ website or the articles contained herein is expressly prohibited without the written consent of the editor.
Authors who publish in the IIJ will release their articles to the IIJ under the Creative Commons Attribution Non-Commercial No Derivatives (by-nc-nd) license. This license allows anyone to copy and distribute the article for non-commercial purposes provided that appropriate attribution is given. For details of the rights authors grants users of their work, see the “human-readable summary” of the license, with a link to the full license:
• Creative Commons licenses: http://creativecommons.org/licenses/• Human-readable summary: http://creativecommons.org/licenses/by-nc-nd/2.5/• Full license: http://creativecommons.org/licenses/by-nc-nd/2.5/legalcode
Publication Policies
• Series A: Community Informatics & Society • Series B: Computing, Communications, Engineering & Internet Technologies
Manuscript Submission Guidelines
Language: The primary language of the Internetworking Indonesia Journal is English due to its international readership. Note that the IIJ does not accept requests for the translation of submitted manuscripts.
Format: Papers should be submitted using the IIJ Template document for MS-Word. The IIJ format is derived from the popular IEEE 2-column text format. The IIJ template can be found on the IIJ website.
Length of manuscript: Papers submitted to the IIJ should not exceed 7 pages of the IIJ 2-column format.
Originality: Papers submitted to the IIJ must not have been previously published or committed to another publisher under a copyright transfer agreement, and must not be under consideration by another journal. Papers previously published at conferences can be submitted to the IIJ, but must be revised so that it has significant differences from the conference version.
Ethics Statement: Authors wishing to submit papers to the IIJ are requested to read the IIJ Publication Ethics Statement, which can be found on the IIJ website.
Internetworking Indonesia JournalISSN: 1942-9703
www.InternetworkingIndonesia.org