ProceedingsMICEEI2010 PartA

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND
MICEEI 2010
Copyright 2010 by MICEEI 2010, Department of Electrical Engineering, Faculty of

Engineering, Hasanuddin University, Makassar
Reproduction or translation of any part of this work

beyond that permitted by MICEEI Committee,
Department of Electrical Engineering, Faculty of Engineering
Hasanuddin University Copyright
Act without the permission of the copyright owner
is unlawful. Request for permission or further information
should be addressed to MICEEI, Department of Electrical Engineering
1
Proceedings of 2nd Makassar International Conference on Electrical Engineering and Informatics (MICEEI),
27-29 October 2010, Makassar Golden Hotel (MGH), Makassar, Indonesia
ISBN: 978-602-8509-15-2, 2010 Electrical Engineering Department, Universitas Hasanuddin
Supported by IEEE Indonesia Section, IEEE APS/MTT Indonesia Joint Chapter, and IEEE Communication Society
(Comsoc) Indonesia Chapter
MICEEI 2010
Foreword
Hasanuddin University, Makassar, INDONESIA and the IEEE Indonesian Chapter have agreed to
jointly sponsor the Makassar International Conference on Electrical Engineering and Informatics
(MICEEI). The First MICEEI was held on November 13-14, 2008, and it was decided to hold the
event bi-annually in the even numbered years in Makassar. This MICEEI being held on October 2728, 2010 is the second one. The other international event jointly sponsored by Hasanuddin
University and the IEEE Indonesian Chapter was the International Workshop on Modern Research
Methods in Electrical Engineering (IWoRMEE). It is held also bi-annually in the odd numbered
years, either in Makassar or elsewhere. The First IWorMEE was hosted by the Hasanuddin
University Department of Electrical Engineering in Makassar last year.
Informatics is a relatively new study program in the Indonesian academia, but it is quite a fast
growing discipline. We decided to integrate this new discipline into the Department of Electrical
Engineering and accommodated research papers in this area in this conference not without
reasons. Several sub-disciplines of Informatics such as software engineering, algorithm analysis and
development, pattern recognition, etc., are very familiar to the Electrical Engineering world,
especially the Computer Engineering specialization. The papers presented in the Second MICEEI.
2010 cover a broad spectrum of both Electrical Engineering discipline (including the Electrical
Power and Energy Engineering, Telecommunication, Control, Electronic and Computer
Engineering) and Informatics. All presented papers as well as the text version of keynote speeches
are printed in the proceedings.
I wish to thank all members of the Organizing Committee, especially the International Advisory and
Program Committee, session chairpersons and authors of all papers, all of whom have contributed
in large measure to the success of this 2nd MICEEI 2010. We all indebted to the distinguished
keynote speakers and the Hasanuddin University's Vice Rector for Academic Affairs and the Dean
of Engineering who have agreed to address the plenary session in the first day of the conference. I
also wish to express my sincere thanks to our Secretary General, Dr. Elyas Palantei, the Department
Chairman and Secretary, Dr. Zahir Zainuddin and Ms. Zaenab Muslimin respectively, for their
enormous work in carrying out many detailed tasks to make this 2nd MICEEI possible. Finally, I
would like to sincerely thank all of our friends and partners in the electrical engineering industrial
world and academia who have contributed greatly to this event.
R. S. Sadjad
Conference Chairperson
2
MICEEI 2010
Organizing Committee
Supervisory Committee:
Dr. Syahrul Yasin Limpo, S.H, M.Si, M.H (Governor of Sulawesi Selatan)
Ir. Ilham Arief Sirajuddin, MM (Mayor of Makassar)
Prof.Dr.dr. Idrus Paturusi, SPBO (Rector of Universitas Hasanuddin)
Dr. Ir. Wahyu H. Piarah, MSME (Dean of Engineering Faculty, UNHAS)
General Chair:
Rhiza S. Sadjad, Universitas Hasanuddin(UNHAS), Indonesia
Co-Chair:
Zahir Zainuddin, Head of Electrical Engineering Department, UNHAS, Indonesia
International Advisory Committee:

Nadjamuddin Harun, Universitas Hasanuddin (UNHAS), Indonesia (Chair)
Muhammad Tola, Universitas Hasanuddin (UNHAS), Indonesia (Co-Chair)
Muhammad Arief, Universitas Hasanuddin (UNHAS), Indonesia
David Victor Thiel, CWMA, Griffith University, Australia (IEEE Senior Member)
Mary-Anne Williams, University of Technology Sydney (UTS), Australia
Kazi M. Ahmed, Asian Institute of Technology (AIT), Thailand (IEEE Senior Member)
Tapio J. Erke, Asian Institute of Technology (AIT), Thailand
Arnold Djiwatampu, (IEEE Indonesia Section Chair)
Burhanuddin Yeop Majlis, Universitas Kebangsaan Malaysia (UKM)(IEEE Senior
Member)
Eko Tjipto Raharjo, Universitas Indonesia(IEEE Indonesia MTT/AP-S Joint Chapter)
Takashi Hiyama, Kumamoto University, Japan (IEEE Senior Member)
Josaphat Tetuko Sri Sumantyo, Chiba University, Japan (IEEE Senior Member)
Manfred Glesner, Darmstadt Univ. of Technology, Germany (IEEE Fellow)
Adang Suwandi, Institut Teknologi Bandung (ITB), Indonesia
Tumiran, Universitas Gadjah Mada(UGM), Indonesia
Dadang Achmad Suryamihardja, Universitas Hasanuddin (UNHAS), Indonesia
International Program Committee:

Salama Mandjang, Universitas Hasanuddin (UNHAS), Indonesia (Chair)
Zulfajri B. Hasanuddin, Universitas Hasanuddin (UNHAS), Indonesia (Co-Chair)
Mazlina Esa, UTM, Malaysia (IEEE Malaysia Section AP/MTT/EMC Joint Chapter)
3
MICEEI 2010
R.M.A.P Rajatheva, Asian Institute of Technology (AIT), Thailand

T.Sanguankotchakorn, Asian Institute of Technology (AIT), Thailand
Tsuyoshi Usagawa, Kumamoto University, Japan
Hassan Bevrani, University of Kurdistan, Iran (IEEE Senior Member)
Budiono Mismail, Universitas Brawidjaya (UNIBRAW), Indonesia
Burhanuddin Mohd Aboobaider, UTEM, Malaysia
Pekik A Dahono, (IEEE Indonesia Joint Chapter of Education, Electron Devices, Power
Electronics & Signal Processing Societies)
Tri Adiono, Institut Teknologi Bandung (ITB), Indonesia
Eniman Syamsuddin, Bandung Institute of Technology (ITB), Indonesia
Wahidin Wahab, Universitas Indonesia, Indonesia
Edi Lukito, Universitas Gadjah Mada (UGM), Indonesia
Anton Satria Prabuwono, Universitas Kebangsaan Malaysia (UKM), Malaysia
Arief Hamdani Gunawan,PT.Telekomunikasi Indonesia TbK.,(IEEE Indonesia Secretary)
Kuncoro Wastuwibowo, IEEE Comsoc Indonesia Chapter
Rini Nur Hasanah, Universitas Brawidjaya (UNIBRAW), Indonesia
Armin Lawi, Universitas Hasanuddin (UNHAS), Indonesia
Syafaruddin, Kumamoto University, Japan
Andreas Vogel, Visiting Lecturer of Universitas Hasanuddin, Germany
Faizal Arya Samman, Darmstadt Univ. of Technology, Germany
Local Organizing Committee:

Secretariat:
Elyas Palantei (IEEE Member)(Chair)
Muhammad Anshar
Fitrianti Mayasari
Merna Baharuddin
Ikhlas Kitta
Andi Ejah Umraeni Salam
Elly Warni
Adi Wahyudi
Mustakim
Electrical Engineering Student Association-HME UNHAS (Member)
Treasure Committee:
Zaenab Muslimin, (Chair)
Novy Nurrahmillah AM
Ansar Suyuti
Sri Mawar Said
4
MICEEI 2010
Publication Committee:
Tahir Ali (Chair)
Muchtar Saleh
Rachmat Santosa
Nien Khamsawarni Nauman
Herman Rombe
Syafruddin Syarif
Electrical Engineering Student Association-HME UNHAS
Local Arrangement Committee:
Gassing (Chair)
Subaer Kanata
Christoforus Yohannes
Andani Achmad
Indrajaya Mansur
A Toyib Rahardjo
Sri Wahyuni Awaluddin (Politeknik Kesehatan Makassar)
Electrical Engineering Student Association-HME UNHAS
5
MICEEI 2010
THE 2nd MAKASSAR INTERNATIONAL CONFERENCE ON ELECTRICAL

ENGINEERING AND INFORMATICS (MICEEI 2010)
OCTOBER 27, WEDNESDAY, MGH GOLDEN BALL ROOM

07.30-08.30
Conference Registration
Makassar Golden Hotel Hall (2nd floor)
08.30- 09.45 Opening Ceremony
Welcome Address-1: Dr. Ir. Zahir Zainuddin, M.Sc (Chairman of Electrical
Engineering Department)
Welcome Address-2: Dr-Ing. Ir. Wahyu Piarah, MSME (Dean of Engineering
Faculty)
Welcome Address-3: Prof.Dr.Dadang Ahmad Suriamiharja, M.Eng (Vice
Rector for Academic Affairs)
09.45-10.00 Coffee Break
Keynote Speaker 1: Prof. Takashi Hiyama, Kumamoto University, Japan
10.00-10.40 Moderator: Dr. Syafaruddin, M.Eng
Keynote Speaker 2: Prof. Dadang Gunawan, University of Indonesia
10.40-11.10 Moderator: Mr. Indrabayu Amirullah, ST., M.Bus.Sys, MT
Keynote Speaker 3: Prof. Mary-Anne Williams, University of Technology
Sydney, Australia
11.10-11.50 Moderator: Mr. Muhammad Anshar, ST., MSc. Research
12.00-13.20 Lunch Time
Keynote Speaker 4: Prof. Eko Tjipto Rahardjo, University of Indonesia
13.20-14.00 Moderator: Dr. Ir. Zulfajri B Hasanuddin, M.Eng
Keynote Speaker 5: Dr. Linus Liu, CST Malaysia
14.00-14.40 Moderator: Merna Baharuddin, ST., M.Tel.Eng, PhD
Keynote Speaker 6: Ir. Syamsul Aliah, MSEE, LEN Industry, Indonesia
14.40-15.20 Moderator: Dr. Ir. Andani Achmad, MT
Invited Speaker 7: Dr. Syafaruddin, M.Eng, Kumamoto University, Japan and
Electrical Engineering Department, Hasanuddin University, Indonesia
15.20-16.00 Moderator: Novy Nur Rahmillah Ayu Mokobombang, ST., MsTM
16.00-16.20 Coffee Break
Technical Tutorial 1: Circuit Design and Fabrication, SolidBase Technology
16.20-17.50 Technical Tutorial 2: Antenna Modeling and Fabrication, CST Malaysia
6
Page
13
14
16
17
19
20
21
MICEEI 2010
OCTOBER 28, THURSDAY

07.30-08.00
Conference Registration
Technical Session 1
TCE1:Telecommunication Engineering (1)
Chairs: Dr.Ir. Zulfajri B. Hasanuddin, Dr. Yuli Fitri Zulkifli, MSc
Venue: Golden Ball Room/Pakarena Room
08.00-08.20
TCE1-1
08.20-08.40
TCE1-3
08.40-09.00
TCE1-4
09.00-09.20
TCE1-5
09.20-09.40
09.40-10.10
10.10-10.30
10.30-10.50
10.50-11.10
11.10-11.30
11.30-11.50
Page
Multiband Microstrip Antenna for RFID Reader
Fitri Yuli Zulkifli, Arifin Djauhari, Dandy Farhan Nugraha,
25
Eko Tjipto Rahardjo
Printed Microstrip Antennas (MSAs) Suitable for MS and BTS 29
WiMAX Applications
E. Palantei, Rahmatullah M, A.E. Putra, and A. Achmad
Performance of Low Power, Short Distance and Narrow 34
Bandwidth Remote Mobile Sensor Network at Various
Wireless Environments
E. Palantei, J.T. Pakadang, H. Sampealang, and N.N.R.A.
Mokobombang
35
Remote Patient Monitoring Network Cluster
Santosh Shedge and Anup Kale

Design of Microstrip Branch Line Coupler with T-Junction for
Mobile WiMAX at Frequency of 2,3 GHz
TCE1-6
Gunawan Wibisono and Indah Riyanti
Coffee Break
Microstrip Diplexer Design for Mobile WiMax at 2,3 GHz dan
2,5GHz Using Parallel-coupled, Half-wave Resonator
TCE1-7 Bandpass Filter
Gunawan Wibisono and Sakti Pratama
A Short Text Compression Scheme based on Arithmetic
TCE1-8 Coding
Iwan Handoyo Putro, Petrus Santoso, and Maya Basoeki
A Query by Singing/Humming System Using Autocorrelation
TCE1-9 Function and Hidden Markov Model
Indrabayu, Abdul Aziz, and Dadang Gunawan
Handover Schemes for Real-time Session over Broadband
TCE1-10 Wireless Networks
Muhammad Niswar and Elyas Palantei
Subarray of Elliptical Microstrip Antenna for CircularlyTCE1-11 polarized Synthetic Aperture Radar
Merna Baharuddin and Josaphat Tetuko Sri Sumantyo
7
36
37
38
39
40
41
MICEEI 2010
CEIE1:Control, Electronics and Instrumentation Engineering (1)

Chairs: Prof. Dr. Ir. Nadjamuddin Harun, MS; Dr. Ir. Andani Achmad, MT.
Venue: Bira Room
08.00-08.20
CEIE1-3
08.20-08.40
CEIE1-5
08.40-09.00
09.00-09.20
09.20-09.40
09.40-10.10
CEIE1-6
Dynamixel Motor-based Actuator For Mobile Robot Design

Muh. Anshar and Ambo Aco
Matlab based Simulation of Measuring Height Level,
Moisture, and Temperature in Miniature Industry
Andani Achmad, Rhiza S Sadjad, Dadang A Suriamiharja,
and Joshapat Tetuko Sri Sumantyo
Product Quality Inspection Based on Machine Vision System
Habibullah Akbar, Anton Satria Prabuwono, Hasniaty A.,
Zulkifli Tahir, and Taufik
Multi Agent System Modeling Using JADEX

Irawan, Atthariq, and Surya Sumpeno
Preliminary Study of Developing Flood Early Warning System
CEIE1-8 Zahir Zainuddin
Coffee Break
Page
48
49
50
51
CEIE1-7
55
CIE1: Computer and Informatics Engineering (1)

Chairs: Dr. Armin Lawi, M.Eng
Venue: Suki-52 Room
08.00-08.20
08.20-08.40
08.40-09.00
09.00-09.20
09.20-09.40
09.40-10.10
10.10-10.30
Page
Optimization of Pre-determined Object Ordering
61
Amil A. Ilham and Kazuaki Murakami
A Decision Support System for De Joglo Restaurant Using 62
Frequent Pattern Tree Data Mining
CIE1-3
Andreas Handojo, Gregorius Satia Budhi, and Nadia Andreani
Dwiyono
Software Framework for Intelligent Micro-Factory
63
CIE1-4
Anup Kale
Color Image Reconstruction From Compressive Sensing Using 64
CIE1-6 Iteratively Reweighted Least Squares- lp-M
Endra
Effect of Spectral/Spatial Transformation on Remote Sensing
65
CIE1-7 Image for NDVI-Based Drought Detection Analysis
F. Akbar, N. Suryana and B. Hussin
Coffee Break
3D Fabric Feature Extraction and Defect Classification Using
66
Low-Cost USB Camera
CIE1-8
Fikri Akbar, Habibullah Akbar, Nanna Suryana and
Muhammad Husni
CIE1-2
8
10.30-10.50
CIE1-9
MICEEI 2010
Book Loan Recommendation System for Petra Christian

University Library using Prefix Span and Generalized
Sequential Pattern Algorithm
Gregorius Satia Budhi, Andreas Handojo, and Stefano Gerry
Sutrisno
71
PE1:Power Engineering (1)

Chairs: Prof. Dr. Ir. Salama Mandjang, MT ; Dr. Syafaruddin, M.Eng
Venue: Kayangan Room
08.00-08.20
08.20-08.40
08.40-09.00
09.00-09.20
09.20-09.40
09.40-10.10
10.10-10.30
10.30-10.50
10.50-11.10
11.10-11.30
11.3011.50
Transient Stability Analysis of Java-Bali 500KV

Page
Inteconnection System using Modified Equal Area Criterion
PE1-1 Combined with Time-Domain
75
Adi Soeprijanto, Muhammad Abdillah
Optimization Techniques of Hybrid Renewable Energy 76
System: A Review
PE1-3
Dilruba Siddiqi, Anton Satria Prabuwono, Hasniaty A.,
Zulkifli Tahir, Taufik
Optimal Placement of TCSC for Optimizing System
77
PE1-4 Loadability by PSO
I Made Wartana and Sasidharan Sreedharan
An Approach for Real Power Generation Scheduling to
78
Improve Transient Stability in the Java-Bali 500 kV
PE1-5 Interconnection Power System
Indar Chaerah Gunadin, Adi Soeprijanto, and Ontoseno
Penangsang
Maximizing Available Transfer Capability of Existing Systems 79
PE1-6 using HVDC System
Rachmat Santosa
Coffee Break
Variable Speed Power Generation for Remote Area Power
80
PE1-7 Supply
Tajuddin Waris and C.V Nayar
81
PT Badak NGL Black Out Start Up System Improvement
PE1-8
Nasrul Syahruddin and Rizqy Fajar Arifianto
82
A Simplified Model of Light Illumination Meter
PE1-9
Muh. Anshar and Usman
The Influence of Infeed Factors on Setting Distance Relays in 83b
150 KV Power Transmission Line of MINAHASA GORONTALO System Interconnection
PE1-10
Salama Manjang and Yohanis S. Rompon
9
MICEEI 2010
Technical Session 2
TCE2:Telecommunication Engineering (2)
Chairs: Dr. Ir. Zulfajri B Hasanuddin, M.Eng; Merna Baharuddin, MTel.Eng, PhD
Venue: Golden Ball Room/Pakarena Room
13.0013.20
TCE2-1
13.2013.40
TCE2-2
13.4014.00
TCE2-3
14.0014.20
TCE2-4
14.2014.40
TCE2-5
14.4015.00
TCE2-7
15.0015.20
15.2015.50
15.5016.10
16.1016.30
16.3016.50
16.5017.10
17.1017.30
TCE2-9
Elliptical Annular Ring Microstrip Antenna for Circularly

Page
Polarized Synthetic Aperture Radar
84
Merna Baharuddin and Josaphat Tetuko Sri Sumantyo
Audio Quality Improvement of FM Radio Receiver through the 85
Noise Level Reduction Based SDR
S. Syarif, E. Palantei, A. Thayyib, and H. Thomassawa.
Statistical Analysis of an Indoor Radio Propagation Channel
86
Path Loss Measurement
Abdullah Zainuddin and Syafaruddin CH
Wideband 3 Array Double Slot Bowtie Microstrip Antenna for
87
Wireless Communication
Bualkar Abdullah, Yono H Pramono, and Eddy Yahya
Study of Adaptive Radio Resource Allocation Using Cross
88
Layer Design in OFDMA Systems
Arfianto Fahmi and Dadang Gunawan
Multiresolution Analysis Using Quaternion Matrix-Valued
89
Wavelets
Mawardi Bahri
93
Compressing Digital Image using Quantization Method
Armin Lawi and Hendra
Coffee Break
Time Reduction In Encrypting and Decrypting With RSA
TCE2-10 Algorithm Using Distributed Computing
Jhonsong Hoya and Armin Lawi
EDFA Simulation on the Fiber Optic Communication System
TCE2-11 A. Achmad, N.N.R.A Mokobombang, R.R.P. Muhlis, J.J.N.
Ferdiansyah, and Z. Muslimin
Digital Image Steganography Under Cipher Block Chaining
TCE2-12 Operation on AWGN Channel
Ansar Suyuti, Sitti Wetenriajeng Sidehabi
MoPING A Cellular Network Data Service Measurement
Scheme
TCE2-13
Ady Wahyudi Paundu, Zahir Zainuddin, Zulfajri Basri
Hasanuddin, Zaenab Muslimin
TCE2-14
Analysis of Sea Surface Temperature using MODIS Data

Dodi Sudiana, Rifqi Annas
10
97
98
99
103
104
MICEEI 2010
CIE2a: Computer and Informatics Engineering (2a)

Chairs: Dr.Ir. Zahir Zainuddin, MSc; Dr. Armin Lawi, M.Eng
Venue: Suki-52 Room
13.20-13.40
13.40-14.00
14.00-14.20
14.20-14.40
15.00-15.30
The Design and Development of Facebook Based Agenda

Application With BlackBerry Integration
Ibnu Gunawan, Yulia Kendengis, Deny Hendrata
Performance Analysis of Dijkstra, A* and Ant Algorithm for
CIE2a-2 Finding Optimal Path Case Study: Surabaya City Map
Leo Willyanto Santoso, Alexander Setiawan, Andre K. Prajogo
A Novel of Hybrid Maintenance Management Models for
Industrial Applications
CIE2a-4
Zulkifli Tahir, M.A. Burhanuddin, and Anton Satria
Prabuwono
The Application Maximum Likelihood Estimation for
Unbalanced Data to Determine of Clean Water Debits from
CIE2a-5
PDAM Makassar
Georgina M. Tinungki
Coffee Break
Page
CIE2a-1
112
113
114
115
CIE2b: Computer and Informatics Engineering (2b)

Moderator : Dr. Yuli Fitri Zulkifli, MSc
Venue: Bira Room
13.20-13.40
CIE2b-3
13.40-14.00
CIE2b-5
The Ramsey Numbers for Odd Complete Bipartite Graphs

Hasmawati
Survival Analysis of High Dimension Microarray DNA Gene
Data
Arniati J. Kalatasik and Armin Lawi
Classifying Stadium Class of Serviks Cancer using Mamdani
Fuzzy Method
Greby Febriastusi and Armin Lawi
Measurement Quality of Service In Virtual Server
Ida Nurhaida, N.N.R.A. Mokobombang and E. Palantei
14.00-14.20
CIE2b-6
14.20-14.40
CIE2b-7
14.40-15.00
Coffee Break
Page
125
129
130
131
132
11
MICEEI 2010
He received his B. E., M. S. and Ph. D. degree all in electrical

engineering from Kyoto University in 1969, 1971 and 1980, respectively. He
joined to Kumamoto University in 1971 and has been a Professor in
Department of Computer Science and Electrical Engineering, Kumamoto
University, Japan since 1989. His current research interests include the
intelligent systems applications to power system operation, control and
management and also the applications of renewable energy power sources.
He is a senior member of IEEE, a member of IEE of Japan, and Japan Solar
Energy Society. He is also the former Dean of the Graduate School of
Science and Technology, Kumamoto University, Japan.
Speech Outline: Prof. Takashi Hiyama [IEEE Member (1986), IEEE Senior Member (1993)]
Intelligent System Applications to Power System

In the last twenty years, intelligent systems applications have received increasing attentions in
various areas of power system operation, control, and management. A quite big number of research
papers indicate the applicability of intelligent systems to power systems. While many of these
systems are still under investigation, however, there already exist a number of practical
implementations of intelligent systems in Japanese utilities. In conventional schemes, power
system operation, control, and management are based on human experience and mathematical
models to find solutions, however, power systems have many uncertainties in practice. Namely,
those mathematical models provide only for specific situations of the power systems under
respective assumptions. With these assumptions, the solutions of problems existing in power
systems are not trivial. Therefore, there exist some limitations for the mathematical model based
schemes. In order to overcome these limitations, applications of intelligent technologies such as
knowledge based expert systems, fuzzy systems, artificial neural networks, genetic algorithms, and
the other intelligent technologies have been investigated in wider areas of power systems for
reliable and high quality power supply at low cost. In addition, recent research works indicate that
more emphasis has been put on the combined usage of intelligent technologies for further
improvement of the operation, control and management of power systems.
In my keynote speech, after a brief introduction of the state-of-art intelligent techniques
utilized in Japanese Utilities, current research activities in our Electric Power Systems Laboratory
will be introduced including several messages to younger researchers and graduate students.
12
MICEEI 2010
Prof. Dadang Gunawan

Professor Dadang Gunawan holds a Doctor of Philosophy (PhD) in Electrical & Electronics
Engineering from University of Tasmania, HobartAustralia, graduated in 1995. He holds a Master
of Engineering (M.Eng.) in Electrical Engineering from KEIO University, TokyoJapan, graduated
in 1989. And, he also holds Insinyur Degree (Ir.) from University of Indonesia, graduated in 1983.
Since 1983 till now, he served as a Lecturer in Electrical Engineering Department University of
Indonesia. His interested areas include Wireless Communications and Multimedia Communication,
Signal Processing and Wavelet Transform, Image and Video Compression, Coding and
Transmission, Audio and Speech Coding, Magnetic Resonance Imaging, Object Classification,
Signal Recovery, Noise Cancellation.
From 2008-now, professor Dadang served as Head of Telecommunication Laboratory Electrical
Engineering Department University of Indonesia. From 2004-07, he served as the Vice Director for
Academic Venture Institution Faculty of Engineering University of Indonesia. From 2000-04, he
served as the Vice Dean for General administration & Finance Faculty of Engineering University of
Indonesia. From 1998-2000, he served as Secretary of Postgraduate Electrical Engineering Study
Program University of Indonesia. From 1996-98, he served as the Assistant for Academic of Head
Electrical Engineering Department University of Indonesia.
Since 1998 till now, he served as Reviewer of Journal Technology Faculty of Engineering
University of Indonesia and Quality of Research Faculty of Engineering University of Indonesia.
He also served temporarily as reviewer for Dissertation of Doctor in Bandung Institute of
Technology, and some overseas University. He also served temporarily as reviewer for National &
International Seminar/Conferences.
Prof Dadang has published more than one hundred papers in National & International Journal,
Proceeding of Seminar/Conferences as either first Author or second and third Authors.
Professor Dadangs awards include the award of services Satyalencana Karya Satya XX; Multi
Years Research Grand Award of University Research for Graduate Education (URGE) Higher
Education of Directorate General Ministry of Education & Culture Republic Indonesia; Research
Grand from TORAY Indonesia. The Best Researcher 1995 Awards, for research activity in Science
and Engineering, University of Indonesia; The Best of Selected Publish in International Journal
1995 Awards, University of Indonesia; Finalist of the Young Researcher 1995/96 Awards, the
Indonesia Institute of Science & Technology (LIPI).
Professor Dadang holds as a Senior Member of the IEEE.
13
MICEEI 2010
Speech Outline:
The Application of Wavelet Transform for an Image Compression
Nowadays, video applications such as digital laser disc, electronic camera, videophone and video
conferencing systems, image and interactive video tools on personal computers and workstations,
program delivery using cable and satellite, and high-definition television (HDTV) are available for
visual communications. Many of these applications, however, require the use of data compression
because visual signals require a large communication bandwidth for transmission and a large a
mounts of computer memory for storage. In order to make the handling of visual signals cost
effective it is important that their bandwidth be compressed as much as possible. Fortunately, visual
signals contain a large amount of statistically and psychovisually redundant information. By
removing this unnecessary information, the amount of data necessary to adequately represent an
image can be reduced.
We present an overview on wavelet transform and its applications, the results of our previous work
and ongoing research for image compression, and some considerations for application wavelet
transform in the future.
14
MICEEI 2010
Professor Mary-Anne Williams

Mary-Anne Williams is a Professor at the University of Technology, Sydney, Director of the
Innovation and Enterprise Research Laboratory, and Associate Dean (Research and Development)
in the Faculty of Engineering and Information Technology. Mary-Anne was a Visiting Professor
at the AI Lab at Stanford University in 2009-2010 and she currently holds a Guest Professorship
at University of Science and Technology China (USTC), a prestigious research university which
was established by the Chinese Academy of Science and one of the top universities in China.
Mary-Anne is a leader in Artificial Intelligence and Knowledge Representation research. She has
over a hundred publications, some in the most prestigious and highest impact research outlets
including in the leading Journal of Artificial Intelligence, and prestigious conferences such as the
International Joint Conference in Artificial Intelligence. Mary-Anne works closely with industry
partners including IBM and Visualrisk to ensure her research has relevance and impact. She has
developed several new models for knowledge management in complex dynamic environments in
a wide range of application areas from robotics to risk management. Mary-Anne is the Team
Leader of the highly successful robot soccer team UTS Unleashed! (www.unleashed.it.uts.edu.au)
and the joint UTS-USTC robot soccer team (www.wrighteagleunleashed.org). Mary-Anne is Past
President and a Director of Knowledge Representation Inc (KR Inc) a US based International
Scientific Research Foundation. Mary-Anne serves on the Artificial Intelligence Journal Editorial
Board, and the American Artificial Intelligence Association/MIT Press Editorial Board among
others, and since 2009 she has been a Fellow at the Stanford Centre for Computers and Law.
Speech Outline: Designing and Developing Robots with People Skills
Robots are coming of age and becoming more and more widespread in society. You can find them
assisting doctors in operating theatres, plugging oil leaks on the bottom of the ocean, interacting
with people at home and at work. Increasingly robots are being developed to interact with
ordinary people in wider society in open environments, not just scientists in constrained
environments. As a result important questions about their design, capabilities and legal status are
being raised. Already robots can perceive things beyond human capability, they can collaborate
with each other in ways that are not possible for humans, e.g. they can share sensor data directly,
imagine if you could see exactly what I see, smell, hear and feel. The age of robots will transform
the way we live, the way we interact, communicate and collaborate with each other. This talk will
explore some of the major challenges we will face in designing and developing robots with people
skills - robots with an awareness of people and an ability to anticipate people's actions and
intentions.
15
MICEEI 2010
Prof. Eko Tjipto Rahardjo

Eko Tjipto Rahardjo, PhD. was born at Pati, Indonesia, on April 22, 1958. He received the Ir.
Degree from the Universitas Indonesia, Jakarta in 1981, the MS degree from the University of
Hawaii at Manoa, Honolulu in 1987, and the PhD. degree from the Saitama University, Urawa in
1996, all in electrical engineering. He joined to the Department of Electrical Engineering
Universitas Indonesia since 1982 as a teaching assistant. Since 2005 he has been appointed as
Professor in electrical engineering.
He has been Head of Electrical Engineering Department, Universitas Indonesia (2004 2008). He
was Executive Director of the Quality Undergraduate Education (QUE) in the Department of
Electrical Engineering Universitas Indonesia (1999 2004), and was Head of Telecommunication
Laboratory Universitas Indonesia (1997 2004). Since 2003, he has been Director of the Center for
Information and Communication Engineering Research (CICER), Universitas Indonesia as well as
Antenna propagation and Microwave Research Group (AMRG) leader.
His research interests include antenna engineering, wave propagation, microwave circuits and
communication system and regulations. He has been published and presented more than a hundred
research papers both national and international journals and symposiums. He has been recipient of
the Indonesian government scholarship through MUCIA (1984 1987); the Hitachi Scholarship
(1992 1996); and the Young Researcher's Award from the Universitas Indonesia (1996).
Dr. Rahardjo is members of IEEE Antenna and Propagation Society (AP-S), IEEE Microwave
Theory and Technique Society (MTT-S), IEICE Japan and IECI Indonesia. He is now serving as
chairman of IEEE Joint Chapter MTT-S/AP-S, IEEE Indonesia Section.
Speech Outline:
A Survey on Reconfigurable Antennas
Recently much attention has been made for the development of reconfigurable antenna. Many
wireless and radar systems use separate and different antennas; however there is a trend to
incorporate several systems by using a single antenna. A reconfigurable antenna is an antenna that
should change their radiation characteristics such as frequency, bandwidth, polarization and
radiation pattern independently to accomplish their operating requirements.
16
MICEEI 2010
In this talk we survey the emerging area of reconfigurable antennas from fundamental concepts to
advanced techniques and examples that offer important new capabilities for future applications. As
we aware antennas are necessary and critical components of communication and radar systems, but
sometimes their inability to adjust to new operating condition can limit system performance.
Therefore by making antennas reconfigurable, their behavior can be adapted with changing system
requirements or environmental conditions so that can eliminate these restrictions and provide
additional levels of functionality for any system.
Several examples on reconfigurable antennas for both single radiator and array will be described to
clarify the concepts.
17
MICEEI 2010
Dr. Linus Liu

Linus Lau obtained First Class honor BEng in Mechatronic Engineering, Master of Research in
Mechanical Engineering and PhD in Microwave Engineering from University of Manchester
Institude of Science and Technology (UMIST) in between 1995 2003. He returned to Malaysia
joining Intel Penang as Senior Test module Development Engineer (2004-2005). He was working in
the area of Test module development for Intel Centrino WIFI Transceiver chipset; measurement
verification study on test related challengers, i.e. Signal Integrity, EMC, thermal and mechanical
test. He is now attached to Computer Simulation Technology (CST), the market leader in 3D EM
Time domain simulator as the main technical consultant for South East Asia region. Hes
responsible for both business development and Technical support for customers in this region and
currently base in Kuala Lumpur, the capital of Malaysia.
Speech Outline:
Computation Electromagnetic for Antenna applications

Modern electronic systems are showing a constant growth in complexity combining with faster
signal transmission rates and higher operating frequencies, the same challenge goes to the Antenna
elements. Having several revolutions in prototyping a design often becomes too expensive in terms
of dollar and time cost. Consequently, a fill 3D EM simulation tool which helps providing better
insight and understanding of the Electro-Magnetics behavior of the system becomes an industrial
standard in shortening the product design cycle. An overview will be given on the possible
simulation methods, discussion of the important points to consider during a simulation and
computational electromagnetic implementation on several Antenna applications.
18
MICEEI 2010
Ir. Syamsul Aliah, MSEE

He received his BE (Ir) from Hasanuddin University in Electronic and Telecommunication
Engineering, and MSEE from University of Alabama in Huntsville in the field of Computer
Engineering. He joined Indonesia National Institute of Science (LIPI) in 1988 as researcher, and
currently working as Defense System Advisor for PT. LEN Industri (Persero). His Current interests
include Defense System and Learning System. His current projects are Combat Management System
for Parchim and Van Speijk Class, Tactical Radio Communication, and Curriculum Development for
Civil Air Transportation School. He also founder of education institutes and electronics companies
such as Cendekia Muda Elementary and Junior High School, Bandung Empowering System (learning
content development company), and Xenon Indonesia (software engineering company).
Speech Outline:
Electrical and Computer Engineering Based Company in Indonesia
In the era of globalization which supported by the ease of communication and interaction today, many
large companies based engineering are facing the difficulties to survive. Some companies were
replaced by small companies which are more flexible and focus. This actual phenomena has and will
influence also in Indonesia, such as decreased activity of several engineering companies in the field of
technology, and shifting their competency. In line with those trends, a number of interesting question
are required to be addressed such as what kind of the electronic-based engineering company in
Indonesia may sustain in the global competition ?", This critical question is important to be explored
and discussed amongst the professional, practiciant, academician and technocrat working in the related
area of study. One of the main outcomes generated will be related to the availability of human
resources. In conjunction with the needs of electronics-based industries in the future the educational
institutions can take a part as the main player. This responsibility should be maintained and developed.
This talk will cover all the above issues.
19
MICEEI 2010
20
MICEEI 2010
21
MICEEI 2010
22
MICEEI 2010
23
MICEEI 2010
24
MICEEI 2010
25
MICEEI 2010
The Application of Wavelet Transform for an Image Compression

Dadang Gunawan
Wireless and Signal Processing (WASP) Research Group
Electrical Engineering Department University of Indonesia
guna@ee.ui.ac.id
I. Introduction
The rapid growth of digital imaging applications, such as image and interactive video tools
on personal computers and workstations, desktop publishing, multimedia, teleconferencing, highdefinition television (HDTV), digital laser disc, electronic camera, program delivery using cable
and satellite, are available for visual communications. Many of these applications, however, require
the use of data compression. In order to make the handling of digital imaging signals cost effective
it is important that their bandwidth be compressed as much as possible. Fortunately, imaging
signals contain a large amount of statistically and psychovisually redundant information. By
removing this unnecessary information, the amount of data necessary to adequately represent an
image can be reduced.
There are numerous ways to achieve compression in image compression techniques.
Predictive coding and transform coding are widely used for image compression techniques.
Transform coding especially using the Discrete Cosine Transform (DCT) is an international
standard and one of the most powerful compression techniques [1]. Data or bandwidth compression
is achieved mainly by discarding weak frequency components whose amplitudes are below a certain
visible threshold and by quantizing the remaining non-zero coefficients with as few bits as possible.
Psychovisual image compression techniques involve the discarding of information that the human
eye is likely to see or to perceive. The aim here is to find an image compression algorithm which is
optimized to match the human visual system (HVS). The technique will be adaptively tuned to the
local characteristics of the image, such as the image activity or business and luminance or
brightness.
This paper describes the results of previous worked for psychovisual image compression by
using wavelet transform [2] and highlight for the future image compression i.e., medical imaging.
II. Wavelet Decomposition

In spite of being a relatively new area of mathematics, wavelet theory has actually been
known since beginning of this century [3]. It was, however, only recently developed as a unified
framework in the mid-eighties by a group of French geophysicists, theoretical physicists and
mathematicians namely Morlet, Grossman, Meyer and Lemarie; and in the late eighties by a French
signal processing group, namely Daubechies and Mallat. Wavelet theory represents signals by
breaking them down into scaled and translated wavelets [4]. The breaking down process is termed
as wavelet transform or wavelet decomposition. Wavelet reconstruction or the inverse wavelet
transform involves a weighted summation of the scaled and translated wavelets so that the original
signal is retrieved. The wavelet transform is basically a multiresolution or multi-scale approach to
signal analysis [5]. In various signal processing applications, the wavelet transform has become a
useful tool that a signal into an alternative domain, i.e. the wavelet domain, so that certain features
of the signal can be more easily identified.
26
MICEEI 2010
The wavelet transform is defined as

1
x b
WTf x
f x
(1).
dx
a
a
This transform can be seen as a mathematical microscope whose position zooms on location b with
a magnification a1 and whose optical characteristic is described by the mother wavelet x . In
most digital signal processing applications, we are only interested in the discrete dyadic wavelet
transform in which scales and spatial translations vary as a power of 2. This dyadic transform is well
matched to multiresolution analysis and fast computations [6]. It has been shown that the dilations
and translations [5]:
jk x 2 j / 2 2 j x k
(2)
j, k Z 2
of a mother wavelet x can be used as an orthonormal basis for the multiresolution

decomposition of signals into octave subbands (by means of dilation) with an excellent spatial
location property (by means of translation). The translation index k is measured in terms of the
wavelets support width. In multiresolution analysis, one also defines the dilations and translations
of the lowpass filtering or scaling function x as
j, k Z 2
jk x 2 j / 2 2 j x k ,
(3).
The wavelet transform of a given signal may be interpreted as the decomposition of the signal into a
set of time frequency functions by the use of translated and dilated basis functions of a mother
wavelet.
A signal f x may be decomposed into detail signals at various scales 2 j and various
locations k with corresponding coefficients d jk and a coarse lowpass component with coefficients
S jk , i.e.,
J
f x d jk jk x S jk jk x
j 1 k
(4)
where J is maximum level of decomposition.

III. Characteristics of Wavelet
Wavelets are characterized by a number of parameters, often mutually exclusive. These are
spatial compactness, orthonormality, regularity or smoothness, and symmetry or antisymmetry.
Symmetry and antisymmetry is not possible for real compactly supported orthornormal wavelets except
for the Haar basis [7]. However, biorthonormal wavelet bases in which analysis (decomposition) and
synthesis (reconstruction) filters, and , are different can be symmetrical or antisymmetrical [8].
They can also be designed to be compactly supported. In general, orthonormal compactly supported
wavelet bases are obtained from the iteration of a dilation equation having a finite number of
coefficients, and therefore the wavelets can be at most piece wise smooth and cannot be expressed in a
closed analytical form e.g., the Daubechies wavelets [6]. Compactness is obviously desirable for efficient
computation and accuracy of detecting localized irregularities. Unfortunately compactness and
smoothness are mutually exclusive for orthonormal wavelets. The trade- off between spatial and
frequency characteristics prevails in all transforms. Spatial Compactness requires a broad frequency
spectrum. Compactly supported wavelets therefore have high frequency content and cannot be smooth.
27
MICEEI 2010
Smoothness of wavelets however, is important in minimizing quantization error in images.

In signal and image coding applications, orthogonality is most desirable because it allows
maximum data compression and independence of coding error and other errors between coefficients.
This requirement however, precludes many other desirable properties and narrows down the choice of
wavelets. For example, linear phase filters are also very desirable for image processing. The filters
therefore are preferably symmetrical. It is important to realize that orthogonality is not essential in
signal representation. Frames of wavelets can therefore be designed to represent signals with specified
degrees of accuracy. In the detection of multiscale edges and discontinuities for signal representation,
compactness and symmetry (or antisymmetry) are more important. The uniqueness of representing a
signal using extrema or zero-crossings of its wavelet transform however, requires exact reconstruction
[3].
As mentioned above biorthonormal wavelet bases, in which analysis and synthesis wavelets are not
the same, may be designed to provide both compactness and symmetry. However, the disparity between
the lengths and coefficients of the analysis and the synthesis filters in the bi-orthonormal scheme results
in more expensive product implementation. The disparity in the frequency responses of the two filters
also causes complications in the frequency selective coding, eg the subjective quantisation and
thresholding of wavelet coefficients incorporating human visual characteristics as shown in [2]. It also
has been shown in [2] that it is more effective to apply the frequency selective coding on the output of
the less smooth one of the two filters. Therefore a better image reconstruction in a subjective coding
scheme is achieved when the synthesis wavelet (or scaling function) is more regular than its analysis
counterpart, i.e., the sensitivity for the coefficients of the transform by is
j
S S
j
2
J
(5)
where S is the frequency sensitivity of the human visual system. This is because in the human visual
system, high frequencies are less sensitive and subjective coding involves a de-emphasis of the high
frequency lobes of the wavelet spectrum. For a smooth wavelet which already has very low high
frequency lobes, a de-emphasis of these lobes will not produce much improvement.
IV. Psychovisual Compression of Wavelet Coefficients
It is well known that the human visual system (HVS) has a bandpass sensitivity characteristic
S with a peak sensitivity between 3.0 to 3.5 cycles per degree (cpd). The sensitivity is reduced slightly
at low frequencies but drops off very sharply at the higher frequencies. Therefore low amplitude high
frequency components may be discarded. The visibility of errors is a function of the local luminance
(Webers law) and the local activity of the scene denoted by A f . The later is also known as a spatial
masking effect. Errors are less visible in bright and busy (in term of edges and discontinues) areas of image.
In these areas, coarse quantization may be used resulting in fewer bits per pixel and a better compression.
j
The visible threshold used to discard weak transform coefficients in subband 2 with visual sensitivity
S j is [9]:
28
Tj
K
Sj
1
N 2j
MICEEI 2010
(6)
jk
where K is a proportionality constant to be determined from a subjective viewing of the quality of the
j
reconstructed image, d jk are wavelet coefficients at subband 2 , and N j is the subband dimension. The
value of K is increased until the level at which a difference between the original image and
reconstructed image is just noticeable by a human viewer. The quantization step size for coefficients
in subband 2 j is given by [10]:
Qj
AF j T j
(7)
j
where AF j is the activity function in the subband at scale 2 .

Available sensitivity function S from psychovisual experiments are usually expressed as a
continuous function of one-dimensional spatial frequency. Current subjective quantization and thresholding
schemes [9], for DCT coefficients have made use of the sensitivity function of the HVS to achieve a very
high compression ratio. Since each DCT coefficient represents a single discretised frequency, the
continuous functions S can be used directly. This is not so straight forward for wavelet transform
coefficients. Each wavelet coefficients represents an octave subband with significant spectral overlapping
j
with neighbouring subbands. We therefore compute the sensity function for a wavelet at scale 2 such as
S j ke
S , , d d
2
1
2
(8)
where k e is an energy constant factor, S x , y is the HVS sensitivity function, j x , y is the

wavelet frequency response corresponding to a highpass filter. To calculate the sensitivity for the lowpass
subband, j x , y is replaced by j x , y .
The non-zero wavelet coefficients, X j m, n are then quantized to X j m, n using the following
operation
X j m, n T j
0 .5
X j m, n 0
Qj
X j m, n
(9)
X j m, n T j 0.5
X j m, n 0
where x is the closet integer value to x. The lowest quantized level is set half a step uniform [9].
V. Simulation Results
The simulation is carried out by the first is to calculate the subband sensitivity for the chosen
wavelet at scale 2 j . The simplest technique to obtain the subband wavelet function j x , y in
equation (8) computationally, is to use inverse wavelet transform followed by a Fourier transform of a
29
MICEEI 2010
unit impulse coefficients being placed appropriately in the subspace at scale 2 j . A comparison of the
compression performance of various wavelet bases using subjective coding is given in this simulation.
In particularly, we used Mallats recursive algorithm [5] for the multiresolution signal decomposition.
The sensitivity function S 0.31 0.69 e 0.29 which has apeak sensitivity at 3.0 cpd
[11] and four level decomposition are used for this simulation. The standard Lenna image of
dimension 512 x 512 is used to test the algorithm. The bit-rate of entropy coding is set to be 0.40 bpp.
Table 1 below shows the simulation results for reconstruction of the image represented by the
weighted peak signal-to-noise ratio (WPSNR) for various wavelet bases.
Table 1: Performance of the reconstructed image using various wavelet bases,
where entropy is set to be 0.40 bpp.
Wavelet Bases
Daubechies (orthonormal)
L=2
L=4
L=4
L=6
Near Linear Phase (orthonormal)
L=4
L=6
Coiflet (orthonormal)
L=2
L=4
B-spline (biorthonormal)
L=2& L 4
L=2& L 8
Laplacian (biorthonomal)
a = 0.3125
a = 0.5000
a = 0.0625
a = 0.7500
WPSNR [dB]
Filter length
(analysis side) and (synthesis side)
38.48
38.68
38.71
38.85
44
66
88
12 12
39.15
39.45
88
12 12
38.51
39.23
66
12 12
40.5
39.22
39
97
36.93
38.42
38.98
39.30
57
57
57
57
VI. Summary & Discussion

It can be seen from simulation results for subjective coding images using orthonormal wavelet
bases show that the length of the filters in general has only a slight effect on the WPSNR performance.
Regularity or smoothness (in term of more long of the filter length) of the wavelet bases is therefore
not as important as symmetry in image coding. This is confirmed by the high values of WPSNR for
the case of the linear phase (less smooth) and Coiflet wavelets as can be seen in Table 1. The
desirability of symmetry or phase linear filters in image analysis and synthesis is well known.
However, the application of HVS sensitivity which de-emphasizes the high frequency components,
tend to smooth out any roughness in wavelet bases and hence decreases the importance of smoothness
30
MICEEI 2010
of wavelet in subjective image coding.

The important of symmetry or linear phase of the wavelet transform is also demonstrated by
the high WPSNR performance of biorthonormal wavelet bases e.g., B-spline and Laplacian wavelets,
compared to the orthonormal bases. The smoothing effect of the HVS used in subjective coding of
images, also has an important implication on the choice of wavelet for the analysis side. Because of
this effect the shorter and hence less smooth wavelet of biorthonormal pair should always be
employed for the analysis side.
It can be remarked that psychovisual image compression can be used to reduce the amount of
bandwidth data without degrading the visual quality of the image to an unacceptable level. For our
future work, psychovisual image compression can be considered and applied to medical imaging
compression. It is well known that medical imaging is widely used in variety of physical phenomena,
ranging from x-ray attenuation to acoustic wave propagation, to measure a staggering number of
variables relating to the patients health [12]. That means medical imaging has a huge of raw
bandwidth data (Giga/Tera-byte), and for this reason the compression is necessary needed, not only
for storage but also for transmission purposes.
References
1. G. Wallace, The JPEG still picture compression standard, Communication ACM, Vol. 34, pp. 30-44,
Apr. 1991
2. Dadang Gunawan and Thong Nguyen, Psychovisual Image Coding Using Wavelet Transform,
Australian Journal of Intelligent Information Processing System, Vo. 2, No. 1, pp 45-52, Autumn
1995.
3. Dadang Gunawan, Psychovisual Video Coding Using Wavelet Transform, Ph.D Dissertation
University of Tasmania, 1995.
4. R. K. Young, Wavelet Theory and Its Applications, Kluwer publisher, Boston, 1993.
5. S. G. Mallat, A Theory for Multiresolution Signal Decomposition: The wavelet Representation,
IEEE Transaction on Patt. Anal. & Mach. Intelligence, Vol. 11, pp. 674 693, Jul., 1989.
6. Thong Nguyen and Dadang Gunawan, Wavelets and wavelets Design Issues, IEEE Singapore
International Conference on Communication Systems, ICCS94, pp. 188 194, Nov., 1994.
7. I. Daubechies, Ten Lecture on Wavelets, Society for Industrial and Applied Mathematics
Philadelphia, Pennsylvania, 1992.
8. Cohen, I. Daubechies and J. C. Feauveau, Biorthogonal Bases of Compactly Supported Wavelets
Communications on Pure & Applied Mathematics, Vol. XLV, pp. 485 560, 1992.
9. D. L. McLaren & D. T. Nguyen, Removal of Subjective Redundancy from DCT-Coded Images, IEE
Proc. Part I, Vol. 138, No. 5, pp. 345 350, Oct. 1991.
10. N. Netravali & B. G. Haskell, Digital Pictures: Representation and Compression, Plenum Press,
New York, 1988.
11. K. N. Ngan, K. S. Leong and H Singh, Adaptive Cosine Transform Coding of Images in Perceptual
Domain, IEEE Transaction on Accoustics, Speech, and Signal processing, Vol. 37., No. 11, pp. 1743
1750, Nov. 1989.
12. M.N. Wenick, C.A. Bouman, R.M. Leahy and J.S. Duncan, The Role of Signal Processing in Medical
Imaging, IEEE Signal Processing Magazine, Vol. 27, No. 4, Jul. 2010.
31
MICEEI 2010
32
MICEEI 2010
33
MICEEI 2010
34
MICEEI 2010
35
MICEEI 2010
36
MICEEI 2010
37
MICEEI 2010
38
MICEEI 2010
39
MICEEI 2010
Printed MSA Suitable for MS and BTS WiMAX

E. Palantei, Rahmatullah M., A.E. Putra and Andani A.
Telematics Lab., Electrical Eng. Department, Universitas Hasanuddin (UNHAS), Makassar, Indonesia,
Phone/Fax: +62-411-590125, E-mail: elyas_palantei@unhas.ac.id ; e_palantei@yahoo.com
Abstract This paper presents the creative and innovative design of microstrip antenna (MSA) applicable in
mobile station (MS) and base terminal station (BTS) of the WiMAX network 2.5 GHz band, respectively. The
designed antennas were investigated through numerical computation and experimental evaluations. Initially, the
printed MSA was built from the rectangular microstrip antenna (RMSA) model. Advanced modification and
optimisation of the RMSA by diagonally chopping two opposite corners of rectangular structure to be
approximately 1/3 of the length L and 1/3 or 1/2 of the width W, respectively, results the significant alteration of
the antenna characteristics such as patterns, fields distribution and operating frequency. The modified antenna
exhibits the capability to work on dual-bands 2.5 and above 3.5 GHz, respectively. The single printed MSA
provides more than 50 MHz of impedance bandwidth. VSWR at the resonance frequency 2.5 GHz is 1.18. The
linear array of printed MSA is discussed as well.
1. Introduction
The antenna as an important part for enhancing the quality of network services is critical in many wireless
designs. The high quality of constructed antennas installed at a particular wireless terminal unit (i.e. MS or
BTS) is an important aspect to maintain an efficient of RF power radiation or reception. A good antenna
design must be capable to illuminate the maximum RF energy to the intended coverage during the
transmission phase. In the received mode, it must have an excellent sensitivity to capture the RF signals from
the vicinity of the wireless terminal.
A tremendous development of various antenna technologies during the last decade has challenged the
new requirements to employ this wireless part for capable of multifunctional operations. In addition, the
future requirement of the wireless mobile networks includes overall size and weight reduction. Thus, the
fabricated antennas must be low cost and low power consumption, integrated with other parts, compact/
small, and user-friendly [1]. These issues are the main concern of the designers, engineers, practicians, and
professionals in order to achieve the sophisticated design aims. Some wireless applications, such as RFID,
sensor network, wireless medical monitoring, WiMAX, and many more, even require more complex
technical specifications. For instance, the antenna must be able to incorporate with RF circuit lay out or
acting as an internal antenna of some mobile terminals (e.g. PDAs, wireless modem, laptop, and mobile DTV
receiver).
This paper outlines the current progress of the R&D antenna design to be deployed in Indonesia WiMAX
network. This will be set-up throughout the country within the next year. The constructed antenna was built
using the simple and well known approaches by modifying the initial RMSA to suit the specified frequency
operation of WiMAX 2.5 GHz through the numerical computation. This executed by utilising FEM-HFSS
programming. The constructed MSA consists of the single element and a linear array configuration. The
material to fabricate the MSA utilised a double layers PCB which widely available in local market. Despite
the disadvantages of the printed MSA especially constructed using PCB material [2-4], however, the
incorporation of the antenna part into the electronic circuit of WiMAX device can be done straightforwardly
[1]. The production cost can be significantly reduced using this approach. This also may enhance the
efficiency of RF power transmission and reception.
40
MICEEI 2010
2. Printed MSA Constructions

2.1 Rectangular Microstrip Antennas (RMSA): Initial Model
A model of the initial rectangular microstrip antenna designed to build the printed MSA for mobile
station terminal and its array configuration developed later for BTS is depicted in Fig.1. The actual antenna
dimension was considered to fit inside the most recent handheld communication terminal such as handphone,
PDAs, palmtop, and GPS receiver. All the antenna physical parameters as shown in Figure 1, such as the
width (W), the length (L), and the thickness (h) were computed using the procedures well explained in [3].
The results of the parameters calculation which determine the antenna size are tabulated in Table 1.
In principle, the patch structure consists of two peaces of rectangular conducting elements (each of size
38 mm x 30 mm and separated by 2 mm distance), ground plane (38 mm x 62 mm), dielectric material (38
mm x 62 mm) using FR4-epoxy, r = 4.4, and small rectangular conducting layer (2 mm x 2 mm). The small
rectangular conducting layer is used to connect the active RMSA elements and to solder SMA connector as
part of the RF port.
Structures
Dielectric Material:
FR4 Epoxy, r = 4.4
Ground Plane
Rectangular Patch
Conducting Wire:
cooper
Small Rectangular
Patch to connect
double rectangular
elements
Figure 1: Initial rectangular patch model to construct the

printed MSA using HFSS numerical computation.
Thickness
(mm)
h=1
Dimension
Width- LengthW(mm) L (mm)
38
62
0.05
0.05
38
38
0.05
Radius
(mm)
-
62
30
(~L/2)
-
Table 1: Antenna structure and dimension
Figure 2: The field distributions of the printed MSA for WiMAX 2.5 GHz application: (a) The electric field distribution, (b) The magnetic
field distribution.
2.2 Single Element MSA

The creative innovation of the printed MSA applicable for WiMAX network 2.5 GHz band was
developed from the initial RMSA depicted in Fig.1. In practical design the printed antenna prototype was
initially modified and optimised to obtain the best antenna configuration suitable for that network
environment. These implemented strikeforwardly in such away by diagonally chopping two opposite corners
of both rectangular structures of Fig.1. The results are an antenna prototype as shown in Fig.2. The optimum
size of the conducting part that must be removed to shift the resonance frequency from 2.3 GHz to 2.5 GHz
is determined by the length L and the width W of top conducting layer. Supposing two corners of the RMSA
41
MICEEI 2010
radiating parts will be chopped by approximately 1/3W and 1/6L as shown in Figs.2 (a) and (b), therefore,
the total area of radiating elements removed at these parts is about 1/18WL mm2. The other two corners as
drawn in Figs. 2 (a) and (b) required to remove will be approximately 1/12WL mm2. The impacts of this
antenna modification were found on the shifting of the operation frequency, and the alterations of both
radiation patterns and field distributions.
The electric field intensity slightly alters while the antenna modified from its initial RMSA model to
create the printed MSA model. In this case, E-field varies from 5.7 x10-3 to 2.8 x 103 [V/m] at the first type
and it varies from 2.8 x10-2 to 5.9 x 103 [V/m] on the new model. Meanwhile, at the initial model the
magnetic field intensity varies from 1.2 x10-3 to 1.9 x 102 [A/m] and it varies from 4.1 x10-2 to 1.7 x 102
[A/m] when the model modified.
The significant property variations appear on its frequency operation and radiation pattern while the
radiating structure modified. There is significant alteration on the resonance frequency which is single band
at 2.3 GHz becomes dual bands at 2.5 GHz and above 3.5 GHz, respectively. These resemble that removing
the area of top conducting layer by (5/36)WL mm2 (or 372.2 mm2) will shift the resonance frequency upward
about 300 MHz and generate the new band above 3.5 GHz. Reducing the area of top conducting layer does
not really affect the impedance matching and bandwidth. In overall, at 2.3 and 2.5 GHz the antenna could
provide the same bandwidth of greater than 50 MHz, while at 3.5 GHz it provides the bandwidth greater than
60 MHz. The reflection coefficients at 2.3 and 2.5 GHz, respectively, are excellent to be less than -20 dB.
This S11 value is equivalent to VSWR 1.18.
(a)
(b)
ReflectionCoefficient(dB)
-5
-10
-15
-20
ModifiedDouble
MSA Rectangular
Modified
Double Rectangular
Initial MSA - Rectangular
-25
1.5
2.5
3.5
Frequency (GHz)
(c)
Figure 3: MSA properties: (a) initial RMSA azimuthal pattern, (b) printed MSA
azimuthal pattern, (c) the reflection coefficient (S11).
The effect of the antenna modification is also encountered at the pattern property where two side
lobes disappeared after modification performed. This will reduce the interference come from the
42
MICEEI 2010
side lobe and increase the efficient transmission of RF energy. Another important impact of the
antenna reconstruction is on the antenna pattern which becomes wider than before. This contributes
to the larger coverage area.
2.3 Array Configuration of the Printed MSA
The next step of the antenna construction in this research is to combine a number of single MSA
to form an array antenna. The fabricated photograph a 4x1 element array of MSA antenna is shown
in Fig.4. A number of interesting results of this modification were partly published in [2].
Figure 4: Array configuration of the printed MSA suitable for BTS
3. Experimental Evaluations
The numerical computation using HFSS software has produced a number of interesting antenna
properties including the electric and magnetic field distributions, the radiation pattern, the
impedance bandwidth, S11, and VSWR. Several data from this numerical investigation were
discussed in the previous sections. Both the single and the array elements were examined,
practically. The results from the actual measurements were then compared with the corresponding
simulated parameters. However, in this paper only the performance of an array structure will be
extensively discussed. A number of plots regarding the performance of the designed array MSA are
drawn in Figs.5 (a), (b) and 6.
(a)
(b)
Figure 5: The impedance matching properties of the printed MSA array: (a) S11, (b) VSWR.
43
MICEEI 2010
Fig. 5 illustrates the impedance matching properties of the printed MSA array in terms of the
return loss and VSWR. The printed MSA array has a different point of resonance frequency
between the simulated (2.575 GHz) and the measured ones (2.253, 2.42, and 2.525 GHz,
respectively). At, the S11 threshold level of 10 dB, the designed array obtained from the simulation
provides a very narrow bandwidth. It is approximately 60 MHz. However, the fabricated prototype
of the printed array can provide very large practical impedance bandwidth, totally, more than 2 GHz.
In general, both simulated and measured testings obtain the VSWR value at each resonance
frequency point achieved 2:1.
Fig.6: Azimuthal pattern properties of the printed MSA array
The simulated and measured pattern properties of the printed MSA array are shown in Fig.6.
These azimuthal patterns represent the capability of MSA array to illuminate better power radiation
into 00 direction (front side). However, the antenna is able to radiate the strong RF signal to other
directions. The different of the maximum and the minimum power is about 40 dB which resemble
very good sensitivity the antenna performs.
4. Conclusions
A typical MS antenna built from the initial RMSA has been demonstrated. Reducing the area of
top conducting layer by approximately 1/6 of the total area of RMSA will affect some antenna
parameters such the operation frequency, the radiation pattern, and the field distributions. The
antenna reconstruction does not really affect the bandwidth, the impedance matching and reflection
coefficient. The main purpose of this antenna modification and size optimisation is solely done to
obtain the most suitable antenna design that working at MS terminal 2.5 GHz. Later on, an array of
single printed MSA was built.
The array configuration of printed MSA itself is potential to be deployed as the BTS antenna in a
particular of WiMAX network. A number of efforts are required to achieve this goal. The extensive
optimization process must be performed to obtain an excellent directional BTS antenna array. The
effort is also required to enlarge the operating frequency around 2.5 GHz and very good shielding
and antenna packaging must be considered to bring the constructed array MSA into the practical
arena.
44
MICEEI 2010
Acknowledgements
The authors would like to thanks to the Directorate of Standardization, General Directorate of Post
and Telecommunication, Ministry of Communication and Information for providing a number of
high quality equipments to allow us to perform some testing of the constructed antennas. The
authors would also extend their sincere thanks to anyone who contributed to the establishment of
the MoU between CWMA, Griffith School of Engineering, Griffith University, Australia and
Telematics Laboratory, Department of Electrical Engineering, Universitas Hasanuddin (UNHAS),
Indonesia. The collaboration is mainly focusing in the R&D and education activities.
References
[1] E. Palantei, D. V. Thiel, dan S.G. OKeefe, Rectangular Patch with Parasitic Folded Dipoles: A Reconfigurable
Antenna, on the Proceedings of the IEEE International Workshop on Antenna Technology (IEEE iWAT) 2008:
Small Antennas and Novel Metamaterials, 4-6 March 2008, Chiba, Japan.
[2] E. Palantei, S. Buwarda, Rahmatullah M, A.E. Putra, D.R.S. Suryono and I.S. Areni, Microstrip Antenna for
Mobile Terminals Operating at 5 GHz Band, accepted for oral presentation at Indonesia-Malaysia Microwave
Antenna Conference (IMMAC) 2010, June 11-12, Universitas Indonesia, Depok, Indonesia.
[3] C.A. Balanis, Antenna Theory: Analysis and Design, John Wiley & Sons, Inc. 3rd Edititon, 2005.
[4] G. Kumar and K.P. Ray, Broadband Microstrip Antennas, Artech House, Inc. 2003.
[5] (2005) HFSS v9.3. Ansoft Corp.[Online Available]: http://www.ansoft.com/
45
MICEEI 2010
Performance of Low Power, Short Distance and Narrow

Bandwidth Remote Mobile Sensor Network under LOS and
NLOS Environments
E. Palantei, J.T. Pakadang, H. Sampealang, and N.N.R.A. Makobombang
Telematics Lab., Electrical Eng. Department, Universitas Hasanuddin (UNHAS), Makassar, Indonesia,
Phone/Fax: +62-411-590125, E-mail: elyas_palantei@unhas.ac.id ; e_palantei@yahoo.com
Abstract This paper presents the novel construction of a typical remote mobile sensor network
consisted of one master unit and a number of slave units. This network was developed to operate at the
ISM band 2.4 GHz. The master unit was built from one set of RF transceiver part connected to PC
desktop/ Laptop. Each slave unit consisted of an antenna, RF transceiver, programmed
microcontroller, and sensors array. Wireless sensor network was tested under two link environments,
both LOS and NLOS. The experimental evaluations were performed including the sensing accuracy
and the coverage service that the sensor network can provide. The programmed computation
algorithms that resemble the network operation mechanism were also examined, at both the master
and slave units.
Keywords: sensor network, computation algorithms, ISM band
1. INTRODUCTION
For the last decade, there have been very significant number of wireless applications to measure and to
monitor the environment phenomena [1-3] including the earth quake, the tsunami, the active volcanoes,
flooding, landslide, the building/bridge, and a nuclear reactor. In most cases, the appearance of human being
in these particular locations is the most crucial issues to do the continuous measurement and monitoring.
This is mainly caused due to the real security problem. The similar situation is happening in the medical
applications, for instance, while there are a limited number of nurses working at the night ship but a number
of patients under medical treatment spreading in a large number of rooms. It is hard to imagine that the
hospital could perform the standard requirement of services without any support from the remote wireless
monitoring technologies. Through the utilization of this technology media perhaps a group of nurses/doctors
could be enough to monitor the patient condition, remotely, from a centre computing room of the hospital.
Since its invention there are some great advancements have been gained on the recent R&D of wireless
sensor networks (WSNs) worldwide. In the future, the design of sensor network must have an efficient
power, low cost manufacturing, installation and maintenance, appropriate packaging technology to minimise
the sorrounding materials effects, low environmental impact on the manufacturing process, and required an
optimisation approach on the network design such as the antenna optimisation and the computation
algorithms at both the master and slave units[1-6].
In practical applications, there are several technical difficulties in this area required to be investigated
and to find some comprehensive solutions. In [1], the traditional issues of the sensor network constraint will
be much related to the limited power sources. This is usually exist in the remote mobile sensors where the
network must capable to adapt to their environment, dynamically. One solution of the power issue, the
energy-efficient MAC protocol was proposed [4]. Another main challenging on the wireless sensor
development was concerning its ability to provide good service/signal to a particular area. The coverage
difficulty is a multidimensional issue in the network [5]. Several interesting solutions were investigated to
improve WSNs coverage including the negotiation and resolution strategies to integrate the data, especially,
46
MICEEI 2010
in applications such as to track the moveable terminals and to handle any obstacles [5]. A proper design of
computation algorithms might have significant influence to improve the network coverage.
Slave - 3
Slave - 4
Fig.1: The Wireless Sensor Network Configuration

There are several R&D projects related to the development of the WSNs which are currently under
investigation at the Electrical Engineering Department, Universitas Hasanuddin. Some of these include
sensor network for early warning systems such as flooding estimation and landslide sensing [7], wireless
environment monitoring, and sensor network for medical applications. This paper presents a simple design of
the WSNs which will be potential to develop for the telemetering purposes as the above mentioned. The
main issue discussed in the WSNs construction including the power limitation, the sensor accuracy, coverage
problems and computation algorithms.
2. SENSOR NETWORK CONSTRUCTION

A typical remote mobile sensor network which consisted of one master unit and a number of slave units
is depicted in Fig.1. Ideally, this mesh network configuration could extend the coverage of services
approximately double comparing to the similar network consisted of one master and two slave units. This
network was developed to operate at the ISM band 2.4 GHz. The master unit was built from one set of RF
transceiver part connected to PC desktop/Laptop and the signal converter. Each slave unit consisted of an
antenna, RF transceiver, and programmed microcontroller connected to the sensors array.
II.1 Antenna System

The constructed antenna used in the network, both at the master and slave units, was designed using the
HFSS modeling tools and then manufactured by printing the final lay out onto the double layer PCB material.
The antenna model is the optimized form which suitable to work in the desired frequency operation 2.4 GHz.
This antenna design is shown in Fig.2.
47
MICEEI 2010
Fig.2: Microstrip antenna for the Sensor Network

In principle, the above microstrip antenna was initially designed from the rectangular microstrip antenna
of 62x40 mm2. The small rectangle and triangle holes were created to adjust the resonance frequency. All the
antennas mounted on both the master and slave units have the same model. They are assumed to have the
identical properties, for instance, the reflection coefficient (S11) and the radiation pattern. These properties
are illustrated in Fig.3. The impedance bandwidth of the antenna was 70 MHz (at below -10 dB level). The
calculated VSWR at 2.4 GHz was approximately 1.18. The pattern property of all the sensor systems (master
and slave units) is approximately omnidirectional pattern. This means that all units can transmit and receive
the same level of power to/from all directions. This antenna pattern contributes to inefficient energy
transmission/ reception and therefore may limit the coverage services. The use of steerable directional
pattern can significantly improve the coverage area of the sensor network. The future direction of the sensor
network development may use the smart antenna to tackle the coverage problems [8-11].
(a)
(b)
Fig.3: The properties of the antenna system: (a) the simulated S11, (b) the measured pattern using ED-3200
antenna trainner.
48
MICEEI 2010
II.2 RF Transceiver
All data collected from the sensor array which are connected through the microcontroller must be
transferred to transceiver part (i.e. Xbee Pro). To keep the Xbee Pro chip working properly it must be
supplied with an input voltage Vcc in the range of 2.8-4.3 Volts. In practical implementation, both master and
slave units are supplied with 9 Volts AA battery. To obtain the suitable voltage level for the transceiver unit
therefore the voltage regulator required. LM 317 chip can be used as one option to implement the whole
circuit as shown in Fig.5. Two external resistors, 240 and 394 ohms respectively, are used to act as the
voltage divider. The main function of the transceiver is different between the transmission and reception
modes. In transmit mode, it will transfer the collected data from microcontroller to be suitable for RF energy
transmission to the master unit. However, in receive mode, it converts the RF-reception from master unit into
the message data that will be understand by the local processing and collector unit (i.e. microcontroller and
PC desktop/Laptop). An identical microstrip antenna was mounted on each transceiver unit via separated
SMA connector 50 Ohms.
There are some parameters of XBee PRO must be set up to guarantee this transceiver module can
communicate with other modules. Actually, there are 24 AT commands could be used for this purpose.
However, the most common use of the AT commands, in practical situation, are only five commands. The
usual procedure to set up XBee PRO parameters was described on the following syntax [12]:
request=+++ //open AT Command
response=OK
request=atmy1 //self addressing =1
response=OK
request=atdl2 //transmit address
response=OK
request=atchc // RF channel connection
response=OK
request=atid3328 //Personal Area Network
response=OK
request=atbd3 //Baud rate 3 = 9600bps
response=OK
request=atwr //write in the Xbee memory
response=OK
request=atcn //close AT Command
response=OK
Fig.4. Instruction to set-up XBee-Pro AT Commands.
II.3
Programmed Microcontroller (Minimum System)
Microcontroller is the most important part in the slave unit. There are various types of microcontroller
that can be utilized to perform as a remote processing unit and data collector. As shown in Fig.2 AT853516PI microcontroller was deployed in the slave unit. The minimum system of this controller is triggered by a
crystal oscillator 11.0592 MHz which combined with 2 pieces capacitor 33 pF. To keep the RST pin of
microcontroller is always in its low logic level during the execution of the controller programming
algorithms the combination of resistor and capacitor may be used and connected to ground. AT8535-16PI
chip is supported by the internal ADC unit to convert the input analog signal (from sensor) into the bit
sequences suitable for transmission in RF transceiver.
To capable for handling various and wide applications the sensor network can have a large number of
sensor nodes (slave units). The configuration of each sensor node design is depending on the type of the
data/information collected. The preliminary construction of the sensor network only a limited number of
sensor devices was used i.e. temperature and light sensor devices. The main goal of this design step is to test
49
MICEEI 2010
whether the whole sensor networks working properly or not. This is also intended to check the quality of the
computing algorithms not only in the slave unit but also at the master unit. Each sensor node consists of four
main units, i.e. power supply (AA battery 9 Volts and IC regulator AN7805), remote data processing unit
(microcontroller AT8535 16PI), RF transceiver part (Xbee Pro), and sensor array.
Fig.5: Slave Unit
3. DATA COLLECTION AND PROCESSING

In most applications, WSNs have the main functions to monitor the environment/system in order to collect,
to process (and some times to store), to display the output of the desired data/information into any type of
visualization/ form (such as text, numeric, graph and moving picture), remotely. The roles of the slaves and
the master units to perform these functions are of the most important things. The standard requirements in
terms of high quality communication link between these nodes are very crucial problem. Even though the
programmed slaves and master units have an extremely different computation algorithm, however they must
capable to communicate each other during the sensing and transferring cycles of the collected
data/information.
Fig.6: The flow diagram of master unit computation algorithms
50
MICEEI 2010
Figure 6 shows the flow diagram of the computation algorithms incorporated in the master unit. After a
short period of time it turned-on, the master unit performs the self initialization (individual reconfigurability).
This is followed by generating the interrogation messages. These messages are then transmitted via the RF
transceiver system to reach the slave units (sensor nodes). Each sensor node will be interrogated sequentially.
Any data responses from these nodes will be automatically plotted on the PC/Laptop monitor. If there is no
response detected the master will send the next messages and wait the sensor node response. In this stage,
there is no change on the monitor display until the next response arrived.
Fig.7: The flowchart of the slave unit computation algorithms

The flow diagram of the computation algorithms set-up in each sensor node (slave unit) is described in
Fig.7. While the sensor node is turned-on then it will immediately execute the self configuration. This task is
done in the microcontroller. As the interrogation messages arrived from the master unit, the microcontroller
will collect all the data from each sensor device (sensor array). During the data collection, the local signal
processing task can be inserted such as filtering and memorizing (data storage). The filtering is required to
eliminate any disturbance effects, i.e. noise and interference. In some cases, the data storage is necessary in
such traffic collision situation and since the need of reliable and continuous data sensing is the main concern
in the data monitoring.
The flowchart of the computation algorithms in the sensor node informs that the data will be collected
from each sensor device, sequentially, at a certain period of time. This data collection is automatically done
as the interrogation messages detected and executed by the microcontroller.
4. PERFORMANCE EVALUATION
To visualize/display the collected data/information from several sensor networks, the monitor of PC
desktop or Laptop is used. This device is connected to Xbee-PRO module via RS232 and MAX232.
MAX232 itself is used to balance the voltage level from the sensor module and computer pheripheral. The
software program used to display the received data is X-CTU software. The real protoype of both master
and slaves units are shown in Figs. 8 and 9, respectively.
51
MICEEI 2010
The experimental investigation related to the performance of the constructed WSNs consisting of one
master unit and two slave units was done in both LOS and NLOS environment at Tamalanrea Campus,
Universitas Hasanuddin. LOS environment in this testing was carried out on the football yard where there
was no obstacles between the master and slaves units. Experimental testing under NLOS conditions were
assumed that environment may be close to the high building, in the forest area, indoor building and along
the university road.
In term of the coverage service, the sensor network connection can approach the distance 200 meter
between the master and slaves, in LOS condition. However, for NLOS situation the networks can cover the
area of 50 meter radius. In case by considering some obstacles exist in the network such as high building
with various thickness the testing shown the interesting results. If the thickness less than 15 cm the radius of
sensor networks coverage can approach the distance up to 30 meter. If thickness > 15 cm the distance is only
about 15 meter.
XBee-PRO
Transceiver
AA Battery 9 Volts
MAX232
RS232
Fig. 8: Implemented Master Unit (Collector Point)

The practical observation for each individual sensor to check the network sensing accuracy and
sensitivity have exhibited the good results. For testing the temperature sensor, the PC monitor was able to
display the collected data in the range of 20 oC-80 oC. Meanwhile, to detect the light intensity from the
sensor device the PC monitor is capable to plot the equivalent light data intensity from 4 to 1500 lumens.
XBee-PRO
Transceiver
AA Battery 9 Volts
Voltage Regulator
AT8535 16PI
Microcontroller
Sensors
Fig.9: Implemented Slave Unit (Sensor Node)
52
MICEEI 2010
5. CONCLUSIONS
The construction of a typical WSNs prototype has been discussed. Its potential applications can be found
in various area of study including wireless environment monitoring, telemedical applications (i.e.
electrocardiograph), and robotic application with wireless control capability. The performance of WSNs may
varies depending on some factors such as the wireless environment variations (LOS or NLOS), the electronic
sensor device sensitivity, the antenna and electronic circuits construction, RF feeding techniques and the
computation algorithms. To improve the WSNs performance further development of this research must
maintain all those constraint factors.
ACKNOWLEDGEMENT
The authors would like to thanks to several faculty members (R&D members of Depkominfo), the Electrical
Engineering Department, Faculty of Engineering, Universitas Hasanuddin, Makassar, Indonesia for their
constructive comments and encouragement concerning the future R&D in this area. The authors would like
to extend their sincere thanks to all the Celebes Microelectronics Crew, especially, Mr. Umar Dani for the
great discussion and helping on the programming stuffs.
REFERENCES
[1]. M.Ilyas and I.Mahgoub, Handbook of Sensor Networks: Compact Wireless and Wired Sensing
Systems,CRC-Press,05.
[2]. D.V. Thiel, Progress in Wireless sensor Design and Networking, Proceedings of 1st Makassar
International Conference on Elecetrical Engineering and Informatics (MICEEI) 2008, Universitas
Hasanuddin (UNHAS), Makassar, Indonesia, pp.1-3, 13-14 November 2008.
[3]. G.W. Allen, K. Lorincz, and M. Weish; M. Marcillo and J. Johnson; M. Ruiz and J. Lees, Deploying a
Wireless Sensor Network on an Active Volcano, IEEE Internet Computing, pp.18-25, March-April 2006.
[4]. W. Ye, J. Heidemann, and D. Estrin, An Energy-Efficient MAC Protocol for Wireless Sensor
Networks, Proceedings of 21st International Annual Joint Conference of the IEEE Computer and
Communication Societies, New York, USA, pp.1-9, 2002.
[5]. S. Meguerdichian, F. Koushanfar, M. Potkonjak and M.B. Srivastava, Coverage Problems in Wireless
Ad-Hoc Sensor Network, IEEE Infocom, January 2001.
[6]. S. Madden, M.J Franklin, J.M. Hellerstein, and W. Hong, TAG: a Tiny Aggregation Service for AdHoc Sensor Networks, Proceedings of the 5th Symposium on Operating Systems Design and
Implementation, Boston, Massachusetts, USA, December 9-11, 2002.
[7]. E. Palantei, R.S. Sadjad, M. Anshar, Indrabayu A., N.N.R.A. Mokobombang, A. Suyuti, Z. Zainuddin,
Z. Muslimin, and N.Harun, Research Prospects in Electrical Engineering and Informatics at Hasanuddin
University: Trends and Challenges, Proceedings of 1st International Workshop on Modern research
Methods in Electrical Engineering (IWoRMEE) 2009, 3-6 August, Horison Hotel and Universitas
Hasanuddin Campus, Makassar, Indonesia, pp.51-59, 2009.
[8]. E. Palantei and D.V. Thiel,Symmetry Problems in Switched Parasitic Smart Antennas, IEEE Antennas
and Propagation-Society (APS) International Symposium, Honolulu, HI, USA, 10 -15 June 2007: pp. 33603363.
[9] E. Palantei and D.V. Thiel, The Impact of Bias Voltage on the Performance of a P.I.N. Diode Loaded
Smart Antenna, Journal of the Japan Society of Applied Electromagnetics and Mechanics (JSAEM), Vol.15,
No.3, pp.274-277, September 2007.
[10] E.Palantei, D.V.Thiel dan S.G.OKeefe, Rectangular Patch with Parasitic Folded Dipoles: A
Reconfigurable Antenna, IEEE International Workshop on Antenna Technology (IEEE iWAT) 2008: Small
Antennas and Novel Metamaterials, Chiba, Japan, 4-6 March 2008.
53
MICEEI 2010
[11] E.Palantei, Constraint Factors on Implementing the Control Mechanism of Adaptive Antenna System,
Proceeding of the 1st Makassar International Conference on Electrical Engineering & Informatics (MICEEI),
4-5 November 2008, Kampus Tamalanrea, Makassar, Indonesia.
[12] E. Palantei, Microcontroller Programming for Smart Antennas: Implementation, Proceedings of 1 st
International Workshop on Modern Research Methods in Electrical Engineering (IWoRMEE) 2009,
Makassar, South Sulawesi, Indonesia, pp.31-40, 3-6 August 2009.
[13] Maxstream Inc., XBeeTM/ XBee-ProTM OEM RF Modules, Product Manual v1.xAx Protocol, 31
May 2007.
54
MICEEI 2010
Remote Patient
Monitoring
Network
Cluster
#
1
2
Santosh Shedge , Anup Kale
Member IET, santoshshedge@yahoo.com

2
FEIT, UTS, 15 Broadway,
Ultimo, NSW, Australia 2007, Anup.V.Kale@student.uts.edu.au
1. Introduction
Remote patient monitoring is rapidly growing area of research with technological growth in Wireless Sensor
Networks. There are several challenges present in this type of research domain. One of the principle
concerns in this domain is bandwidth availability in remote areas and scarce power resources of Wireless
Sensor Networks [1]. In case of remote geographies reliable connectivity of network plays important role in
network based applications. But looking at near future policies for network expansion in Australia will boost
availability of bandwidth in remote areas [2]. This expansion will provide big advantage to network based
applications. But still limitations of power in wireless sensor network demands special approach to design
ad-hoc network to tranceive signals from them. Apart from technological limitations limited availability of
medical staff to take update of patients with long term illness creates new challenge. In this case
Telemedicine application with special heuristic to measure and visualize the patient routine health can be an
answer. This paper presents a solution based on an ad-hoc wireless network and an algorithm to monitor
patterns of patient's health. System architecture of the solution based on Service Oriented Architecture (SoA)
is proposed to work in a distributed environment. This architecture includes Wireless Sensor module,
Computer Vision Module and Visualization and analysis module. Wireless Sensor Network module monitors
patient health parameters periodically and checks for any possible health deterioration. The Computer
Vision module is provided to establish a person to person communication between medical staff and patient
remotely. The Visualization module visualizes trends and patterns of patient health parameters for
understanding of medical staff.
Following sections of paper provides details on previous related research for same domain, proposed
solution for remote patient monitoring, experimental work and conclusions.
2. Previous Research
This section outlines fundamental literature behind proposed research area. Here basic and brief discussion
on Telemedicine, Wireless Body Area Network (WBAN) and Surveillance is conducted.
2.1 Telemedicine in practice
Telemedicine [3] is a branch of medical science in which a patient is diagnosed and treated for a disease
from a remote location. Due to the current rapid developments in Telecommunication, Internet and
Computer hardware advancement, it is possible to receive measurement data with low power signals
measured from patient body and send emergency treatment measures to the patient's mobile or a PC. Thus,
telemedicine gives treatment to the patients which are located in remote places. It avoids cost to the patient
and time for the medical professional to travel to the patient. . Recent developments in Wireless Sensor
Network technology provides low cost wireless sensor units for measuring patient health parameters.
2.2 Wireless Body Area Network in Telemedicine

Establishment of IEEE 802.15 Task Group 6 [4] has enabled specialised approach to Body Area Network
communication standards to optimise low power of sensors and their operation for human body environment.
Other approaches with current wireless network standards includes measuring heart signals [5], Tele-
55
MICEEI 2010
rehabilitiation [6], Telemedicine Monitoring [7], Movement Analysis in physiotherapy [8] and many more.
Thus Wireless Sensor has already started playing big role in Telemedicine applications and can be feasible
for a long term solution.
2.3 Surveillance Camera for Telemedicine

Tele-health can be monitored by using a surveillance camera which can be made possible by video
conferencing[9] technology, or camera based remote monitoring [10]. In this case a patient can communicate
with medical staff using web based video streaming application. This approach provides more confidence to
patients and connects them with the medical staff virtually. Thus, this approach can be very useful to
patients in emergency unattended situations..
3. Proposed Solution
Network based software design model (principle focus on Algorithm and Adhoc-Network)
Figure 1: Conceptual Design of Patient Monitoring system

Figure no. 1 shows block diagram of proposed conceptual design for remote patient monitoring. Neural
Network based monitoring system is proposed to monitor sensory data received from a patient. This sensory
data will include signals received from wireless sensors for measurement of patients physical parameters.
Selection of parameter measurement will depend on an individual health queries or disease specific
symptoms. The standard diagnosis data for a specific selection will be applied to the neural network module
as a training data. Neural network module includes three layers of neurons for receiving input sensory data,
to analyse combined information and to generate output pattern or disease pattern. In proposed conceptual
design multiple patient monitoring clusters can be connected the fuzzy controller.
The fuzzy controller includes set of rules or knowledge base for analysing severity of disease pattern.
This rule base is expert knowledge base on each disease populated and authorized by medical association or
medical experts in a particular medical specialization field. The inference module will be responsible for
making comparative analysis of fuzzified disease pattern and feed from expert knowledge base. It generates
alerts based on this analysis. Defuzzified data will be transmitted to the Central Monitoring System for
medical staff's attention and advice. Central Monitoring System is a Business Logic and Visualization server
has the capacity to monitor multiple fuzzy clients. This hierarchical and selective disease information
monitored from different geographies will be visualized at the central monitoring system. Thus this system is
56
MICEEI 2010
intelligent enough to analyse the symptoms and communicate the emergency measures only in case of high
priority situations.
4. Remote Patient Monitoring using Web Services- Case Study:
Figure 2: Implementation Architecture of Remote Patient Monitoring System
Wireless sensors monitor the different parameters like temperature, pressure of the patients body. Wireless
Body Area Network uses IEEE 802.15 TG 6 standard. The initial communication is in between the wireless
sensor(s) and the sensor gateway. The wireless sensor and the sensor gateway act a single unit; the wireless
sensors providing the raw data to the sensor gateway and the gateway collecting the data from various
wireless sensors. The neuro-fuzzy logic algorithm is executed by the application that collects the data from
the sensor gateway. Tangible and conclusive results can be provided by appropriate web-services that can
be accessed over the internet by the central monitoring server.
A Web-Service can serve as an independent unit to provide its service of the requested parameter from the
client. It can be implemented using Microsoft .NET service framework or Enterprise Java Beans using Java
Technology specifications. Since web-services can be accessed over the internet geographically from any
part on the globe with internet connectivity the patient data can be accessed by the central monitoring server
over the internet. The central monitoring server is accessible to the medical professionals and counsellors.
Thus the medical information of the patients can be visualized on the screen visually.
In case of individual pairs of sensor network cluster and sensor gateway, Secured Socket Layer [SSL]
technology would encrypt the data to reliably transmit it to the Central Monitoring System. In case of bunch
of wireless sensor clusters say in case of hospital chains or clinics a virtual private network (VPN) would be
the best way to associate the hospital/clinic chains with the central monitoring system. Securing the data
over the internet is a major challenge and it can be taken care of using the VPNs. To tackle the reliability of
57
MICEEI 2010
the critical data monitoring to happen over distance apart in the central monitoring room a virtual-private
network (VPN) will give the security and reliability of the sensitive data being transferred on the internet.
To tackle the performance throughput of the virtual private network the internet service provider can tag the
(VPN) in gold class of service to high prioritize and reliably transmit the data. A VPN creates secured
tunnels of internet connectivity between computer networks of two or more geographical locations. In our
case it is can be branches of the same hospital or clinic geographically apart.
5. Conclusion
The work done so far proposes the conceptual design and discusses a case study for implementing this design.
In this work we apply a simple design with the available network and internet connectivity standards using
WBAN standard and web-services using .NET technology. A concept of the infrastructure has been put
forth. To refine this work further, more specialized approach is required for diagnosis using more specific
sensors for specific symptoms. In future a feedback loop from the fuzzy-controller to the patient monitoring
system can be given to provide assistance in case of trivial suggestions when medical practitioner is not
available. In case of emergencies the fuzzy controller will be bypassed and the data will be fed to the central
monitoring server and receive the necessary medical consultancy from the medical experts.
References
[1] Mainwaring, A., et al., Wireless Sensor Networks for Habitat Monitoring, in ACM International
Workshop on Wireless Sensor Networks and Applications (WSNA02). 2002.
[2] www.nbnco.com.au
[3] W. Xiaomin, Z. Weidong, and W. Yulin, The development and application of a provincial telemedicine
center, in proc. of the 2nd Joint EMBS/BMES Conf., Houston, TX, pp. 1843-1844.
[4] www.ieee802.org/15/pub/TG6.html
[5] Nopparat, V.; Keeratiwintakorn, P.; "The three-lead Wireless ECG in Sensor Networks for mobile
patients," Networked Sensing Systems, 2008. INSS 2008. 5th International Conference on, vol., no.,
pp.249-249, 17-19 June 2008.
[6] P. Bonato, Advances in wearable technology and applications in physical medicine and rehabilitation,
Journalof NeuroEnginnering and Rehabilitation, February 2005, Volume 2.
[7] P. Thorne, K. Kulhankova, M. Yin, R. Cohn, S. Arbes, D. Zeldin, Endotoxin Exposure is a Risk Factor
for Asthma: The National Survey of Endotoxin in U.S. Housing, 21st International Conference on
American Journal of Respiratory and Critical CareMedicine, December 2005; 172(11) pp. 1371-1377.
[8] Kifayat, K.; Fergus, P.; Cooper, S.; Merabti, M.; , "Body Area Networks for Movement Analysis in
Physiotherapy Treatments," Advanced Information Networking and Applications Workshops (WAINA),
2010 IEEE 24th International Conference on , vol., no., pp.866-872, 20-23 April 2010.
[9] Celik, Nuri; Baker, James; Youn, Hyoungsun; Iskander, Magdy F.; , "An internet based interactive
telemedicine system for remote healthcare," Antennas and Propagation Society International Symposium
(APSURSI), 2010 IEEE , vol., no., pp.1-4, 11-17 July 2010
[10] Nazeran, H.; Setty, S.; Haltiwanger, E.; Gonzalez, V.; "A PDA-based flexible telecommunication
system for telemedicine applications," Engineering in Medicine and Biology Society, 2004. IEMBS '04.
26th Annual International Conference of the IEEE, vol.1, no., pp.2200-2203, 1-5 Sept. 2004
58
MICEEI 2010
Design of Microstrip Branch Line Coupler with TJunction for Mobile WiMAX at Frequency of 2.3 GHz
Gunawan Wibisono and Indah Riyanti
Department of Electrical Engineering, Faculty of Engineering
University of Indonesia Kampus Baru UI Depok 16424, INDONESIA
e-mail : gunawan@eng.ui.ac.id, indahriyanti25@yahoo.com
Abstract
Mobile WiMAX system in Indonesia operates at 2,3 GHz frequency by using Time Division
Duplex (TDD) technology. In TDD, the transmitter and the receiver use the same single frequency
band. In TDD, thing that must be considered is the separation between the transmitted signals with
the received signals. The function of the coupler in RF WiMAX is to separate the transmitted signal
to the received signal. At the branch line coupler there are two equal output values with the phase
difference of 90. The phase difference is used as the separator between the transmitter and the
receiver. In this paper, a microstrip branch line coupler with a T-Junction will be designed which is
operated at 2,3 GHz frequency with two equal output values of - 3dB and 90 phase differences. In
here also the length and width for each impedance line of the microstrip branch line coupler using
FR4 substrate with a dielectric constant of 4,6 and a thickness of 1,6 mm are calculated. It is shown
from the results that the microstrip branch line coupler operates at 2,300 GHz, the output port on
port 2 is ? 2,754 dB and at port 3 is 3,298 dB, phase shift between the two output is 90,038, the
return loss is -30,689 dB, the isolation is -31,185 dB, the VSWR is 1,060, and the bandwidth is 200
MHz.
59
MICEEI 2010
60
MICEEI 2010
61
MICEEI 2010
62
MICEEI 2010
63
MICEEI 2010
64
MICEEI 2010
65
MICEEI 2010
A Short Text Compression Scheme

based on Arithmetic Coding
#Iwan Handoyo Putro1, Petrus Santoso2, Maya Basoeki3
Electrical Engineering Department, Faculty of Industrial Technology, Petra Christian University
Siwalankerto 121-131, Surabaya 60236
1
iwanhp@petra.ac.id, 2petrus@petra.ac.id, 3m23404047@john.petra.ac.id
Abstract - This paper discusses the making of a short text compression application to increase the number of
characters in short text message service using arithmetic compression algorithm. Due to Java 2 Mobile
Edition specifications limitation, this application is not integrated with the built-in short text application on
the mobile phone. The application is developed based on Java 2 Mobile Edition technology and it could be
installed on a mobile phone that supports MIDP 2.0 and CLDC 1.1. The experimental results revealed that
the percentage of message compression as well as decompression process is strongly influenced by the use of
characters in accordance with a predetermined probability level. It is also showed that this application,
compare with normal short text capability, could send up to 176, 341 and 517 characters for a single, double
and triple short text delivery respectively.
Keywords: short text compression, arithmetic coding, mobile phone
1. Introduction
The use of mobile phone as a means of communication and information has become an important need.
Apart from being used as a voice communications device, mobile phones are widely used for sending short
text message. This short text message has become a favourite service as it has considerably cheaper cost
rather than voice communication.
In general, a short text message consists of text with a size equal or less than 160 characters for each delivery.
If the customers would like to send more than 160 characters then they have to pay it as a multiplication of
160 characters. For example, a message with 161 characters would be calculated as 2 messages. Therefore,
this project was initiated in order to address that limitation and hopefully could give benefits to the mobile
phone customers.
2. Short Text Compression using Arithmetic Algorithm

Briefly speaking, several projects regarding short text compression has been done previously using different
compression algorithm as well as targeted platform [1,2]. However, work on short text compression using
arithmetic algorithm combined with mobile phone, especially the one that supported Java application is
rarely found. Perhaps because of the significant decreased of short text rate together with numerous special
short text rate scheme offered by operators makes this idea looks like a not worth it project to be done.
Nevertheless, it could not be denied that couple characters addition still becomes an interesting idea for a
short text addict.
66
MICEEI 2010
Arithmetic Coding Algorithm is one of the examples of lossless compression algorithm. In this algorithm, a
string of characters would be represented using a fixed number [3]. Similar to Huffman coding algorithm, a
frequently used character would be stored with fewer bits and the other way round in order to keep the total
bits lower. However, arithmetic coding implements a different method in terms of encode the process.
Unlike Huffman algorithm which separates the input into component and replaces it with code, the full string
is encodes using Arithmetic algorithm into a single number between 0 and 1 [4,5]. The flowchart of coding
and decoding process in Arithmetic algorithm could be seen in Figure 1 and Figure 2.
Start
Start
Find the probability of

each character
(between 0-1)
Comparecodevaluewith
probabilitytable
Define the value of
low_range and high_range
from each character
Extract thefirst character

low = 0
high = 1
code_range=simbol high_range- simbol low_range

Take the first character
Code range = high - low
encoded_number =(encoded_number - simbol low_range) / code_range
high = low + code_range x simbol high_range

low = low + code_range x simbol low_range
Compareencoded
number withprobability
table
False
Take the next character
True
Any character
left ?
Findthenext character
False
End
Achievinghighrange (1)
True
End
Figure 1. Encoding process
Figure 2. Encoding process
67
MICEEI 2010
3. Application Design
Compression process will convert the message into a single code. This single code is a set of numbers
obtained from calculations based on the possibility of occurrence of each letter that is predetermined.
Messages that are typed in the TextBox will be collected and stored in the message string and will be
processed to obtain a single code before it is sent to the recipient. The process of encoding or compression is
to use a static table that has been determined. In the table there are 90 characters that may appear in the
message together with its probabilities. These characters are the characters A-Z, a-z, 0-9, and 28 additional
characters such as @, &, * and so forth.
String of message typed in the textbox will be calculated starting from the first characters until the last
character using the probability value. The probability values are ranging from 0 to 1 with a variety of each
different character. After being calculated through the last character, a single code will be displayed. That
code number, in the double data type, is sent to the recipient as a string message.
Decompression process performed by taking a string of messages sent and separating the first code that
indicates the number of characters in the message and the second code is the code that will show the
character message sent. Decompression process uses a table that has been determined. Block diagram of the
process is carried out in Figure 3.
Figure 3. Flowchart of the system
4. Experimental Results and Discussion

Testing is done by sending a series of short message with the number of characters more than 160 characters.
The test is performed using Sun Java Wireless Toolkit 2.5.1 then continued with testing on a Mobile Phone
using Nokia 6681.
68
MICEEI 2010
This test is expected to know the maximum number of characters that can be sent each time it send short text
message. As it is widely known, a normal short text message could send maximum 160 characters, while 2
and 3 short text message could deliver maximum 320 and 480 characters, respectively.
The mobile phone used in this test is using prepaid scheme. Therefore, as part of verification process, before
and after each test the amount of mobile phone credit is checked. The purpose of this stage is to see how
much credit has been used. The experimental test result is shown in Table 1.
Testing number
Number of
characters to be
send
Normal short
text delivery
Number of
short text
delivery using
Application
Table 1 Short text delivery results
162
163
165
169
172
176
179
180
190
10
197
11
312
12
321
13
330
14
341
15
352
16
370
17
498
18
500
19
517
20
580
It can be seen from the Table 1, the application is able to increase short text capacity up to 16 characters or
capable to send as many as 176 characters compare with the original number of characters that can be sent in
a normal text message, 160 characters.
In contrast with standard long text message, for instance: sending 2 short text messages continuously which
is 320 characters maximum, this application can send up to 341 characters. While in the experiment of
sending 3 messages in a row, the number of characters sent is increased up to 517 characters from just 480
characters in standard short text message delivery.
69
MICEEI 2010
5. Conclusions
1. Due to Java 2 Mobile Edition specifications, this application could not access the mobile phone built-in
inbox.
2. Compare with the number of characters sent on normal short text delivery, this application could send up
to 176 characters on a single SMS delivery, 341 characters for sending two short text message and 517
characters on 3 short text message delivery.
References
[1] Islam, M.R., An Enhanced Short Text Compression Scheme for Smart Devices, Journal of Computers,
Vol. 5, No. 1, January 2010
[2] Rein, S., Guhmann, C., Fitzek, F., Compression of Short Text on Embedded Systems, Journal of
Computers, Vol 1, No. 6, September 2006
[3] Howard, P.G., Vitter, J.S., Analysis of Arithmatic Coding for Data Compression, Information
Processing and Management, Vol 28, No 6, November 1992
[4] Langdon, Jr., G.G., An Introduction to Arithmatic Coding, IBM J. RES. DEVELOP, Vol 28, No 2,
March 1984
[5] Witten, H.I., Neal, R.M., Cleary, J.G., Arithmetic Coding for Data Compression, Communications of
the ACM, Vol 3, No 6, June 1987
70
MICEEI 2010
A Query by Singing/Humming System Using

Autocorrelation Function and Hidden Markov Model
#
Indrabayu 1, Abdul Aziz 2, Dadang Gunawan3

Electrical Engineering, University of Indonesia
Kampus Depok Universitas Indonesia
indrabayu.amirullah@ui.ac.id; 2az_e06@yahoo.co.id; 3guna@eng.ui.ac.id
Abstract
Query by Singing/Humming (QBSH) is a most natural way for music recognition. A QBSH system can help find songs
by matching a part of melody users sing or hum. Many Music Information Retrieval (MIR) techniques have been used to
carry out the QBSH goal. In this paper the goal is set to the Meaan Reciprocal Rank (MRR). The built system utilizes
autocorrelation function for pitch tracking. It also employs Hidden Markov Model as matching engine. Two influencing
parameters i.e. Duration and Gender are investigated for best performances in MRR.
1. Introduction
Distribution of music collection in digital format is very common today. With digital format, music
can be organized, collected and created in playlist with ease. The trend has effect in the increase of digital
music transactions. Nowadays, a song provider or distributor can have millions of music collection. Hence, a
proper way to handle abundant of titles need to be studied. The technology of managing those music
database is lately known as Music Information retrieval (MIR).
Consequently, the music information retrieval and then start getting the spotlight. Chai Wei [1] was
interested in making an application system with Internet-based music retrieval. Wei, in his thesis, making a
song based on melody retrieval system as an input query. Melody that is used as input is the humming made
by the user to the proficiency level in these systems. The query than match to the database to find a song that
has some similarities with those inputs. The system is then known as Query by Singing / Humming
In [1], the similarity of the input melody with the database is calculated based on the Euclidean
distance, cosine similarities, Longest common sub-sequence (LCS), Longest Common consecutive of subsequence (LCCS) and modified LCS. Jang et.al. [2] has proposed Hidden Markov model as matching engine
and showed that system worked better in accordance with the number of database. Furthermore in HMM,
Helen et.al. [3] conducted research on similarities using the probability function features. However, most
research in QbSH made use of western or Chinese songs as database. Hence in this research, a QbSH model
is made for Dangdut (Indonesian Ethnic Songs) under HMM Model.
The paper is organized as follows. Section 2 describes proposed the process of MIDI database
collection, pitch tracking and matching engine used in this paper. Section 3 presents the QbSH system design
for simulation. Experimental results given in Section 4 and conclusion are drawn in Section 5.
2. Parameters used in QbSH

2.1 Query by Singing/Humming
Query by Singing / Humming (QbSH) is one branch of the Music Information Retrieval. QbSH is a
search method by using the humming or singing of a user. Search by humming/singing a song is needed
because sometimes the searchers do not know or forget the song title or artist name, and simply knowing
71
MICEEI 2010
several lyrics or some part of melodies. In QbSH system, the query (humming/singing) is matched with the
song database based on its melodies, then the system will provide a ranking of the closest melodies with
songs that referred to in the first rank. The purpose of ranking is given instead of only one candidate to
anticipate when there are multiple versions of existing tracks or songs that have similar melodies.
2.2. Pitch Tracking

Pitch tracking is a method to obtain the fundamental frequency, F0 of a sound signal, which later
became a feature to the representation of melody. Several methods can be used in the pitch tracking, such as:
Time Domain Method, domain Frequency Method, and Auditory Model-Based Method. Time Domain
Method is the most often methods used to estimate the F0. Some functions that can be used in this method,
e.g., ACF (Autocorrelation Function), AMDF (Average Magnitude Difference Function), and SITF (Simple
Inverse Filter Tracking). In this paper ACF is proposed as the pitch tracking. ACF method was chosen
because its simplicity, fast and very suitable for making a fundamental frequency of human voice. ACF
method is a form derived from the cross correlation method. The working principle of the cross correlation
method is one multiplies the signal with other signals. While the method of autocorrelation function (ACF)
signal is the original signal and one other signal is a signal that is already experiencing a shift in time.
2.3.
Matching Engine
Pitch tracking technique was used to obtain the fundamental frequency of hummed queries by the
users. While the representation of a melody to match queries on the database is tracked using the Matching
Engine. There are several methods that can be used to match the melody representation. Generally used
methods are Dynamic Time Wrapping (DTW), Hidden Markov Model (HMM), Linear Scaling (LS), Note
Interval Matching and N-Gram Matching.
In this paper HMM is used as matching engine. HMM is utilize due to its error resistance. Several
error that may occur when taking the query from users are a fast tempo of query, tempo variation, and the
appropriateness of fundamental query. Several other beneficial of using HMM method are the high level of
accuracy and shorter matching process.
3. System Design in QbSH

There are three main process in QbSH system as can be seen in figure 2. The three respectively
process are dealing with the making of MIDI database, query feature extraction and representation of
matching melody.
3.1. MIDI Database

MIDI (Musical Instrument Digital Interface ) database creation process is conducted in two stages as
shown in figure 1. Stage 1 is converting polyphonic MIDI files into monophonic files. The conversion is
necessary since the polyphonic MIDI consist of several channels whereas the only channel required in the
process is vocal/melody channel. Figure 3.2 shows the process of making a database of songs.
Figure 1. The Process of Database Collection
72
MICEEI 2010
In stage 2, all these monophonic files are then stored in a single file in a line matrix form of 1xN
where N shows the number of monophonic MIDI collection. A MIDI toolbox is used in this process taken
from Ken Schutte [4]. In this paper Dangdut songs are used and can be downloaded freely in
www.indowebster.com . From the huge amount of polyphonic MIDI downloaded only 15 MIDI files can be
converted to monophonic that is contain vocal channel.
Figure 2. System Design of QbSH
3.2. Process of Collecting Queries

Audio signal that formed the queries are recorded using a microphone. These recorded audio are then
sampled at a rate of 8000 Hz with a single channel (mono). Six of balanced gender users are asked to sing or
hum only 10 from 15 songs provided. This allows users to sing songs that are familiar to them since Dangdut
is quiet difficult to deliver.
Autocorrelation Function, a variant of pitch tracking method in time domain is then applied. The main
concept behind this method is to measure correlation of a signal with its time-shifted version. The input
signals are divided into several frames which each frame consist of 256 sample points.
Pitch Vectors are acquired from the process of pitch tracking. A vector is compared to its neighbouring
vector to decide a notation. A notation consist of two information i.e. pitch and duration. In this paper, only
the pitch information is takes to form the melody representation. The query processing is conducted using
provided toolbox i.e. recognition toolbox [5], speech and audio processing toolbox [6], dan utility toolbox [7].
3.3.
Matching The Melodies
In the process of matching melodies, each notation is represented as a state. These states are then
modelled in probability density function (PDF that has is a Gaussian Distribution with Mean (x) and
Variance (2) formulate as follows:
73
MICEEI 2010
(1)
Where x is the value of pitch observed.

There are two option of queries in the process of melodies matching. Firstly, the queries are taken from the
record of singing or humming of users that has been discussed previously. Secondly, the queries are taken
from live recording which in this case is limited to 10 seconds duration.
3.4.
MRR (Mean Reciprocal Rank)
Till now, the system is still far from a perfect QbSH, Therefore, the results are often displayed still in the
form of a list of best results that close to the query. The MUSART define a value that states QbSH results
using various methods. This value is the mean reciprocal rank (MRR), which formulated as follows[8]:
1 n 1
MRR
n i 1 ri
(2)
where, ri is the rank of precise query to-i for a number of times queries taken (n).
4. Simulation Results
The system is tested by comparing the melody representation (pitch) between the singing queries and the
songs from the database. As mention previously the songs in the database are in MIDI format. Hence all
pitch from queries are converted to semitone before the comparison. Unfortunately there is a quite difference
of pitch range between queries and database. The range of semitone in the queries is 50-70 while in database
the value varied from 69 to 85. To cope this situation, a normalisation is conducted. The normalisation taken
by divided each pitch with the total energy. Table 1 shows the results of having normalisation before
comparison.
Table 1. Example of having normalisation
Before
Pitch Query
52
51
56
57
56
59
55
57
Normalisation
Pitch from DB
74
76
75
78
74
71
74
72
After
Pitch Query
0,94
0,98
1,02
1,02
0,99
1,03
1,06
1,01
Normalisation
Pitch from DB
0,98
1,01
1,00
1,04
0,98
1,01
0,94
0,98
This paper accommodates two variables of how the QbSH system work best which are the influence of
gender and the length of input duration. The queries are taken from 3 males and 3 females without musical
background (amateur). Eight songs are selected from the database which categorized as the most
popular/familiar songs to the users. The result is shown in figure 3.
74
MICEEI 2010
Figure 3. Comparison of average MRR by Gender

Most of the best MRR are sang by female singers, particularly the ones that have MRR more than
0.5. Two songs are insignificantly won by males singer are Goyang Dombret and Penasaran in MRR of less
than 0.2. The justification of this condition is the closer semitone of queries to the database the more likely to
have better MRR.
Table 2. Pitch Comparison of Males and Females
Pitch
Songs
Darah Muda
Jablai
Sekuntum Mawar
MIDI
MIDI
MIDI
71
54
60
69
53
61
82
53
60
69
57
60
68
55
63
84
56
60
72
56
60
69
58
61
84
57
60
71
58
59
71
55
64
85
57
63
69
58
61
71
55
64
84
56
58
Table 2 shows the comparison of pitch from 3 selected songs. It can bee seen that females pitch has a
closer value to the MIDI pitch in database than males. This validates the reason of females have better MRR
than males. This should not become an issue if all user can sing on the right pitch which might be difficult
for amateur singers.
Another parameter that observed in this paper is the influence of input duration to better performance
(MRR). Two option of duration, 5 and 10 seconds are implemented in the query system. Naturally, the
longer duration should have closer outlook to the database. However, performances in QBSH system not
only based on how close the system shows the song desired but also on how fast the system response to the
query. Faster response can be achieve by shortened the queries which accordingly lesser the processing time.
75
MICEEI 2010
Figure 4. MRR comparison base on duration

Three best identified songs (Darah Muda, Kopi Dangdut and Sekuntum Mawar) show 10s duration
gives better MRR compare to 5s duration. During the recording stage, silent state often occur where users
thinking of the right notation instead of directly singing. This also demand a longer query for the system to
have right recognition.
5. Conclusions
A Query by Singing/Humming system has been built for Dangdut songs. Two influencing parameters,
gender and duration, are explained in this paper. Somewhat the queries from the singers have lower pitch
than MIDI database. This inevitably cause the female queries have better performance compare to males
queries. The input duration also contribute to system performance that the longer the queries will have
better recognition. Several normalization methods might be investigated for further research and addition
of collection for database and queries would give improved precision of results.
Acknowledgement
The first author is a lecturer in Hasanuddin University and now pursuing his doctorate degree at University
of Indonesia under JBIC-Hasanuddin University Scholarship.
References
[1] Wei Chai, Melody retrieval on the web. Thesis. Massachusetts Institute of Technology, 2001
[2] J.R. Jang, C.L. Hsu, H.R. Lee, Continuous Hmm And Its Enhancement For Singing/Humming Query
Retrieval, ISMIR, 2005.
[3] M. Helen, T. Virtanen, "Audio Query by Example Using Similarity Measures between PDF Features,
EURASIP Journal on Audio, Speech, and Music Processing, Volume 2010.
[4] Schutte, Ken.MATLAB and MIDI, available from the link at authors homepage at
http://www.kenschutte.com/midi.
[5] J.R. Jang, "Melody Recognition Toolbox", available from the link at the author's homepage at
"http://mirlab.org/jang".
[6] J.R.Jang, Speech and Audio Processing Toolbox, available from the link at the author's homepage at
"http://mirlab.org/jang".
[7] J.R. Jang, "Utility Toolbox", available from the link at the author's homepage at "http://mirlab.org/jang".
[8] Muller, Meinard, Information retrieval for music and motion. Berlin: Springer, 2007.
76
MICEEI 2010
Handover Schemes for Real-time Session over

Broadband Wireless Networks
#
Muhammad Niswar 1, Elyas Palantei 2
Graduate School of Information Science, Nara Institute of Science and Technology

8916-5 Takayama, Ikoma, Nara 630-0192, JAPAN, niswar-m@is.naist.jp
2
Department of Electrical Engineering, Universitas Hasanuddin
Perintis Kemerdekaan Km.10 Tamalanrea, Makassar, Indonesia, elyas_palantei@unhas.ac.id
1. Introduction
Future ubiquitous networks will consist of various wireless network technologies and mobile users
need an ubiquitous connectivity over heterogeneous wireless technologies. Recently, mobile WiMAX (IEEE
802.16e) has gained serious attention as a means of providing wireless broadband access to mobile users in a
wide area, and it provides QoS for various applications. On the other hand, Wireless LAN (WLAN,
IEEE802.11a/b/g/n) has been the dominant wireless technology and is extensively deployed today to provide
high data rates in a limited area. Therefore, in the near future, since WLAN and WiMAX networks are
intermingled while complementing each other, a mobile node (MN) will be likely to execute many
handovers (HOs) between WLAN and WiMAX networks with different IP subnets.
Meanwhile, there is a huge demand for Voice over IP (VoIP) service over wireless networks. However,
delivering VoIP over wireless has many challenges because VoIP is a delay and packet loss sensitive
application. In a mobile and wireless environment, typically, two main factors degrade VoIP quality over the
wireless network: (1) degradation of wireless link quality and (2) wireless network congestion. First, because
an MN moves freely, the communication quality degrades due to inherent instability of the wireless link
condition. Second, as VoIP is a bi-directional communication, an AP of WLAN and a base station (BS) of
WiMAX become a bottleneck with the increase of VoIP calls.
In this paper, we propose HO schemes for both WLAN-WLAN and WLAN-WiMAX networks. We propose
an end-to-end HO scheme for VoIP considering the wireless link condition and congestion state of network.
The proposed HO schemes aims to preserve VoIP call quality during HO. This paper first describes the HO
triggers for WLAN and WiMAX and followed by HO algorithms for WLAN-WLAN and WLAN-WiMAX
with the simulation experiments to verify the effectiveness of our proposed HO schemes.
2. Handover Triggers
HO trigger plays an important role in deciding and executing HO in order to maintain on-going
session while HO. We need some HO triggers indicating the condition of wireless link and congestion state
of network. From reference [1] [2] and [3], we found that request to send (RTS) retries and RTT between AP
and MN (W-RTT) for WLAN as well as Carrier-to-Interference-plus-Noise-Ratio (CINR) and MN interface
buffer queue length for WiMAX can be significant HO triggers to indicate the wireless network condition.
In terms of HO triggers for indicating congestion state of WLAN, Fig. 1(a) shows the relationship among WRTT, AP queue length and mean opinion score (MOS). Note that, in this study, we use MOS [4] for
assessing VoIP quality and MOS > 3.6 indicate an adequate VoIP call quality. From Fig. 1(a), we can see
that W-RTT < 200 ms satisfies an adequate VoIP quality. On the other hand, Fig. 1(b) shows the
characteristic of HO triggers for indicating the wireless link condition of WiMAX, i.e., RTS retry ratio. From
77
MICEEI 2010
the figure, we can see that RTS retry increases as increase of distance between AP and MN and RTS retry
ratio < 0.6 indicate the adequate VoIP call quality. Therefore, we can set the RTS retry ratio of 0.6 as a
threshold (RTS_Thr) as well as W-RTT of 200 ms as a threshold (W-RTT_Thr) for initiating HO in WLAN.
.
(a). W-RTT, AP queue length & MOS
(b). RTS retry vs. MOS
Figure 1. Characteristics of HO triggers for WLAN
(a). #MN, MN queue length & MOS
(b). CINR vs. MOS
Figure 2. Characteristics of HO triggers for WiMAX

In terms of HO triggers for indicating congestion state in WiMAX, i.e, MN queue length, Fig. 2(a) shows the
relationship among the number of MN establishing VoIP call, MN queue length and MOS. From this figure,
we can see that MN queue length < 12,000 bytes indicate an adequate VoIP call quality. On the other hand,
Fig. 2(b) shows the relationship between CINR and MOS. This figure shows that CINR > 26 dB indicate the
adequate VoIP call quality. Therefore, we can set the CINR level of 26 dB as a threshold (CINR_Thr) as
well as MN queue length of 12,000 bytes as a threshold (QL_Thr) for initiating HO in WiMAX.
3. Handover Scheme for WLANs

We proposed an HO scheme between WLAN-WLAN networks. Our proposed scheme employs
multi-homed MN, i.e., MN has two interfaces (IFs), in order to support soft-HO. HO Manager (HM)
implemented on transport layer controls and initiates HO based on the HO triggers. The HM switches
between single-path and multi-path transmission modes in response to wireless network condition. Singlepath transmission mode means that an MN communicates with a CN using only one IF. Multi-path
78
MICEEI 2010
transmission, on the other hand, means that an MN sends duplicated packets to a CN through two IFs for
supporting soft-HO.
3.1. HO Algorithms
(a) Switching from Single to Multi-Path
(b) HO based on transmission rate
(c) HO based on retry rate

Figure 3. Handover Algorithms for HO between WLAN and WLAN networks
Figure. 3(a) shows an algorithm of switching to single/multi-path transmission when an MN moves into an
overlap area of two APs (AP1 and AP2). An MN associated with two APs transmits a probe packet to both
APs at 500 ms intervals to estimate AP queue length of each AP. If both W-RTT values are below an WRTT threshold (W-RTT_thr), an MN detects that both APs are not congested. Then, the MN investigates the
RTS frame retry rate of the current (single) active IF since its movement also affects wireless link condition.
If the RTS frame retry ratio reaches a retry ratio threshold of single-path (R_Sthr), an HM switches to multipath mode to investigate the wireless link condition of both IFs as well as supporting soft-HO. On the other
hand, if the W-RTT of AP1 reaches W-RTT_thr, i.e., detection of congestion at AP1, an MN switches to the
AP2 directly without switching to multi-path mode because multi-path mode may cause more serious
congestion in WLANs. Finally, if both measured W-RTTs reach W-RTT_thr, an MN then investigates the
wireless link condition by using the RTS frame retry ratio of the active single IF.
79
MICEEI 2010
In multi-path transmission, to maintain VoIP quality, an MN sends the same data packets to both IFs. Hence,
the MN needs to switch back to single-path transmission as soon to prevent redundant network overload. As
shown in Fig. 7, an algorithm of switching to single-path transmission works as follows. An MN measures
W-RTTs of both APs at all times. If either of the W-RTTs is below the W-RTT_thr, the MN switches to an
IF with the smaller W-RTT. If both W-RTTs are simultaneously below the W-RT T_thr, the
MN
then
compares the RTS frame retry rate of both IFs. Figure. 8 shows an algorithm to compare RTS frame retry
ratios of both IFs. If both frame retry ratios of the IFs are equal, the MN continues multi-path mode. On the
other hand, if either of the frame retries is below the ratio threshold of multi-path (R Mthr), the MN switches
to single-path mode through the IF with the smaller frame retry ratio.
3.2 Simulation Experiments

We conducted simulation experiments to verify the effectiveness of our proposed HO scheme
between WLAN and WLAN networks using Qualnet 4.5 [5]. This simulation experiments aims to see
whether our proposed HO scheme can preserve VoIP quality during HO. Figure 4(a) shows the simulation
scenario where MNs randomly move between two AP coverage areas at a speed of 1 m/s while establishing
VoIP call with CN. From Fig. 4(b), we can see that our proposed scheme can maintain adequate VoIP
quality. Although some degradation points exist, the MOS can be recovered very quickly.
(a) Simulation scenario
(b) Simulation result
Figure 4. Simulation experiments for HO between WLAN-WLAN networks
4. Handover Scheme between WLAN and WiMAX

This section describes our proposed HO scheme for WLAN and WiMAX networks. The HO
architecture is similar to section 3 using multi-homed MN and HM in transport layer. The HM switches
between single-path and multi-path transmission modes in response to wireless network condition.
4.1.Handover Algorithms
Figure 5 shows an algorithm of switching to multi- path transmission when an MN is located in an overlap
area of WLAN and WiMAX (AP and BS). Basically, our algorithm first examines the wireless link condition
by observing RTS retry of WLAN IF and mean CINR level of WiMAX IF. After that it examines the
congestion state of wireless network by observing W-RTT of WLAN IF and MN queue length of WiMAX
80
MICEEI 2010
IF. For example, when the initial transmission path of an MN is via WLAN IF, the HM monitors the RTS
retry ratio to detect the wireless link quality. If the RTS retry ratio exceeds threshold RTS_THR (0.6), the
MN switches transmission paths to multi-path mode; otherwise, it examines W-RTT to detect the congestion
of WLAN network. If W-RTT exceeds threshold RTT_THR (200 ms), the MN switches transmission path to
multi- path mode; otherwise, it stays in single-path mode via WLAN IF. When the initial transmission path
of the MN is via WiMAX IF, the HM monitors downlink CINR level to detect the wireless link quality. If
CINR level is less than CINR_THR (26 dB), the MN switches transmission paths to multi-path mode;
otherwise, it examines the MN queue length MN QL to detect the congestion of WiMAX network. If
MN_QL exceeds the threshold QL_THR (12,000 bytes), the MN switches transmission path to multi-path
mode; otherwise, it stays in single-path mode via WiMAX IF.
(a) Switching from single to multi-path
(b) Switching from multi to single-path
Figure 5. Handover Algorithms for HO between WLAN and WiMAX networks
4.2 Simulation Experiments

We conducted simulation experiments to verify the effectiveness of our proposed HO scheme
between WLAN and WiMAX networks. This simulation experiments aims to see whether our proposed HO
scheme can preserve VoIP quality during HO. Figure 6 shows the simulation scenario where An MN, which
is establishing a VoIP call with a CN, moves from the WLAN area to the WiMAX area at 1 m/s, and vice
versa. From Fig. 7(a) and (b), we can see that our pro-posed method can obtain average MOS values of 4.29
for uplink and downlink, respectively, when an MN moves from WLAN to WiMAX. MOS is decreased
sometimes below 3.6 at simulation time of 200 seconds. This is mainly because the MN located in the edge
of WLAN network at 200 seconds where link condition is degraded and also effect of fading model
introduced in the simulation scenario. On the other hand, from Fig. 7(c) and (d), when an MN moves from
WiMAX to WLAN, average MOS values are 4.06 and 4.286 for uplink and downlink, respectively. From
Fig. 7(c), the average MOS of CN is a bit decreased compared to MOS of MN due to uplink latency of
WiMAX, which is mainly raised from the uplink BW request contention and scheduling delays.
Figure 6. Simulation scenario for HO between WLAN and WiMAX networks
81
MICEEI 2010
Figure 7. Simulation results for HO between WLAN and WiMAX networks
5. Conclusion
In this paper, we have proposed HO scheme for VoIP over WLANs networks as well as intermingled
WLAN and WiMAX networks. The HM is implemented on the transport layer of the MN and obtains
effective HO triggers from the lower layer using a cross-layer approach. The HM exploits RTS retries and
W-RTT of the WLAN as well as CINR level and MN queue length of the WiMAX as HO triggers. Our
proposed method aims to preserve VoIP quality during HO between networks with different IP subnets. Our
simulation results show that our proposed HO scheme can preserve VoIP call quality during HO.
References
[1] M.Niswar, et al., Handover Management for VoWLAN based on Estimation of AP Queue
Length and Frame Retries, IEICE Trans. on Information and System, Vol.E92-D, No.10, pp.
1847-1856, October 2009
[2] Niswar, M.; Kashihara, S.; Taenaka, Y.; Tsukamoto, K.; Kadobayashi, Y. & Yamaguchi, S. (2009b).
MS-initiated handover decision criteria for VoIP over IEEE 802.16e, Proceedings of IEEE Pacific Rim
Conference on Communications, Computers and Signal Processing (PACRIM'09), CD-ROM, August
2009
[3] Niswar, M.; Kashihara, S.; Taenaka, Y.; Tsukamoto, K.; Kadobayashi, Y. & Yamaguchi, S. (2010).
Seamless vertical handover management for VoIP over intermingled IEEE 802.11g and IEEE 802.16e,
Proceeding of 8th Asia-Pacific Symposium on Information and Telecommunication Technologies
(APSITT2010), CD-ROM, June 2010
[4] ITU-T:G.107, http://www.itu.int/rec/T-REC-G.107/en.
[5] Scalable Network Technologies (2006), http://www.scalable-networks.comA.
82
MICEEI 2010
Subarray of Elliptical Ring Microstrip Antenna for

Circularly Polarized Synthetic Aperture Radar
#
1
Merna Baharuddin 1, Josaphat Tetuko Sri Sumantyo 2
Department of Electrical Engineering, Faculty of Engineering, Hasanuddin University

Jl. Perintis Kemerdekaan, Kampus Tamalanrea
Makassar 90211, Indonesia
Email : merna.baharuddin@gmail.com
2
Center for Environmental Remote Sensing, Chiba University
1-33, Yayoi, Inage, Chiba 263-8522 Japan
Email : jtetukoss@faculty.chiba-u.jp
1. Introduction
The design of a circularly-polarized microstrip antenna subarray intended for use in circularlypolarized synthetic aperture radar (CP-SAR) systems operated in L-Band is presented. The CP-SAR sensor is
a new sensor for Earth observation applications, with the advantage of sensitive measurements on Earth
object characteristics, providing greater amount of information than conventional linear-polarization systems.
Microstrip antenna has the advantages of lightweight, low profile, and low volume, as well as the relatively
low fabrication cost [1]. In spite of its narrow bandwidth and low gain, microstrip antenna is a suitable
choice for airborne SARs. The subarray is the essential component that will compose the full array CP-SAR
antenna. The full antenna, in turn, will be implemented on an Unmanned Aerial Vehicle (UAV) CP-SAR,
currently under development in the Microwave Remote Sensing Laboratory, Chiba University. The array
antenna requirements are (i) broad-band input impedance, (ii) sufficient Axial Ratio bandwidth, (iii) a
cosecant-squared beam pattern [2] in the elevation plane (E-plane), (iv) a pencil beam [2] in the azimuth
plane (H-plane), (v) relatively good gain and (vi) low cross polarization levels. There are a number of
technical challenges in achieving such specifications required for the CP-SAR sensor. The subarray is
comprised of CP microstrip elements. The microstrip element is an elliptical patch antenna [3] which is
radiating circularly-polarized wave. A single proximity-coupling method is applied for feeding the antenna
element. The subarray design will involve the right spacing between elements in order to minimize the
mutual coupling and to achieve the required pattern. A well-matched feeding system [4] is also indispensable
for the implementation of the dual-polarized circularly-polarized subarray system.
2. Antenna Parameter
The configuration of the radiating element, together with the microstrip line feed and ground plane,
is shown in Fig. 1(a), where important parameters are labeled. The dimensions of the radiator and the ground
plane for the elliptical patches are a = 46.1 mm, b = 43.8 and la lr = 540 225 mm. Spacing between
elements is s = 140 mm. Side view is depicted in Fig. 1(b). The geometry model is implemented on two
substrates, each with thickness t = 1.6 mm, conductor thickness tc 0.035 mm, relative permittivity r =
2.17 and dissipation factor 0.0005. The parameters of the microstrip line feed are l1 = 48 mm, l2 = 49 mm, l3
= 11 mm, l4 = 36 mm, l5 = 10 mm, d = 10 mm, w1 = 3, w2 = 8, w3 = 5, w4 = 8, and w5 = 5. With the width of
the microstrip line feed of 5 mm, the characteristic impedance is approximately 29.9 .
The radiator will generate LHCP by rotating the patch by -45o around the center of the patch.
Simulations with a finite-ground-plane model have been undertaken to optimize the size parameters using a
full-wave analysis tool (IE3D Zeland software) based on the method of moment (MoM) algorithm. A 1 4
83
MICEEI 2010
subarray is actually fabricated and tested. In the following section, the simulation and measurement results
will be described.
Figure 1: Configuration of the 1 4 subarray with proximity coupled feed; (a) top view and (b) side view.
3. Results and Discussion

The array performance is greatly influenced by the element spacing. If the spacing is very small, it
will decrease the gain due to the strong mutual coupling between the feed line and the patch elements. There
is a reduction of mutual coupling whenever the spacing is increased, which will improve the array
performance. However, it the spacing is increased beyond a certain value, the surface wave increases rapidly
and will reduce the gain.
Fig. 2 shows the frequency dependence of the S11-parameter (reflection coefficient) obtained from
the simulation. The bandwidth (S11 <-10 dB) is 59.6 MHz (4.6%). The input resistance and reactance are
shown in Fig. 3.
In Fig. 4, the antenna gain and AR at angle = 0o are plotted against the frequency. The gain of the
antenna is simulated to be around 10.1 dBic at 1.296 GHz. As for the frequency dependence curves of AR, a
crucial parameter for circularly-polarized antenna operation, the values of 3-dB AR bandwidth simulated is
10.1 MHz (0.78%).
Spatial distributions (radiation pattern) of the gain and AR at an azimuth angle Az = 0o (and 180o, xz plane) is shown in Fig. 5 at the frequency of f = 1.296 GHz. From the simulation at the azimuth angle Az =
0o, the 3-dB beamwidth is 20o while the side lobe level is 13.2 dB.
84
S 11 - Reflection Coefficient (dB)
MICEEI 2010
0
-5
-10
-15
-20
-25
-30
-35
1.25
1.275
1.3
Frequency (GHz)
1.325
1.35
Figure 2: Frequency dependence of simulated reflection coefficient.
Zin - Input Impedance (ohm)
Resistance
Reactance
60
45
30
15
0
-15
-30
1.25
1.275
1.3
Frequency (GHz)
1.325
1.35
Figure 3: Frequency dependence of simulated input impedance.
85
10
G - Gain (dBic)
10
4
Gain
Axial Ratio
2
0
1.25
1.275
1.3
Frequency (GHz)
AR - Axial Ratio (dB)
MICEEI 2010
0
1.35
1.325
Figure 4: Frequency dependence of simulated gain and AR at angle = 0o.
LHCP Gain (co-polarization pattern)

RHCP Gain (cross-polarization pattern)
G - Gain (dBic)
10
0
-10
-20
-30
-40
80
60
Az= 180
40
20
20
Theta angle (degree)
40
60
Az = 0
80
o
Figure 5: Simulated gain vs. theta angle (radiation pattern) in the theta plane (Az = 0o and 180o) (x z plane)
at f = 1.296 GHz.
4. Conclusion
A circularly-polarized subarray antenna has been developed for implementing antenna for circularlypolarized synthetic aperture radar (CP-SAR) operated in L-band. The design and optimization process was
carried out using an MoM analysis software. The result has shown that the subarray has met some of the
86
MICEEI 2010
required specifications. Further adjustment, measurement, and test will be carried out to achieve the best and
most optimized results.
References
[1] Garg, R., Bhartia, P., Bahl, I., and Ittipiboon, A., 2001, Microstrip Antenna Design Handbook, Artech
House Inc.
[2] Vetharatnam, G. , Kuan, C.B., and Teik, C.H., 2006, Microstrip Antenna for Airbone SAR
Applications, Available online at:
http://www.remotesensing.gov.my/images/default/publication_3rdmicrowave/3rdmicrowave_paper5.pdf
[3] Baharuddin M,et al., Elliptical microstrip antenna for circularly polarized synthetic aperture radar. IntJ
Electron Commun (AEU) (2010), doi:10.1016/j.aeue.2010.01.012.
[4] Levine, E., Malamud, G., Shtrikman, S., and Treves, D., 1989, A Study of Microstrip Array Antennas
with the Feed Network, IEEE Trans. Antenna Propagation, Vol. 37, No. 4, pp. 426-434.
87
MICEEI 2010
Dynamixel Motor-based Actuator For Mobile Robot

Design
#
1|2
Muh. Anshar 1, Ambo Aco 2
Department of Electrical Engineering, University of Hasanuddin

Jl. Perintis Kemerdekaan KM.10 Unhas Tamalanrea
1
muh.anshar@gmail.com
2
aaj.e05ftuh@gmail.com
1. Introduction
Technology advancement in the area of robotics has a tremendous impact factor in shaping humans
civilization, from industrial to domestic applications. Two elements of the robotics development supporter
are majorly from the implementation of artificial intelligence into the robot mechanisms, and the
development of electronic technology, including sensors, actuators and computers.
From engineering perspective, robot can be divided into three major elements, controller, sensors and
actuators. Actuators, for instance, have been developed for many years and electric motor-based actuators is
the most favourable one to be used around the robot hobbyist, yet some industrial-based robots are built with
this typical actuator.
Various types of motors, particularly, DC motors have been developed and marketed. Recent motor
technology achievement is the integration of processors and sensors hence increases the motor capabilities.
Dynamixel motors, which are developed and introduced by a Korean robot company, ROBOTIS [12], for
instance, has integrated a reduction gear mechanism, a microcontroller-based processor, temperature and
speed sensors, into one mechanism connected in one bus system [8]. For communication, this motor supports
half-duplex serial communication, which in turn reduces the wiring to the motors.
As this motor has its own protocol system, this paper introduces an approach that can be used to synchronise
with this protocol. The proposed mechanism is embedded into a mobile robot called Junior which utilises
two dynamixel motors as the actuator. This project stresses on the monitoring capability and the simplicity in
the hard-wiring of motors. These objectives contribute to the reliability of the monitoring system and, at the
same time promote the aesthetic value of the robot physical appearance.
2. Microcontroller and Robotics

There is an increasing trend in the microcontroller usages, particularly in the area of robotics as more
features blended into a single chip. Some programming languages also support communication through
parallel, serial port or other communication ports by which the microcontroller technology is able to
comprehend with [10]. The AVR microcontroller class, for instance, has been widely used since its
flexibility with its programming environment, such as the C-compiler available. Let say, the Codevision C
AVR is currently being the most favourable of programming environment to be used around the robot
hobbyists in Indonesia [3, 5]. Some works related to microcontroller-based robot applications have been
carried out and are reported in [4, 6, 11]. Recent development has been achieved through the integration of
artificial intelligent onto the robot working mechanisms [7, 8].
88
MICEEI 2010
3. Velocity Conversion
By referring to [7], liner velocity is defined as the comparison of the distance taken for a material to
travel to the elapsed time. The mathematical deduction equation can be formulated as follows:
v r .Equation 1
where v is the linear velocity (m/s), is the angular velocity (1/s), r is the radius of the wheel (m)
4. Related Works
This work is an integrated research part of the research effort to be carried out in the Department of
Electrical Engineering; University of Hasanuddin with an early work is reported in [1]. The work utilises dc
motor as the mobile robot actuator and the research focused on how the robot reacts to various conditions set
by sensor activation timing. There were no a monitoring system available for the robot and the wiring was
complicated as the copper cable used to connect the motors to the main controller.
Some works are reported in [13] which have utilised dynamixel-based motor type. In that reports, some
research are carried out by personal, robot hobbyists, and some are from companies and educational
institutions. Another work is carried out by a cognitive research group at the University of Technology
Sydeny [2] in which they developed the cognitive aspect of a bear robot actuated by dynamixel-based
actuators. Currently, there is no any available report on research on dynamixel-based robot actuator in
Indonesia. Hence, this paper brings forward this idea to be implemented into a mobile robot actuator
mechanism named Junior mobile robot.
The coming sections will explain in details of the robot design along with the experiment stages and the
analysis of the results.
5. Junior Mobile Robot Design

The hardware design is a mobile robot-type with a dual-motor-driven mechanism. The design will
put the microcontroller-based technology as the main element, the robot brain and display unit as additional
feature to display the internal condition of the two-dynamixel motor. The AVR microcontroller-based,
ATMega 8535, is selected to be the robot brain as this microcontroller type is classified as RISC processor.
For data communication, the robot brain utilises UART method with half duplex which is compatible with
the dynamixel motor. Hence, all data transmission from the robot brain to the motors or vice versa will use
this bus system. For identification purpose, both motors is assigned different identification number, ID, as to
avoid data coalitions. The two motors are capable of being accessed through the data bus by the
microcontroller. The accessing forms can be to receive command data and to transmit data upon the request.
In addition, the power supply will be integrated into the robot system to increase the aesthetic value of the
design. The architecture of the Junior mobile robot is depicted in Figure 1, below.
Figure 1: Junior Mobile Robot Architecture
89
MICEEI 2010
It can be inferred from the figure above that the data length appeared in the data bus remained the same for
the two-mode of data operation, transmit and receive mode.
The two-wheel driven mechanism is implemented by attaching the two motors at the rears of the robot and
one free-wheel at the front to direct the moving of the robot. The 3D graphic design can be illustrated in Fig.
2.
Figure 2: The 3D Design of Junior Mobile Robot (dimension shown in cm)
6. Experiment Process
The sequences of experiments are set up to meet the goal, i.e. to analyse the relationship between
the moving speed which is taken from the moving speed register and the measured speed on the dynamixel
motors. This goal is achieved through analysing the comparison of three different distance assignments
theoretically and practically. The stages of the experiments are arranged as follows:
1. The Junior mobile robot is assigned to three different speed settings by changing the values of the moving
speed register according to motor setting register table as shown in Table 1, below.
Table 1: Moving Speed Data Packet Register Setting Dynamixel AX-12+
Instruction Packet of Dynamixel AX-12+ With ID 1 and ID 2

Rotation
0xFF 0xFF
ID
Length Instruction P1
P1
P3
CS
(dec)
1
2
1
2
1023
0xFF 0xFF 0x01 0x02 0x05
0x03
0x20 0xFF 0x03 0xD4 0xD3
768
0xFF 0xFF 0x01 0x02 0x05
0x03
0x20 0x00 0x03 0xD3 0xD2
512
0xFF 0xFF 0x01 0x02 0x05
0x03
0x20 0x00 0x02 0xD4 0xD3
256
0xFF 0xFF 0x01 0x02 0x05
0x03
0x20 0x00 0x01 0xD5 0xD4
2. Motor speed is measured by using a tachometer. The collected data is then calculated by incorporating
Equation 1 mentioned in Section Three.
3. The Junior mobile robot moves with three different path length.
4. The data, collected from stage two and stage 3 are then compared for further analysis.
90
MICEEI 2010
7. Experiment Results and Analysis

This section is divided into several sub sections and each sub section provides the results of the
experiments followed by their analysis.
7.1 Speed Register Values

The measurement results of the angular velocity of the motors in respect to the moving speed
register can be illustrated in the Fig. below.
Figure 3: Comparison of the Moving Speed Register and the Angular Velocity the Motors
It can be seen from the figure above that with the 500-decimal value of the moving register, the resulting
speed is around 23 rpm. By doubling the value of this register, the achieved speed is around 44 rpm, almost
twice the previous speed. Hence, the value of the moving speed register inside the motor processor has a
linear relationship to the speed of the motor.
7.2 Linear Path Scenario

As set up before, the robot moves along a linear path within three different speed setting, 44 rpm, 33
rpm and 22 rpm. Table 2, Table 3 and Table 4 depict the results of the experiments.
Table 2: Experiment at a speed of 44 rpm
Trial
1
2
3
Averag
e
Error
(%)
Travel time (s)

Practical
Theoretical
1
1m 2m 3m
2m 3m
m
11,3 17,2
5,74
0
8
11,3 17,1
5,61
0
7
11,2 17,0 5,7 11,5 17,3
8
7
5,70
0
1 9
11,2 17,1
5,68
7
5
1,85
2,71
1,25
91
MICEEI 2010
Trial
1
2
3
Averag
e
Error
(%)
Travel time (s)

Results of
Results of
measurement
calculation
1m 2m 3m 1m 2m 3m
15,0 22,8
7,37
7
2
15,4 23,1
7,49
2
6
15,4 23,1
15,3 23,2
7,72
7,44
4
6
9
7
15,2 23,0
7,43
9
8
3,71
0,95
0,33
Travel time (s)

Results of
Results of
Trial
measurement
calculation
1m
2m
3m
1m
2m
3m
11,73 23,45 36,06
1
11,35 23,94 36,73
2
11,98 23,99 35,74
3
11,58 23,16 34,74
Average 11,69 23,79 36,18
Error
0,91 2,72 4,12
(%)
7.3 Result Analysis
It can be inferred from the three tables that the increase of value in the moving speed register will
affect the angular velocity of the motor and the speed of the robot eventually increases. The tables also show
that every 255 decimal addition in the moving speed register will increase the angular speed about 11 rpm.
In terms of the reliability of the speed monitoring system, Table 2, 3 and 4 depict the speed setting for the
Junior mobile robot at the speed of 44, 33 and 22 rpm. It can be seen at the speed of 44 rpm, Table 2, for
every one meter addition in distance, the time required increases about five seconds. At the speed of 33 rpm,
depicted in Table 3, for every one meter, the required time is roughly about seven seconds. Finally, Table 4
which shows the result of the test at the speed of 22 rpm, for every twelve seconds, the robot will travel at
one meter distance. It can be seen that for every change in the speed setting will not necessarily affect the
change in a linear trend.
Regarding the error monitoring percentage, the overall experiment results tend to produce insignificant error
reading factor from the internal sensor mechanism of the dyanmixel motor.
In addition, the robot hardwiring that is used to establish a two-way communication, TX-RX data can be
provided through one line cable.
92
MICEEI 2010
8. Conclusions
The study on the implementation of dynamixel motor-based actuator promotes the easy installation
process and with the internal processor, the speed of motor can be monitored. Besides, some other internal
sensors are also available, by which contributes the monitoring system capability.
As the Junior mobile robot is only driven by two dynamixel motors, the wiring is lesser compared to the
other standard DC motors. For motor driver expansion, the wiring remain the same as it just apply the same
principle.
With these early achievements, the future work will dedicate to the implementation of cognitive aspects into
the Junior mobile robot, particularly by incorporating a more dynamic environment to the experiments.
Acknowledgments
Special thank to our research colleagues, Ikhsan and Mr. Christoforus who have provided assistance
in all phases of the whole experiment stages.
References
[1] Anshar, M. & Astuty, T., Robot Basic Berbasis Mikrokontroler 89C51, Department of Electrical
Engineering, University of Hasanuddin, Makassar, 2003.
[2] Anshar, M., Johnston, B., Novianto, R., Stanton, C., Wang, X. & Williams, M.-A., The Bear Project: A
Cognitive Approach to Robotics, the Eighteenth International Conference on Automated Planning and
Scheduling, ICAPS-2008 System Demo Papers, September 14 18, Sydney Australia, 2008.
[3] Bejo, A., C & AVR Rahasia Kemudahan Bahasa C dan Mikrokontroler ATMEGA8535, Graha Ilmu,
Yogyakarta, 2008.
[4] Firmansyah, R., Robot Pengelasan, Surabaya, 2006.
[5] Heryanto, M. Ary & Wisnu, A.P., Pemograman Bahasa C untuk Mikrokontroler ATMEGA8535. C.V.
ANdi Offset, Yogyakarta, 2008.
[6] Iwan S, Darjat & Rano G.N., Pengembangan Sistem Kontrol Navigasi Kursi Roda Cerdas
Menggunakan Arsitektur Subsumption Studi Kasus Door Passing dan Corridor Following. Semarang,
2008.
[7] Lohat, A. S., Gerak Melingkar Beraturan, 2008
[8] Mensink, A., Characterization and modeling of a Dynamixel servo. Netherland, 2008.
[9] Pitowarno E., Robotika, Desain, Kontrol, dan Kecerdasan Buatan, C.V Andi Offset Yogyakarta, 2006.
[10] Prasetia, R. &Widodo, C.E., Interfacing Port Parallel dan Port Serial Computer dengan Visual Basic
6.0., C.V Andi Offset Yogyakarta, 2004.
[11] Prayudha, N.B., Robot Line Follower Berbasis Mikrokontroler dengan Kemampuan Re-Rute, Bandung,
2009
[12] Robotis, Users Manual Dynamixel AX-12, 2006
[13] Online Robot Forum, http://robosavvy.com/forum/
93
MICEEI 2010
MATLAB BASED SIMULATION OF MEASURING HEIGHT LEVEL, MOISTURE, AND

TEMPERATURE IN MINIATURE INDUSTRY
Andani Achmad, Rhiza S Sajaad,

Dadang A Suriamiarja, Joshapat Tetuko Trisumantio
e-mail : andani60@yahoo.com, rhiza@unhas.ic.id
AbstractIn this paper, modelling of control process has been designed representing miniature
industry. Matlab Simulink based equation model has been built to show the characteristic of
the miniature. The miniature consists of 2 silos, 2 conveyers, 2 heater kilns, and 1 cooler. In
this modelling, it is able to simulate physical measurement such as volume, temperature,
moisture and timeline of 5 litre things while being processed in 3 scenarios. Transmitting 5
liter from silo 1 to silo 2 needs 17 minutes and 6 seconds and total time for each cycle is 19
minutes and 8 seconds while there are 0.127 litres material situated outside the silo. In water
added material scenario, the end value of moisture and temperature are approximately same
with the initial value.
Key Wordsmodelling measurement control process
1. INTRODUCTION
Manufacture industries situated in east Indonesian region are dominated by solid material
based industry such as copper and gold, mining, cement, flour processing and packing industry,
tinning tuna industry, etc. East Indonesia region has 72,9 % of Indonesia metal mining [5]. Beside
that, local industry engineer seems hard to know the fast growing of control system technology
because they trust contractor for giving training program. Actually the training is only a method not
science based program. The miniature industry used for modelling the measurement is a miniature
for process control of solid material based industry (Andani Dissertation, 2010)
The easiest way to identify process control loop is by identifying and showing these
elements i.e. process, measurements, evaluations, and control element. [1]
In education world, there are many vendors yielding experimental modul, i.e. Level and
Flow Training System 38-001, Temperature Training System 38-002, Pressure Training System
38-003, pH Training System 38-003 and Industrial Process Fault Finding Training System 34-250.
Feedback Instrument Limited [2] from UK is a process control using software for computer control,
data acquisition, and data management. Tools produced for the education and training are Process
Control Unit for pH, Flow, Temperature, Level and Pressure Control System Edibon from Italy [3].
These companies produce process control tools for industry and the tools were produced using DCS
technology and Advanced process Control (APC) such as Boiler Drum Heat-Exchanger PCTS, Gas
Flow, Pressure, gas temperature and gas pressure PCTS Syntek Group from Malaysia [4].
For simplification on understanding a control system, sometimes it needs to visualize the
control system into a general model i.e. physical and math model.
An important concept in designing a model depends on what is the mission, so it will be
used properly [8].
94
MICEEI 2010
Aggregate formula or equation variables showing fundamental of a system or physical

process are math model. Math model can be expressed in relation functional i.e. dependent
variable= {independent variable, parameter, coercive function} (Chapra, 1988).
Lumped parameter analysis is shown with Ordinary Differential Equation (ODE). Whereas
math tools used for analyzing distribuion paramter is Partial Differential Equations (PDE). Simple
examples that can be ilustrated to show the two methods i.e. ODE which is used to show system
character in one dimension e.g. fluida height level as a function of time and PDE which is used to
analyze temperature profile in a tank. One of the variables used in PDE for analyzing is caused by
different temperature profile on several measurement points. Therefore analyses must be distributed
on several measurement points to get valid model (Labahn, Kohlhaas 1983).
On the other hand, experiment tool needs on some industrial based universities in East Indonesia
regional is really urgent. Hence it is important to build laboratory tools such as experiment modul
and training for process control system which can representing real manufacture industy using solid
material in East Indonesia region.
This paper discuss:
1. Matlab simulink model of the process
2. Simulation result of the model
2. RESEARCH METHOD
In this research, we use modelling and simulation method which consist of
a. Introducing disturbance of input variables. There are no limitations in disturbance but the output
effect must be bigger for allowing identification of a model.
b. Collecting input and output data which doesnt need stability in the end of the experiment
c. Measuring model parameter such as level, temperature and mositure.
a. Research method, Laboratory activity

Laboratory activity generally consist of identifying and categorizing kind of miniature
industry. By this category it will be built dynamic model approximate real system process.
Modelling is built to learn, analyze and evaluate system control representing the industry. Then
math model from the miniature industry is synthesized using Matlab simulink. The simulation is
used to measure parameter like level, temperature, and moisture which representing each part of
measurement on control miniature industry.
For determining the parameters at certain value on the miniature industry, combination of
control tools such as valve, motor, heater, furnace, cooler, dan blower is used. Hence physical
matter of a factory design must fulfill the main part of the industry such as kiln, booler, reactor,
pump, destilation, absorber, evaporator, tank, etc.
The model is then designed based on identified data by considering the aim of a control
system i.e.
a. Diminish external disturbances
b. Guarantee tha stability of process
c. Optimize the performance of the process
The experiment data are taken at Control Engineering Laboratory, Department of Electrical
Engineering Hasanuddin University, Makassar.
95
MICEEI 2010
b. Data processing method

There are 3 scenarios when datum are identified
1. The First scenario, the sand at silo 1 move to silo 2 without heating and cooling process
while valve at silo 2 is closed.
2. The Second scenario, the sand at silo 1 move to silo 2 without heating and cooling process
while valve at silo 2 is opening so the sand will move back to silo 1
3. The Third scenario has the same scenario unless water is added to the sand with heating and
cooling process.
3. MODELLING AND SIMULATION OF CONTROL SYSTEM PROCESSS OF
MINIATURE INDUSTRY
By using modeling theory, the simulation was designed using Matlab Simulink. It is built as
if it operates continuously like miniature industry
The simulation model consists of Silo 1, Conveyer 1, Kiln 1, Kiln 2, Cooler, Conveyor 2 and Silo 2.
Here is the model of miniature industry
Fig.1 Modelling of Minaure System Control

In the simulation it is used volume to level equation
function y = vol_to_lev(x)
vol = x; % [liter/sec]
ht = 100 ; % tinggi tangki [mm]
R1 = 131.5 ; % jari-jari tangli [mm]
hk = 130; % tinggi kerucut kecil [mm]
R2 = 24 ; % jari jari kerucut [mm]
Vt= pi*(R1^2)*ht; % volume tangki [mm3]
Vk = ((1/3)*pi*hk*(R1^2+R2^2+R1*R2)); % volume kerucut [mm3]
V = (10^6)*vol; % volume [mm/sec]
if V >= Vk ;
y = (((V-Vk)/(pi*(R1^2)))+ hk);
96
MICEEI 2010
else
y = ((3*V)/(pi*(R1^2+R2^2+R1*R2)));
end
4.
a.
RESULT AND DISCUSSION
Scenario 1, Silo 2 Valve is closed
The sands used ini this experiment is 5 liter (equivalent with height level 169.3528 mm ),
initial moisture 5.8 % and temperature 260C without addition water. Kiln 1, kiln 2, cooler, and
blower are inactive condition.
Silo 1 has been fullfilled with the 5 liter sand, then the sand move to silo 2 through conveyer
1, rotary kiln 1 and rotary kiln 2 .
The result figure of scenario 1 on silo 1 where valve at silo 2 is closed is shown in the
following figure
Chart 1. Silo 1 measurement

Chart 1 show that cone part of the silo is emptied in 7 minutes and then the cylinder part is
emptied in 17 minutes and 6 seconds. 0 mm height level is happened on 20th minute and will be
constant continuously. The chart shows that at the end of 20th minute , measured temperature is
material temperature, that is 260C, while after 20th minute, the temperature rise to silo temperature,
that is 280C. The measured moisture is the sand moisture, that is 5.8 %.
According to silo 2, the level, temperature, and moisture is show in the following chart
Chart 2 Condition on silo 2 scenario 1

From the chart we can get information that at silo 2 there has been delay which is
accumulated delay of conveyor, kiln1, kiln 2 , and cooler. Silo 2 is filled where filling cone part first
is faster than cylinder part and will be full on 18th minute at level 169.4, equal with 5 litre. The
moisture doesnt change because there is no heating process
97
b.
MICEEI 2010
Scenario 2, valve silo 2 opened at minute 10
5 litre sand (equivalent with level 169.3528 cm) with initial moisture 5.8 % without
addition water move to silo 2 while kiln 1, kiln2, cooler and blower are still inactive. Valve on silo
2 is closed first and then opened on 10th minute. The measurement of silo 1 is shown in the
following chart
Chart 3 Silo 1 conditions where silo 2 opened on 10th minute

Material level decreases continuously until minute 10.1 then start to increase on height level
52.5 which equal with 1.15 litres. This happens because valve at silo 2 is opened on 10th minute at
level 117 cm which equal with 2.62 litres. Hence there are still 1.23 litres material which are outside
silo 1 and silo 2.
Scenario 2 on silo 2 are represented in the following chart
Chart 4. Silo 2 when valve opened at minute 10
From the chart it is seen that the silo start filled after a few delay about 2 minute 2 seconds
and then continuosly increase until silo 2 valve is opened then it will start to decrease and tends to
approximately constant on height level 134.68 mm which equal to 3.2 litrse.
According to temperature measurement, it is seen that whenever the material are still in silo
2 the temperature remain constant to the material initial temperature, that is 260 C. But when the
silo is empty the measured temperature shows the silo temperature, 280 C.
Moisture chart basically has the same concept with the temperature. When there is no
material in the silo the moisture is 0% and when there is material in the silo the moisture is 5.8 %.
98
MICEEI 2010
c. Scenario 3, Valve of silo 2, Kiln 1, Kiln 2 , and Cooler are opened.

This scenario is done by adding water so the moisture is 37.9% while the temperature is
25.450C. Kiln 1 valve is opened 30%, kiln 2 is opened 40% and cooler blower is turned on 64%.
Simulation is running so that chart result of the scenario on silo 1 is shown on the following chart
Chart 5 Silo 1 while valve on silo 2, kiln1, kiln 2, and cooler are opened
It can be seen that the material decrease when simulation has been running for 10 minutes
where the height level seems continuosly approximately constant on 92.6 mm which equal to 2.06
litres.
It can also be seen that the temperature remain constant on 25.5 0C until 10th minute where
water is added on silo 1 and silo 2s valve opened then temperature slowly down and remain
constant to 24.10C on minute 65 to 150. Moisture also changes, from 37.9 % rises to 41.75% and
then falls to 39.05 % on 23th minute then remain constant until 150th minute.
Kiln 1 condition
On chart 6, it can be seen the condition on kiln 1.
Chart 6 kiln 1 condition when valve fo silo2, kiln1,kiln1, and cooler are opened.
From the chart, material volume on kiln 1 flow constantly 0.2476 litre/minutes. It can be
seen to that the temperature seems constant 38.20C on 0.6335th minute until 10.6 then decrease to
36.20C on 3535 minute then remain constant until 150th minute. Acording to moisture, it seems rise
to 11.37% on 10.94th minute then rise again to 12.27% on 11.63th minute and finally fall down and
remain constant to 11.72 on 14.6th until 150th minute.
99
MICEEI 2010
Kiln 2 condition
Chart 7 Kiln 2 condition when valve on silo 2, kiln1,kiln 2, and cooler is opened.
From the chart, material volume on kiln 2 flow constantly 0.2476 litre/minutes. Temperature
seems constant after 0.923 minutes delay on 53.50C until 11.5th minute then fall down and remain
constant to 50.50C on 32.7th until 150th minute caused by the opened valve on silo 2.AccOrding to
the moisture, it semms constant 5.62% until 11.2th minute when the silo 2 valve is opened. The
moisture riseS to 6.13% for a while then fall down again to 6 % on 13th minute.
Cooler Condition
Here is the chart showing the condition on cooler
Chart 8 Cooler condition when valve on silo 2, kiln1, kiln 2, cooler, are opened
From the chart 8, the volume flow 0.2476 litre per minutes. The temperatue remain constant
27.60C after delay until 12.64th minute caused by the opened silo 2 valve on 10th minute then falls
down and remains constant to 26.30C on 32th until 150th minute. According to the moisture, it can
be seen that it also remain constant 5.685 % until 12.2th minute after delay then rise to 6.13% on
12.05th minute and finally falls down again to 6% on 13.54th minute
Silo 2 Condition
According to silo 2 condition when valve is opened is shown in the chart 9.
100
MICEEI 2010
Chart 9. Silo 2condition when its valve is opened on 10th minute
From the chart it can be seen that silo 2 start to be filled by the material after a few minutes
delay between silo 1 and silo2. From the chart the material volume situated in silo 2 never close to
5 litre. After valve of silo2 is opened, the height level reach its maximum at 82.82 whish equal with
1.85 litre and continuously constant until 150th minute.
The temperature in silo 2 shows 280C for 2 minutes and 2 seconds when the material still
doesnt reach silo 2. When the meterial is in silo 2 the temperature will decrease to 27.640C on 8th
until 12th minute then fall slowly and continuously constant to 26,20C on 65th until 150th minute.
According to the moisture, the chart show that when the material is still in the kiln 1, kiln2, and
cooler and not yet in silo 2, the moisture is 0%. When the material have been in silo 2 the moisture
increase to 5.68% on 9th until 12.2th minute. Then it will raise again to 5.97% on 13th minute
because of the opened valve of silo 2 on 10th minute. When the speed material come in silo 2 has
the same amount with the material leaving silo 2, the moisture slowly decrease to 5.86% on 18.6th
minute and constantly remain at that level.
5. CONCLUSION
1.
2.
3.
Math model synthesized by Matlab Simulink consist of two silo, 2 belt and cup conveyer, 2
heater kiln, and cooler.
There are 3 scenarios that can be simulated through this modelling measuring heigth or volume
level, moisture, and temperature not only in silo but also in kiln and cooler.
The measurement result show that by numerical method it can be calculate the time for the
material move from silo 1 to silo 2 i.e. 17 minutes and 6 seconds (scenario 1), and the time for
the material to do one cycle from silo 1 to silo 2 and back to silo 1 again is 19 minutes and 8
seconds (scenario 2) while there are 0.127 litre of material outside the silos, and also moisture
and temperature variation which show 5.86% and 26.20C resepectively.
Recommendation
1. This simulation and modelling can be upgraded by adding automatically controller with
some disturbances
2. The measurement result of the simulation can be used as a reference on measurement result
of the minature controller industry
101
MICEEI 2010
References
1. Smith, Carlos A, and Armando B Carripio, [1997], Principle and Practiceof Automatic
Process Control JohnWiley & Sons. Inc NJ, USA.
2. Andani, Rhiza S Sajad Model Simulink dari Miniatur Sistem Kendali Proses pada
Industri Manufaktur Berbahan Baku Padat Prosiding Seminar National Teknologi Industri
Fakultas Teknologi Industri Universitas Trisakti Maret 2010 ISBN : 978-979-18265-2-5.
3. Andani, Rhiza S Sajad Pemodelan Kendali Proses Industri Manufaktur Berbahan baku
Padat (Jurnal Teknik Elektro Politeknik Negri Ujung Pandang Elektrika Edisi Juli 2010,
Tahun 8 Nomor 2 ISSN 1412-8764).
4. Andani, Rhiza S Sajad Miniatur kendalian industri manufaktur berbahan baku padat
(Jurnal Teknik elektro Trisakti JE Tri Volume 10, Nomor 1, Agustus 2010, ISSN 14120372).
5. Andani Miniatur Kendalian Proses Industri Berbahan Baku Padat Disertasi S3 unhas
2010.
6. Feedback Instrument Limited Electronin catalogue (BESA) 2006 www.fbk.com email :
feedback@fbdk.co.uk.
7. Edibon Technical Teaching Equipment, Edibon International issue 01/06 (Juni/2006)
www. Edibon.com, edibon@edibon.com.
8. Syntek Group specialist in Proses Control Training System Malaysia Webside : www.
Syntekgroup.com, Email : info@neuder.com. (2004)
9. Directorate General of Mineral Coal and Geothermal Indonesia and Mineral, Coal,
Geothermal and Groundwater Statistics, ISSN : 1410 2196 Edision 2008.
10. Lincoln F. Lautenschlager Moro Process technology in the petroleum refining
industry*/currentsituation and future trends Computers and Chemical Engineering 27
(2003) 1303_/1305.
11. Curtis D, Johnson Process Control Instrumentation Technology, John Wiley & Sons
1977.
12. Thomas E Marlin Process Control Designing Processes And Control Systems For
Dynamic Performance Mc Graw Hill.
13. Togar M. Simatupang Pemodelan Sistem Nindita Klaten 1995
14. Thomas Wahyu Dwi Haetanto & Y Wahyu Agung Prasetyo Analisi dan Desain Sistem
Kontrol dengan MATLAB Penerbit Andy Yagyakarta 2002.
102
MICEEI 2010
Product Quality Inspection Based on Machine Vision

System
Habibullah Akbar 1, #Anton Satria Prabuwono 2, Hasniaty A. 3, Zulkifli Tahir 4, Taufik 5
1
Faculty of Information and Communication Technology, Universiti Teknikal Malaysia Melaka

Melaka, Malaysia, Email: habibrown@gmail.com
2
Faculty of Information Science and Technology, Universiti Kebangsaan Malaysia
Bangi, Selangor, Malaysia, Email: antonsatria@ftsm.ukm.my
3,4
Department of Electrical Engineering, Universitas Hasanuddin
Makassar, Indonesia, Email: hasniaty@yahoo.com; zuledm@gmail.com
5
Electrical Engineering Department, Cal Poly State University
San Luis Obispo, CA, USA, Email: taufik@calpoly.edu
1. Introduction
Machine vision has many applications in industrial automation and covers the full range of technical
difficulty in computer vision [1]. The system is developed by integrating multi disciplinary subjects such as
computer science, optics, electrical engineering, software engineering and industrial automation. This
including image acquisition devices, artificial intelligent for image processing software, computer, and
actuator such as robotics arms to dissociate rejected and accepted products if intolerable defect detected.
In the manufacturing process there are several cases that cannot handle by experts. First time new
kind of product is going to be mass produced, few parts should be sampled to confirm whether
manufacturing process running properly or not. While inspection begins, the manufacturing process is still
running to produce parts. Product which has certain degree of complexity such as automotive part, takes a lot
of time to be inspected since experts measure point by point and filling the quality inspection list. If the
result confirms that the parts being produced is out of standard, then the production process will be stopped.
At the time, there will be hundreds or thousands defect parts that had been produced. This will be an
unwanted situation for any manufacturing industry.
The other case has illustration as follow, since the
market increase then productivity process should be improved to reach the target market. Although human
approach is very smart for inspecting products but it is not possible for increasing the productivity speed.
This caused by the limitation of human eyes, low speed movement, illness, boring and human error. Besides,
there are available technological tools for increasing quality control by using machine vision approach.
Although machine vision is a powerful tool and effective for industry as well as described in many literatures
but the cost is very expensive. The cost has range from USD 30,000 up to USD 200,000 or even more [2].
2. Quality Inspection in Manufacturing Process

In industrial application, quality control of product requires visual inspection in order to ascertain the
quality or correctness of manufactured objects and the precise measurement of the size and features of
objects. In addition, product defect rates have been improved from 0.1% in 1970s to 0.001% in 1990s
in modern manufacturing [3]. In many cases assembly line production which are moving at high speed
would be impossible for human inspector to perform inspection tasks.
The problem can be solved by replacing human inspectors with acquisition devices such as high
speed strobe illumination or exposure controlled cameras synchronized and real time image processing
computer-based. The output of image processing software can also be used to triggering actuator such as
robotics arms or pneumatic pipe to reject an unwanted product. In addition, human inspectors can only
use visible light. Besides, machine vision can use another type of electromagnetic radiation. The
application has been used for a lot of application such as x-rays is used in biomedical application to
103
MICEEI 2010
detect disease of human body; infrared is used in thermograph to determine temperature distribution in
metal, and radio wave ray is used in astronomy. Figure 1 shows the table of typical application of
electromagnetic waves.
Gamma
X-ray
UV
Infrared
Radio
Astronomy Biomedical Atmosphere Thermal Astronomy
Figure1: Typical Application of Electromagnetic Wave
There are a lot of methods and techniques that have been developed by researchers to apply machine
vision in industrial application. Most of them are specific depend on the application. A better understanding
of methods that employed machine vision in industrial application can be obtained by considering several
samples.
3. System Structure and Application

Figure 2 shows the structure of a typical machine vision system [4]. A computer is employed to
process the acquired images. This can be achieved by applying special purpose software for image
processing analysis and classification. One or more cameras placed at the scene under inspection to
acquire object images. The positions of the cameras are usually fixed. Normally industrial automation
systems are designed to inspect only known objects at fixed positions. The scene is appropriately illuminated
and arranged in order to facilitate the reception of the image features necessary for processing and
classification. These features are also known in advance. When the process is highly time-constrained or
computationally intensive and exceeds the processing capabilities of the main processor, application specific
hardware (e.g., DSPs, ASICs, or FPGAs) is employed to alleviate the problem of processing speed. The results
of this processing can be used to:
Control a manufacturing process (e.g., for guiding robot arms placing components on printed
circuits, painting surfaces etc.).
Propagated to other external devices (e.g., through a network or other type of interface like
FireWire) for further processing (e.g., classification).
Characterize defects of faulty items and take actions for reporting and correcting these faults and
replacing or removing defective parts from the production line.
The requirements for the design and development of a successful machine vision system vary
depending on the application domain and are related to the tasks to be accomplished, environment, speed
etc.
Figure 2: Structure of a Typical Machine Vision
104
MICEEI 2010
3.1 Confirming Press Part

One application that arises in die stamping industry, press parts is required to have high precision of
dimension. The most common method used for quality control of press part is performed by employing
experts. They use checking fixture to measure dimension such as edge, angularity of two edges, length of an
edge, corner point position, circularity of a cylindrical shot, and rectangularity of rectangular slot. Point by
point measurement is performed by using rulers and other measuring tools. Experts require many steps of
measurement and the result is low performance for the quality control. Although they have wide range
tolerance of environment ambient and background knowledge, but they can suffer of illness, boring and
human error.
The other method for press part inspection system is using coordinate measurement machine (CMM).
The CMM measures 3D coordinates by contacting a probe to the measurement point of part. The 3D
coordinates then display and recorded as database by software. The CMM example as shown in Fig. 3. The
most advanced method is using 3D laser scanner. It scans area of part by transmit laser signal then the
reflectance from surface part focused by lens and then recorded by CCD/PPS sensor. The typical device has
scan speed up to 120,000 measurements per second. Laser scanner also has capability to generate 3D CAD
model of inspected part using integrated high performance software.
Figure 3: Bright-STRATO Series 355 High Accuracy Large CNC CMM
3.2 Defect in Integrated Circuit Packages

Integrated Circuit (IC) application offers unlimited great advantages for electronics application. For
example, advanced integrated circuit such as microprocessors can control many objects such as computer,
hand phone and humanoid robot. One problem faced by manufacturers is that defects may exist on the IC
leads and surface of the IC chip [5]. During packaging operations, chips are transported at a conveyor.
The typical human vision inspection is locating bend leads in the standard IC packages. Experts hold
the object up and examine the reflections of light sources to recognize whether any defect detected. The
inspection process takes a lot of time. The case gets worse when experts have to inspect smaller chip such as
surface mounted device (SMD). Human approach needs microscope to help magnification because human
vision has limited resolution. In the machine vision system, inspection process can be improved faster, more
accurate, and it is unnecessary to use microscope for convenience. This can be obtained by using high
resolution smart camera, computer, and artificial intelligent algorithm for object recognition. In addition, the
machine vision approach allowed real-time inspection system. This resulted in some systems could inspect
chips at rates of over 10 per second, speeds at which the chips are only a blur to human inspectors.
4. Discussion
Human approach is effective for confirming quality of press parts even there are list of disadvantages
as described above. However, performing inspection task using experts is definitely inefficient. To improve
efficiency, manufacturing industry must employ machine vision approach to handle high speed line
production. The human eye could be replaced using camera or sensor. Data captured then transmit to
computer-based system to enter the image processing stages and finally analyzed by Intelligent Software to
perform object recognition.
105
MICEEI 2010
The software system typically consists several stages such as of image acquisition subsystem to get
information of the captured scene; a preprocessing subsystem to removes noises of captured image; an image
enhancement subsystem to optimize image contrast; a segmentation subsystem to divide the images into
regions that presumably correspond to structural units in the scene or distinguish object to interest [6]; a
feature extraction subsystem to localize possible object for further analysis and measurement; a classification
subsystem to identifying objects or regions in the image. Finally, to grading quality of part, artificial
intelligent subsystem can be integrated to analyze whether part is accepted or rejected. The output then
commands rejecting mechanism such as robotic manipulator to dissociate the defect part. This approach is
common for developing automated visual inspection software in industrial application.
5. Conclusion
Manufacturing process often needs machine vision to increase the productivity process. Automated
visual inspection system reduces the cost by replacing human inspector and increases the efficiency by
improving duration of inspection process. Although the quality control is improved, the cost for employing
automated inspection system is expensive. Higher performance, robust, reliable, and intelligent computer
vision-based system causes the equipment getting more and more expensive. However, low cost automated
inspection system is possible and has been developed to create low cost automated inspection system. This
approach is using low cost image acquisition device such as web camera and high efficient proper software
technique for image processing. Of course, the overall inspection system will be not compete with standard
industrial which using high-end devices such as laser scanning. At least but not last, lower level industries
can use it to produce product reasonable quality for customer. This will confirm the existence of the
industries itself. In addition, the implementation of electromagnetic radiation other than visible spectrum for
data acquisition is widely open for further research. As techniques and vision systems developed, low cost
and better application can be expected in the future.
Acknowledgments
The authors would like to thanks Faculty of Information Science and Technology, Universiti
Kebangsaan Malaysia for providing facilities and financial support under Junior Researcher Grant No.
UKM-GGPM-ICT-103-2010 and Research University Operation Budget of Pattern Recognition Research
Group No. UKM-OUP-ICT-36-186/2010.
References
[1] J.M. Bowskill, T. Katz, J.H. Downie, Solder inspection using an object-oriented approach to machine
vision, SPIE Proc. Machine Vision Application in Industrial Inspection III, Vol. 2423, San Jose, CA, pp.
34-45, 1995.
[2] T.S. Newman, A.K. Jain, A survey of automated visual inspection, Computer Vision and Image
Understanding, Vol. 61, No. 2, pp. 231-262, 1995.
[3] Jihong C., Daoshan Y., and Huicheng Z., Avoiding spurious reflections from shiny surfaces on a 3D
real-time machine vision inspection system, Proc. IEEE Instrumentation and Measurement Technology
Conference, St. Paul, Minnesota, pp. 364-368, 1998.
[4] E.N. Malamas, E.G.M. Petrakis, M. Zervakis, L. Petit, J.-D. Legat, A survey on industrial vision
systems, applications and tools, Image and Vision Computing, Vol. 21, pp. 171-188, 2003.
[5] D.P. Mital, T.E. Khwang, An intelligent vision system for inspection of packaged ICs, Proc. the 4th IEEE
Region 10 Int. Conf. TENCON , Bombay, pp. 1003-1006, 1989.
[6] J.C. Russ, The Image Processing Handbook, 5th Edition, CRC Press, Boca Raton, Florida, 2007.
106
MICEEI 2010
MultiAgent System Modelling Using JADEX

Irawan 1, Atthariq 2, Surya Sumpeno 3
1
Jurusan Teknik Elektro, Politeknik Negeri Ujung Pandang 1 Department of Electrical Engineering,
Polytechnic Ujung Pandang
irawan@poliupg.ac.id irawan@poliupg.ac.id
Jurusan Teknik Elektro, Politeknik Negeri Lhokseumawe 2 Department of Electrical Engineering,

Polytechnic Lhokseumawe
ata_riq@mhs.ee.its.ac.id ata_riq@mhs.ee.its.ac.id
3
Jurusan Teknik Elektro, Institut Teknologi Sepuluh Nopember Surabaya 3 Department of

Electrical Engineering, Institute of Technology Surabaya
surya@ee.its.ac.id surya@ee.its.ac.id
Abstract
The agent is something that is capable of sensing its environment through sensors and react to that
environment through effectors. Multiagent system is a system whereby a number of agents perform
a number of tasks to achieve common goals. The agents in multiagent system can interact with each
other either directly (direct, through communication and negotiation) or indirectly (indirect, interact
through their environment). JADEX is an agent-oriented reasoning engine to write the relationship
between agents using XML and Java programming language. So Jadex can display a simple
approach to the relationship between the agent for some tasks. By using JADEX to simulate how
agents communicate with each other cooperatively and competitively. In this study simulate an
example case of the mining process energy sources. With the results of a comparative analysis of
the benefits of work and production expenses. Where there are several agents that cooperate with
each other and there is an agent that acts as an obstacle (barrier). From the results of this study can
be considered for a mining company to simulate those things into consideration when will perform
the mining process.
Keywords: agent, multiagent, JADEX, Java
1. Introduction
Agent technology has become one of the subject of discussion and research is much talked about in
science today. Starting from the use of robotics technology and mimic the functions of the behavior
of living things to the modeling of human social interactions in computer simulations.
The agent is something that is capable of sensing its environment through sensors and react to that
environment through effectors [1]. Basically, an agent is autonomous, since agents operate without
107
MICEEI 2010
direct intervention by humans or others and can control the action and the condition itself against its
environment. Multiagent system is a system where some agents do a number of tasks to achieve
common goals [2]. Agents in multi-agent system can interact with each other either directly (direct,
through communication and negotiation) or indirectly (indirect, interact through their environment).
In this study the development of an agent using a tool Jadex. Jadex is an agent-oriented reasoning
engine to write the relationship between agents using XML and Java programming language. So
Jadex can display a simple approach to the relationship between the agent for some tasks. By using
Jadex Control Centre (JCC) displays all runtime tools that can be accessed and used in accordance
with their respective functions.
2. Simulation Scenario
In this study we create a scenario simulation game about agents Lithium Ore miner. There are some
agents that exist in this story, such as Homebase, Alien, Sentry, Production and Carrier.
In this simulation we analyze the outline of a mining process that requires many resources, among
others, who can we describe here in simple form on the stage of the process of looking for sources
of ore, producing and transporting the result. It describes a simple simulation of how a process of
mining new energy sources (ore) which was initially all agents can be simulated for the detection of
the source of ore mines, each agent will communicate with one another to report on its search
results to the agency Sentry, which in this case This serves to determine the large amount of ore that
can be explored and opened the mining points are to be explored further by the production agent.
For agents who have the mining function will start the process of working from a base camp or
named as a home base here. Sentry agent, production, and the carrier will begin its work by the
time set by the home base. After the allotted time expires, then all the agent will return to home
base.
Figure 1 shows the process flow diagram of ore mining. We add an agent that is named Alien, this
agent has a capability similar to the Sentry, but its the opposite of the Sentry. The purpose of this
agency is to describe an obstacle or problems that may occur in the mining process is simple. This
agent has the ability to remove the sign (mark), which means the point of the mine can not be
explored again.
3. Experiment
After doing this miner agent simulation, we did some experiments by replacing some variables to
determine the optimal configuration taking into account several aspects. In this experimental
configuration we are doing just the addition of the agent (Table 1), so other factors such as the
obstacle, vision sensor, speed and capacity Carrier everything fixed.
In Table 2 indicated several variables (speed, vision, number of agents) that will be simulated in
normal conditions. Variable speed and vision are fixed, while the number of agents will vary.
Variable speed is a variable speed which is owned by each agent to be able to reach each target
operation, this variable will be different for each agent but in this simulation is that it has constant
speed variable. For the vision variables is assumed as a sensor to detect the target variable mining
108
MICEEI 2010
which is well worth the stay. Initial conditions assumed an ore valued at Rp. 5,- the number of
existing ore mining in the area totaling 500 units and exploration of the simulation time is 90
seconds. Within 90 seconds of each agent will work each according to its function and will return to
home base.
Figure 1. Mining process flow diagram

After determining the variables that are fixed and variable, then obtained a result as shown in Fig. 2.
Table 1 Assumptions Agent per unit
Agent Type
Sentry
Production
Carry
Speed
5
10
15
Vision
0.1
0:05
0:05
Price (Rp)
50
35
15
In this experiment, we conducted additional amount of carrier agent (carrier) in each section. We
divide the experiments each agent into 3 major parts, where in each section we perform the
experiment 10 times and write the result as shown in Figure 2.
It appears that in table 2 with the normal condition of all agents amounted to 1 pc comparison of
results obtained between the cost (cost) incurred by the amount of ore obtained. Of the total ore are
placed at all points of fruit ore amounting to 500 targets, the results obtained from this process, the
carrier agent can perform an average of transporting ore are 36.8% or roughly about 184 pieces ore.
These results will be compared with our subsequent experiment with the addition of different agents.
109
MICEEI 2010
Table 2 Experiment 1 (Normal Condition)
Tipe Agen
Sentry
Production
Carry
Alien
(obstacle)
Speed
Vision
Jumlah Agent
A
5
10
15
B
5
10
15
C
5
10
15
A
0.1
0.05
0.05
B
0.1
0.05
0.05
C
0.1
0.05
0.05
A
1
1
1
B
1
1
2
C
1
1
3
0.05
0.05
0.05
Figure 2: On Production Cost Comparison Graph Results
Figure 3: An example of JADEX Simulation Program

4. Conclusion
From the simulation multiagent systems we have done in this case take the example of energy
mining process simulation can be drawn a conclusion that by adding one of the fleet (agent) miners
will lead ore mining products increased, but the increase was also accompanied by an increase in
total cost of expended by the miners to fulfill the request. Comparison between the total cost
110
MICEEI 2010
incurred and the amount of ore produced significant is 1:5. With the addition of an alien agent (in
character as a barrier, obstacle) will result in decrease in the number of mining ore and this has
become one of the factors to be considered. However, if the agent is removed will be obtained a
very ideal conditions (without obstructions) and almost all sources of ore being mined successfully
collected.
Future Work:
This experiment can still be done by making changes in vision sensors, speed of each agent,
communication systems, the addition of (alien) as an obstacle and a combination of production of
each agent, carrier and Sentry to enable maximum results with minimum operational costs down
possible.
Bibliography
[1] Russell, Stuart J. and Peter Norvig. 1995. Artificial Intelligence: A Modern Approach",
Englewood Cliffs, NJ: Prentice Hall.
[2] Weiss, Gerard. Multiagent Systems,: A Modern Approach to Distributed Artificial
Intelligence.Massachusetts: Massachusetts Institute of Technology, 1999.
[3] P, Alexander and Braubach, Lars Jadex User Guide, University of Hamburg, Germany, 2007.
[4] P, Alexander and Braubach, Lars Jadex Tutorial, University of Hamburg, Germany, 2007
111
MICEEI 2010
Preliminary Study of Developing Flood Early Warning System

Zahir Zainuddin
Electrical Engineering Department, UNHAS, Makassar, Indonesia
Abstract
Flood disasters are routinely happened every year in many places in Indonesia. In 2006, one big
flood disaster occurred in Kabupaten Sinjai and killed more than 200 peoples. The flood disaster
like this came very quick without any prediction before people could make any rescue activities.
The victim of people caused by the disaster like this can be avoided if such an early warning
was implemented.
A research in the form of preliminary study and development of some prototypes in establishing
flood early warning system had been done. The system consists of some environmental
monitoring using sensors such as temperature, humidity, rainfall sensors and water flow
instruments. The concept of monitoring sensors configuration is based on wireless network
sensors. Data acquisitions are implemented in real time. Study of communications infrastructure
for wireless network sensor was implemented by evaluating long range wifi usage. The result
shows with some manufacturers equipment communications con be done normally (2MBps) for
less then 25 km. Currently, evaluation of mobile communication for this purpose is being done.
Some algorithms also had been developed to process data acquired from the monitoring system.
The data processing algorithms include water level prediction using Neuro Fuzzy algorithm.
The result from this processing is one day ahead prediction. Also developed in this system is run
off water modeling software that can simulate the area of flooding using D8 and DEM (digital
elevation modeling) algorithm. Now the development of flood early warning system is still in
progress to complete the whole system.
Keywords: Flood disaster, early warning system, wireless network sensors, real-time
monitoring, water level prediction, neuro-fuzzy, D8 and DEM
112
MICEEI 2010
Optimization of Pre-determined Object Ordering

#
Amil A. Ilham 12, Kazuaki Murakami 12
Department of Informatics, ISEE, Kyushu University

Fukuoka, Japan
2
Institute of Systems, Information Technologies and Nanotechnologies
Fukuoka, Japan, amil@isit.or.jp
1. Introduction
The automatic memory management and security advantages of Java programs have encouraged
programmers to use this language to write all kinds of applications. As the popularity of Java has been
increasing among programmers, researchers have put more their attention to improve the performance of
Java applications. This paper is aimed to optimize the order of Java objects in memory to improve the
performance of Java applications. Understanding the memory behaviours of Java is important because as an
object oriented program, Java creates many objects dynamically on a heap, accesses and mutates them during
runtime. Improper order of objects in memory might degrade the performance of Java application especially
when the order of objects in memory is not match with the way the Java program accesses the objects.
This work exploits the existence of garbage collector in Java Virtual Machine (JVM). In Java, object
deletion is done automatically by a garbage collector instead of by programmers as in C or C++. It is
claimed that garbage collector has increased the productivity of programmers by reducing programmer
burden and eliminating sources of errors [5]. Copying garbage collector provides opportunity for object
ordering optimization because it can move live objects during the garbage collection.
We implement a simple pre-determined object copying ordering scheme (Depth First) and an
extension of this scheme which we call Hot-Depth First scheme. Our experiment results show that our
extended Hot-Depth First scheme has a better impact on Java application performance compared with the
Depth First scheme.
2. Background
Generally, there are two main types of garbage collectors [4]: non-moving garbage collector and
moving garbage collector. Due to the limitation of the space, we only briefly discuss moving garbage
collector, the garbage collector we use in our experiments.
2.1 Copying Garbage Collector

An example of moving garbage collector is a semi-space garbage collector as shown in Figure 1.
This algorithm divides heap into two spaces: from-space and to-space.
to-space
to-space
to-space
from-space
from-space
from-space
Figure 1: Semi-space Garbage Collector
113
MICEEI 2010
During runtime (before collection), only the from-space is used for object allocation and the to-space
is reserved. Once the from-space is full, JVM halts the running application and invokes the garbage collector.
The garbage collector starts to find all live objects and moves them to the to-space. For example, the garbage
collector moves object A to the to-space since it has a reference from the root. Object C is also moved to the
to-space because it has a reference from A which is a live object, and so on. After all live objects are
evacuated to the to-space, JVM flips the role of the two spaces, the to-space becomes from-space and vice
versa. The running application is resumed and new objects will be allocated in the free space inside the fromspace (after collection).
Semi-space garbage collector provides opportunity for object ordering because during the
reclamation process, live objects are copied and can be placed in any order in the reserved space.
3. Object Copying Ordering Schemes

We extend the semi-space garbage collector to implement a simple pre-determined object copying
ordering scheme, Depth First, and an extension of this scheme which we call Hot-Depth First scheme. Both
these schemes copy objects based on their connectivity but the Hot-Depth First scheme takes into account
the hotness of the objects.
3.1 Depth First Scheme (DF)

The order of object copying at garbage collection time is performed based on the connectivity of the
live objects. Figure 2 shows an example of object connectivity graph. In this figure, object A is called the
root object, because there is no reference to this object. Object A has direct references to objects B, C, and D.
These objects are called object A`s children and they are siblings. Similarly, object E, F and G are siblings
and they are object B`s children.
DF scheme would copy one child immediately after copying the parent. For example, after copying
object A, DF scheme would recursively copy one of its children, say B, and the one of B`s children, say E,
and so on. The resulting object order of DF scheme in memory is shown in Figure 4.
Figure 2: An example of object

connectivity graph
Figure 4: Objet order in memory

based on DF scheme
Figure 3: An example of object connectivity

graph with hot objects
Figure 5: Objet order in memory

based on HDF scheme
114
MICEEI 2010
3.2 Hot-Depth First Scheme (HDF)

This scheme is our extension to the DF scheme. During runtime, a Java program creates many
objects, accesses and mutates the objects. We assume that there are some objects which are frequently
accessed and mutated by the program. We called these objects as hot objects. Figure 3 shows an example
of object connectivity graph with hot objects: object C, F, J and L. HDF scheme would give a priority
copying of these hot objects. For example, after copying object A, HDF scheme would copy one of its hot
children, say C, and the one of C`s hot children, say J, and so on. The resulting object order of HDF scheme
in memory is shown in Figure 5.
3.2.1 Hot Object Identification
We performed online object profiling to find hot objects created by the application before each
invocation of the garbage collector. We exploit the baseline compiler and the existing adaptive optimization
system in the used JVM to identify these hot objects.
JikesRVM [1] implements two types of compilers: baseline compiler and optimizing compiler.
JikesRVM uses the baseline compiler to compile all Java classes for the first time without optimization to
provide a fast start-up application run. The JVM has an adaptive optimization system which is used to
optimize hot methods in the application. These methods are recompiled with a certain level of optimizations
by the optimizing compiler. The corresponding un-optimized code of these methods produced by the
baseline compiler will be replaced by optimized code produced by the optimizing compiler.
Identification of hot methods is done by runtime measurement component. It installs a listener that
periodically samples the currently executing method. When it is time to take a sample, the listener inspects
the thread's call stack and records a single compiled method id into a buffer. When the buffer of samples is
full, the listener then stops taking samples and wakes the sleeping Hot Method Organizer to update the
Method Sample Data. This data structure maintains, for every compiled method, the total number of times
that it has been sampled since the beginning of execution.
We extend the baseline compiler to identify all potential hot fields in the classes. We then use and
extract information from the runtime measurement component regarding the hot methods. The potential hot
fields will be considered as hot if they access the hot methods and any created objects which contain these
hot fields will be considered as hot objects.
4. Evaluation Environment
We conducted our experiments on a single 3.20 GHz Pentium 4 with 64 byte DL1 and L2 cache line
size and 2GB of main memory. We performed the experiment on 32bit Linux 2.6.18 kernel with perfctr
patch to access the Pentium 4`s on-chip performance counters. The computer run stand alone with all
unnecessary daemons and services stopped. The network interface is also down. Our virtual machine
infrastructure was Jikes RVM 3.1.0, released on June 10, 2009.
For each experiment, we run the JVM once and run each Java application twice. We measured the
performance of Java application at the second run because Eeckhout et al. show that measurements of the
first run of a Java application inside a JVM tend to be dominated by the JVM overheads instead of by
application behaviour [3]. We setup the JVM to invoke semi-space garbage collector with DF scheme or
HDF scheme and we run 11 Java applications from SPEC [6] and DaCapo [2] benchmark suite version
2006-10-MR2. We run all applications on different heap sizes and for each heap size, we run the application
5 times and report the mean.
5. Results
We evaluate the impact of Depth First (DF) scheme and Hot Depth First (HDF) scheme on execution
time, number of cache misses and DTLB misses of the running Java applications and report the results for
heap size 1.5x minimum heap size.
115
MICEEI 2010
Generally, considering the hotness of objects in reordering Java objects in memory improves Java
application performance. Figure 6 shows HDF scheme has a better impact on java application execution time
compare to DF scheme. It improves execution time around 3-9% for all applications. These performance
numbers includes the overheads of finding the hot objects and reordering them during the garbage collector.
As expected, memory and DTLB performance are also improved. L1D cache and L2 cache misses are
improved by HDF from 4 to 11% over DF as shown in Figure 7 and 8 respectively. DTLB gains more
improvement compare to L1D cache and L2 cache by HDF. Figure 9 shows DTLB misses are improved by
5-12% for all applications.
Figure 6: Execution time
Figure 8: L2 cache misses
Figure 7: L1D cache misses
Figure 9: DTLB misses
6. Conclusions
Our extension to the Depth First (DF) object ordering scheme which we call Hot-Depth First (HDF)
scheme has generally improved the performance of Java applications. By giving a priority copying for hot
objects as we implemented in the HDF scheme has reduced cache and DTLB misses and improved Java
application execution time. However the performance gain is still moderate due the overheads of finding the
hot objects.
References
[1] Alpern, B., Augart, S., Blackburn, S., Butrico, M., Cocchi, A., Cheng, P., Dolby, J., Fink, S., Grove, D.,
Hind, M., McKinley, K., Mergen, M., Moss, J., Ngo, T., Sarkar, V., and Trapp, M., The Jikes Research
116
MICEEI 2010
Virtual Machine Project: Buliding an Open-source Research Community, IBM Systems Journal 44, 2, pp.
399417, 2005.
[2] Blackburn, S. M., et al., The DaCapo benchmarks: Java benchmarking development and analysis. In
OOPSLA, 2006.
[3] Eeckhout, L., Georges, A., and De Bosschere, K., How Java programs interact with virtual machines at
the microarchitectural level, In OOPSLA, pp. 169-186, 2003.
[4] Jones, R. E., Lins, R. D., Garbage collection: algorithms for automatic dynamic memory management,
Wiley, Chichester, 1996.
[5] McCarthy, J., History of LISP, In ACM History of programming language 1, pp. 173-185, 1981
[6] Standard Performance Evaluation Corporation. http://www.spec.org/benchmarks.html#java
117
MICEEI 2010
118
MICEEI 2010
119
MICEEI 2010
120
MICEEI 2010
121
MICEEI 2010
122
MICEEI 2010
123
MICEEI 2010
124
MICEEI 2010
125
MICEEI 2010
126
MICEEI 2010
127
MICEEI 2010
128
MICEEI 2010
129
MICEEI 2010
130
MICEEI 2010
131
MICEEI 2010
132
MICEEI 2010
133
MICEEI 2010
134
MICEEI 2010
135
MICEEI 2010
136
MICEEI 2010
137
MICEEI 2010
138
MICEEI 2010
139
MICEEI 2010
140
MICEEI 2010
141
MICEEI 2010
142
MICEEI 2010
143

ProceedingsMICEEI2010 PartA

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

ProceedingsMICEEI2010 PartA

Diunggah oleh

Hak Cipta:

Format Tersedia

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

Copyright 2010 by MICEEI 2010, Department of Electrical Engineering, Faculty of

Reproduction or translation of any part of this work

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

International Advisory Committee:

International Program Committee:

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

R.M.A.P Rajatheva, Asian Institute of Technology (AIT), Thailand

Local Organizing Committee:

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

THE 2nd MAKASSAR INTERNATIONAL CONFERENCE ON ELECTRICAL

OCTOBER 27, WEDNESDAY, MGH GOLDEN BALL ROOM

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

OCTOBER 28, THURSDAY

Santosh Shedge and Anup Kale

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

CEIE1:Control, Electronics and Instrumentation Engineering (1)

Dynamixel Motor-based Actuator For Mobile Robot Design

Multi Agent System Modeling Using JADEX

CIE1: Computer and Informatics Engineering (1)

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

Book Loan Recommendation System for Petra Christian

PE1:Power Engineering (1)

Transient Stability Analysis of Java-Bali 500KV

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

Elliptical Annular Ring Microstrip Antenna for Circularly

Analysis of Sea Surface Temperature using MODIS Data

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

CIE2a: Computer and Informatics Engineering (2a)

The Design and Development of Facebook Based Agenda

CIE2b: Computer and Informatics Engineering (2b)

The Ramsey Numbers for Odd Complete Bipartite Graphs

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

He received his B. E., M. S. and Ph. D. degree all in electrical

Intelligent System Applications to Power System

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

Prof. Dadang Gunawan

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

Professor Mary-Anne Williams

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

Prof. Eko Tjipto Rahardjo

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

Dr. Linus Liu

Computation Electromagnetic for Antenna applications

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

Ir. Syamsul Aliah, MSEE

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

The Application of Wavelet Transform for an Image Compression

II. Wavelet Decomposition

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND

The wavelet transform is defined as

of a mother wavelet x can be used as an orthonormal basis for the multiresolution

where J is maximum level of decomposition.

BRING THE FUTURE DEVELOPMENT OF SCIENCE & TECHNOLOGY ON YOUR HAND