Abstract Proceedings of IEEE SCAP 2013

95
The 2013 IEEE Student Conference on Senior Capstone Project IEEE SCAP 2013 Faculty of Engineering Chulalongkorn University Bangkok, Thailand March 29, 2013

description

Abstract Proceedings of IEEE SCAP 2013 @ Chula Engineering, March 29th, 2013.

Transcript of Abstract Proceedings of IEEE SCAP 2013

Page 1: Abstract Proceedings of IEEE SCAP 2013

The 2013 IEEE Student Conference on Senior Capstone Project

IEEE SCAP 2013 Faculty of Engineering

Chulalongkorn University

Bangkok, Thailand

March 29, 2013

Page 2: Abstract Proceedings of IEEE SCAP 2013

ii

The 2013 IEEE Student Conference on Senior Capstone Project

Organizing committee

Advisory Committees

Boonsom Lerdhirunwong

David Banjerdpongchai

Boonchai Sowanwanichakul

Parames Chutima

Witaya Wannasuphoprasit

Ekachai Leelarasmee

Arthon Sinsawasdi

General Chair

Thavatchai Tayjasanant

Technical Program Chairs

Panuwat Janpugdee

Peerapon Vateekul

Technical Program Committees

Chaodit Aswakul

Nakornthip Prompoon

Pizzanu Kanongchaiyos

Kuntinee Maneeratana

Angkee Sripakagorn

Boonchuay Supmonchai

Arporn Teeramongkonrasmee

Charnchai Pluempitiwiriyawej

Haruetai Lohasiriwat

Wipawee Tharmmaphornphilas

Local Arrangement Chair

Jitkomut Songsiri

Secretary

Amornrat Noicharoen

Webmaster

Apinun Intarachaiya

Page 3: Abstract Proceedings of IEEE SCAP 2013

iii

The 2013 IEEE Student Conference on Senior Capstone Project

Keynote speakers

Title : “Scientists investigate what already is. Engineers create what has never been”

Ekachai Leelarasmee, Chulalongkorn University

Abstracts

Engineering is a field which applies scientific knowledge to solve real world problems or

designing products. This talk is meant to inspire young engineers by showing a video clip

―IEEE: Engineer Inspiring‖ as well as introducing some creative projects from

www.kickstarter.com such as Pebble Watch, 3D Printer, Hidden Speaker and iOS controlled

camera. Then two development projects done at the Embedded System and Integrated Circuit

Design Laboratory (ESID) at Chulalongkorn University will be described. They are Smart

Meter and Remote Solar Farm Monitoring. These two projects demonstrate how electronics

engineers apply their electrical knowledge to help the country fighting the energy problem. Biography Ekachai Leelarasmee finished his undergraduate study in electrical engineering from

Chulalongkorn University in 1974 and went on to do his Ph.D at University of California,

Berkeley USA in 1982. He is now an associate professor and deputy chair of student

activities in the Department of Electrical Engineering, Chulalongkorn University. He was the

IEEE Thailand Section Chair in 2010-11 and is the interim chair of the IEEE Thailand Solid

State Circuit Chapter 2013.

Page 4: Abstract Proceedings of IEEE SCAP 2013

iv

The 2013 IEEE Student Conference on Senior Capstone Project

Table of Contents

Organizing Committees ......................................................................................................... ii

Keynote Speakers ................................................................................................................. iii

Section 1 Sustainable Technology

Remote Agricultural Data Management System for Royal Project Foundation .............................................. 2 Tuchchai Apinuntakulchai, Apinya Phanumphai, Surapont Toomnark, Thagorn Tangmankhong,

Kraikron Settakraikun (King Mongkut's University of Technology Thonburi)

Online shopping mall ................................................................................................................................... 3 Parita Doungngam, Chachanok Somchit, Artit maneesri, Sirapat Chiewchanwattana

(Khon Kaen University)

Comparison of indoor SLAM using ceiling images with different state representations ................................. 4 Piyapat Saranrittichai, Nattee Niparnan, Attawith Sudsang (Chulalongkorn University)

Fabrication of Micropore on Silicon substrate ............................................................................................... 5 Krairat Mairin, Mana Sriyudthsak (Chulalongkorn University)

Development of an Embedded BLDC motor controller using RS485 standard ............................................... 6 Teeratath Ariyachartphadungkit, Kanjanapan Sukvichai (Kasetsart University)

An AU Golf Car: Autonomous Parking System ............................................................................................ 7 Hemmawat Yensuong, Patompong Sulsaksakul, Passaporn Rattanasetyith, Pattanan Detkrut, Pasawat

Laokhonkha, Narong Aphiratsakun (Assumption University)

Improved Material Removal Rate by Abrasive Water Jet Milling .................................................................. 8

Kasidet Wangvithayakul, Ukrit Jampachon, and Sataphol Wattnapornmongkol, Pairat Tangpornprasert

(Chulalongkorn University)

Section 2 Science and Technology

Disruptions to Saccade Planning: Effects on Eye Movements ..................................................................... 10

YueYinglong, Yen Shih-Cheng, Chua fook kee, Shih-Cheng (National University of Singapore)

Parallel Additive Implementation for modified Flexible Interval Representation System ............................. 11

Phoonlarp Mekraksakit, Thanisorn Thanapongsapak, Chawin Wongchindakhun, Athasit Surarerks

(Chulalongkorn University)

A Quadrature Current Controlled Oscillator with 250 – 700 MHz Tuning Range ........................................ 12

Radit Smunyahirun, Ekachai Leelarasmee (Chulalongkorn University)

Real Time People Tracking and Collision Avoidance using Sensors Fusion for an Indoor Omni-directional Wheels Mobile Robot ................................................................................................................................ 13

Bhirawich Pholpoke, Kanjanapan Sukvichai (Kasetsart University)

Design of a Phasor Measurement Unit (PMU) using FPGA ........................................................................ 14

Porjed Sakunjareanpornchai, Thirawut Fueangfu (King Mongkut's University of Technology Thonburi)

Fire Infiltrating Robot with Image Acquisition: A Microcontroller based Urban Search and Rescue

(USAR) Robot .......................................................................................................................................... 15

Ruth A. Silvallana, Russel L. Amado, Kristoni R. Aquino, Carmen Fe C. de Guzman, Moreno,

Kendrick Kent L. (Technological Institute of the Philippines)

A New Algebraically Simple Chaotic Jerk Circuit and System using Memristor Nonlinearity ..................... 16

Jirayu Peetakul, Sawitee Wangthawal, Wimol Sanum (ThaiNichi Institute of Technology)

An AU Golf Car: Obstacles Avoidance System .......................................................................................... 17 Ankur Agrawal, Krittamate Jirundorn, Tidarat Punyachai, and Samuel Hussain Mogakolodi,

Narong Aphiratsakun (Assumption University)

Stability Control of a Self-Driving Bicycle ................................................................................................. 18

Vittavat Kaewmaneekul, Sompong Jarujitjumlern, Whicha Thaitavon, Viboon Sangveraphunsiri

(Chulalongkorn University)

Investigating on Damping Property of Laminated Composite Materials ...................................................... 19

Thanyarat Singhanart, Kritsadaporn Thongsawang, Nattapon Kaewchalam, Nutthasan Srikwanma,

Worapol Phraechinda (Chulalongkorn University)

Section 3 Health and Medicine

Design of an appointment system and bed management for Chemotherapy department ............................... 21

Thanakrit Thanathipanont, Achiraya Eurakarawong, Wipawee Tharmmaphornphilas

(Chulalongkorn University)

Individual Speech Synthesis for ChulaDAISY ............................................................................................ 22

Page 5: Abstract Proceedings of IEEE SCAP 2013

v

The 2013 IEEE Student Conference on Senior Capstone Project

Sivaporn Homvanish, Thiti Watansrimongkol, Natchanon Phachongkitphiphat, Proadpran

P.Punyabukkana, Atiwong Suchato (Chulalongkorn University)

Real-Time Gait Analysis Using Kinect-based Motion Capture System ........................................................ 23

Pattaravut Maleehuan, Pizzanu Kanongchaiyos (Chulalongkorn University)

Gait Phase Detection using Wireless Smart Shoe with Optimized Threshold by Genetic Algorithm ............ 24

Nantawat Pinkam, Itthisek Nilkhamhang (Thammasat University) Ultrasound B-scans Image Denoising via Expectation Maximization-based Unsharp Masking ................... 25

Theerawit Wilaiprasitporn, Chedsada Chinrungrueng, Widhyakorn Asdornwised

(Chulalongkorn University)

An Automatic System for Measuring Electrode-Skin Impedance ................................................................ 26

Prawsiri Veeranarapanich, Arporn Teeramongkonramee, Apiwat Lek-uthai (Chulalongkorn University)

Computer Application using ARIMA and Simulation techniques for managing inventory in hospital .......... 27

Pattranit Khobkun, Watsa Tantiyuth, Wassachon Veerakul, Surapong Sirikulvadhana

(Chulalongkorn University)

Admission and Discharge Processes Improvement...................................................................................... 28

Nopparuth lurkittikul, Tanawut Supparerkapa, Tanawat Jarusiripipat, Wipawee Tharmmaphornphilas

(Chulalongkorn University)

Design and Manufacture Argumentative Mobility Device For Osteoarthritis patients in Thailand ................ 29 Vorapraj Chutintaranond, Wongsatorn Sathitsuksanoh, Pitchaya Sovachinda, Teerinth Changsawang,

Pairat Tangpornprasert, Chanyaphan Virulsri (Chulalongkorn University)

Second Ear: A Hearing Tool for People with Hearing Disability ................................................................. 30

Pattara Sukprasert, Nuttapong Phaisarn, Atiwong Suchato, Proadpran Punyabukkana

(Chulalongkorn University)

Design and Implementation of Surveillance Robot for Disabilities and Elderly ........................................... 31

Surainan Kahmimah, Jakapan Suaboot, Apichat Heednacram (Prince of Songkla University)

Learning Causal Structures of Brain Connectivity in fMRI time series ........................................................ 32

Arnan Pongrattanakul, Puttichai Lertkultanon, Jitkomut Songsiri (Chulalongkorn University)

Patient and treatment scheduling in Dermatological Outpatient Department queuing system ....................... 33

Hataipat Juthasri, Suwat Tansiriyakul, Oran Kittithreerapronchai (Chulalongkorn University)

Section 4 Network and Communications

An Alarm System via Social Network Services .......................................................................................... 35

Theeradet Khamperawat, Gorarit Chandhanimi, Korporn Panyim (Mahidol University)

An Improvement of 3G Network Coverage with Picocell in Mahidol University ......................................... 36

Sanhapit Phatratipakorn, Thanate Srieakpanit, Korporn Panyim (Mahidol University)

Performance Improvement of Wi-Fi Network in Faculty of Engineering, Chulalongkorn University ........... 37

Nipoon Sintoowong, Nonthawat Ungtrakul, Thanach Chotruangprasert, Chaodit Aswakul,

Chaya Limchitti (Chulalongkorn University) Microwave Sensor with Artificial Neural Network Application .................................................................. 38

Narong Borijindargoon, Nuttapon Muangmor, Chuwong Phongcharoenpanich, Sorawat Chivapreecha

(King Mongkut's Institute of Technology Ladkrabang)

Development of Radio Resource Planning Program for 2G and 3G ............................................................. 39

Jiradchaya Kanjanakunchorn, Jidapa Hansawangkit, Pongsatorn Sedtheetorn (Mahidol University)

Communication Emulation using DMSL/COSEM on PC............................................................................ 40

Raivin Srisamrit, Wanchalerm Pora (Chulalongkorn University)

Service Quality Improvement in Telecommunication Service Encounter ..................................................... 41

Konlawat Luengrat, Kritsana Settapittayakul, Natcha Thawesaengskulthai (Chulalongkorn University)

Target Location Searching Application using Ad-hoc Network ................................................................... 42

Kornkanok Khaoampai, Kamonluk Suksen, Kultida Rojviboonchai (Chulalongkorn University) Traffic sensing application using ad-hoc network ....................................................................................... 43

Korakot Sirinam, Krittin Intharawijitr, Kultida Rojviboonchai (Chulalongkorn University)

A Centralized Online Network-based Intrusion Prevention System Control via Web Application ................ 44

Ekgapark Wonghirunsombat , Teewalee Asawaniwed, Vassapon Hanchana, Naruemon

Wattanapongsakorn (King Mongkut‘s University of Technology Thonburi)

Voice On Demand Application over Ad-hoc Network ................................................................................ 45

Jittapon Busarakum, Kultida Rojviboonchai (Chulalongkorn University)

A Study on Chaos in Digital Filter and Its Application to Secure Communication ....................................... 46

RattaponThassanasateankit, WittayaJantaramatsakarn, WisarutPhompung, SorawatChivapreecha

(King Mongkut's Institute of Technology Ladkrabang)

Page 6: Abstract Proceedings of IEEE SCAP 2013

vi

The 2013 IEEE Student Conference on Senior Capstone Project

Section 5 Computer and Information Technology

Data Partitioning to Improve Similarity Join by Prefix Filtering Algorithm ................................................. 48 Methus Bhirakit, Jaruloj Chongstitvatana (Chulalongkom University)

Automatic Screening Angle-Closure Glaucoma using Genetic Algorithm and Genetic Kernel SVM ........... 49

Paopat Ratpunpairoj, Waree Kongprawechnon, Pished Bunnun, Toshiaki Kondo (Thammasat University)

Application of Combinatorial Optimization with Coincidence Algorithm for Flow shop Scheduling

problem ..................................................................................................................................................... 50

Ornrumpha Srimongkolkul, Prabhas Chongstitvatana (Chulalongkorn University)

3D Game Framework for Android Platform................................................................................................ 51

Wirune Kaewjai, Thiti Rungcharoenpaisal (Dhurakij Pundit University)

Development of Ground Planar Segmentation algorithm using 3D Point Clouds Information from

Kenect Depth Image Camera ...................................................................................................................... 52

Thanarat Chaveekolakit, Kanjanapan Sukvichai (Kasetsart University)

Adaptive strategic decision point extraction from Influence maps in games ................................................. 53 Namo Podee, Vishnu Kotrajaras, Boonserm Kijsirikul (Chulalongkorn University)

Tower Defence Game Project ..................................................................................................................... 54

Nattapong Rattasamut, Weerapong Siriyindee, Natasha Dejdumrong

(King Mongkut's University of Technology Thonburi)

Application for Publicity of promotional products online. ........................................................................... 55

Natthaporn Singmanee, Chadaporn Chiansri, Sirapat Chiewchanwattana (Khon Kaen University)

Study on a National SMS HUB Solution for Bulk SMS and National Policies on SMS Regulator in

Sri Lanka ................................................................................................................................................... 56

M.W. P. Maduranga , M. Gamage, Manodha Gamage (Asian Institute of Technology)

Indoor Environment 3D-Image Reconstruction using Microsoft Kinect for Rescue Robot ........................... 57

Piyaphat Phukphan, Vasitphon Pawankiattikun, Toshiaki Kondo, Itthisek Nilkhamhang (Thammasat University)

Multiple-CCTV monitoring system for iOS-based mobile device ................................................................ 58

Rakneung Pichitwitaya-ake, Supavadee Aramvith (Chulalongkorn University)

Development of a real-time object tracking camera..................................................................................... 59

Krittin Pachtrachai, Toshiaki Kondo, Toshiaki Kondo (Thammasat University)

Development of Unification Filesystem for Cloud Storage.......................................................................... 60

Worachate Apichanukul, Thanawat Suknim, Krerk Piromsopa (Chulalongkorn University)

Personalized Assessment, Review and Learning System Project ................................................................. 61

Tawan Junhom, Sawaruk Sirimungklanurak, Siriwat Saiwiwat, Natasha Dejdumrong

(King Mongkut's University of Technology Thonburi)

Proximity Information Exchange based on NFC ......................................................................................... 62

Boripach Khuhaprema, Win Eiwwongcharoen, Supachai Vorapojpisut (Thammasat University)

Section 6 Sufficiency Economy

Markerless Fruit Augmented Reality .......................................................................................................... 64

Songpon Sastrawaha, Sirinad Kapoung, Kumron Sunat (Khon Kaen University)

Design and Implementation of Artificial Prediction Markets Using Logarithmic Market Scoring Rules ...... 65

Pannate Jongpanichkultorn, Prabhas Chongstitvattana, Daricha Sutivong (Chulalongkorn University)

Internal Rate of Return for Probabilistic Cash Flows Project ....................................................................... 66

Phatharaphan Chamnannarongsak, Daricha Sutivong (Chulalongkorn University) Factors Analysis of Bayesian-based Pricing Prediction Markets .................................................................. 67

Nattanont Olarnthatchanant, Daricha Sutivong (Chulalongkorn University)

Optimizing the Liquidity Parameter of Logarithmic Market Scoring Rules Prediction Markets .................... 68

Suparerk Lekwijit, Daricha Sutivong (Chulalongkorn University)

A Comparison of Moving Average and Support Vector Machine for predicting SET index ........................ 69

Poom Saowalukpan, Tanpitcha Chattrakoonpong, Oran Kittithreerapronchai (Chulalongkorn University)

Thai Sign Language Recognition using Data Gloves and Motion Trackers .................................................. 70

Ditsarin Vongasemjit, Jenwit Amonpongitsara, Chotirat Ratanamahatana (Chulalongkorn University)

Vision-Based Thai Sign Language Recognition .......................................................................................... 71

Jakapan Narkbuakaew, Chanaporn Pikulnarkwong, Chawintorn Pathomkasikul,

Chotirat Ratanamahatana (Chulalongkorn University)

An Improved Traffic Incident Detection System Using Fuzzy Logic ........................................................... 72 Jaraspat La-inchua, Sorawat Chivapreecha, Suttipong Thajchayapong

(King Mongkut‘s Institute of Technology Ladkrabang , NECTEC)

Page 7: Abstract Proceedings of IEEE SCAP 2013

vii

The 2013 IEEE Student Conference on Senior Capstone Project

Load Consolidation Management System ................................................................................................... 73

Ratsuda Amatyakul, Manavut Sunthonpaobvong, Rapeepat Sukprasert, Sukritta Viryasiri, Seeronk

Prichanont (Chulalongkorn University)

Design, Analysis and Testing of Semi-Monocoque Frame for Student Formula ........................................... 74

Kulanun Chutisemachai, Kasemphan Siriployngam, Kiatnathee Dilokthonsakun, Jintasarn Sanchai,

Thanyarat Singhanart (Chulalongkorn University)

Section 7 Energy and Transportation

Design of a Pole-Slipping Protection for Loss of Synchronizing of a Generator........................................... 76

Supachai Sabua, Veerayut Pojui (King Mongkut's University of Technology Thonburi)

Mathematical modeling of radiant cooling room with radiation ................................................................... 77

Jarudon Pattarabancha, Jetarin Chareonpornpoj, Kuntinee Maneeratana (Chulalongkorn University)

Line current time shifting behavior when fault occurred on branch circuit of radial distribution system by

using ATP program .................................................................................................................................... 78

Swanya Sittiprasertphol, Weerapun Rungseevijitprapa (Chulalongkorn University) Analysis of Voltage Fluctuation based on Thai Utility Regulation ............................................................... 79

Paranee Chailungka, Thavatchai Tayjasanant (Chulalongkorn University)

Reducing Electricity Costs in the Tire Manufacturing by Load Shifting Technique and Using High

Efficiency Motors ...................................................................................................................................... 80

Aleena Sakulsanchartthai, Hadsakoon Boriphonmongkol (Chulalongkorn University)

Design and Construction of a Test Rig for the Cable-Pulley Power Transmission System ........................... 81

Pao Wichayaketsophorn, Poompat Pornwissanukul, Songkran Taepaisitpong, Suksit Wongwanich,

Phongsaen Pitakwatchara (Chulalongkorn University)

Design of a stator ground fault protection for a generator directly connected to a busbar ............................. 82

Sathichai Padungsiripanich, Kailerk Ruangsakul

(King Mongkut‘s University of Technology North Bangkok) Feasibility Study and Human Machine Interface (HMI) for 'Automatic Control and Monitoring of

a Micro Hydro Power Plant' ....................................................................................................................... 83

Isuru Mudannayake, Anuradha Colombathanthri, Erik Bohez (Asian Institute of Technology)

Analysis of Electric Power Quantities under Sinusoidal, Nonsinusoidal, Balanced or Unbalanced

Conditions ................................................................................................................................................. 84

Phonlawat Wannasiwaporn, Thavatchai Tayjasanant (Chulalongkorn University)

Impact of the Distributed Generator to Protection on Distribution Power System ........................................ 85

Tanyaluck Anusin, Hadsakoon Boriphonmongkol (Chulalongkorn University)

Application of Phasor measurement Unit for Power System Voltage Stability Monitoring ........................... 86

Puriwat Suttitham, Naebboon Hoonchareon (Chulalongkorn University)

Forecasting and Hedging Natural Gas Price for Hedging Electricity Price in Thailand ................................. 87

Sanyapong Petchrompo,Chonawee Supatgiat, Surapong Sirikulvadhana (Chulalongkorn University) A New Single Solar Tracking System for Concentrating Linear Fresnel Reflectors ..................................... 88

Nopharut Laopornpichayanuwat, Sunhaporn Sribanditmongkol, Sukrit Chuntaravisuid, Supachart

Chungpaibulpatana (Thammasat University)

Page 8: Abstract Proceedings of IEEE SCAP 2013

Section 1

Sustainable Technology

Page 9: Abstract Proceedings of IEEE SCAP 2013

2

The 2013 IEEE Student Conference on Senior Capstone Project

Remote Agricultural Data Management System for Royal Project Foundation

Tuchchai Apinuntakulchai and Apinya Phanumphai

Advisor: Surapont Toomnark, Thagorn Tangmankhong and Kraikron Settakraikun

King Mongkut‘s University of Technology Thonburi, Bangkok, Thailand,

e-mail: [email protected]

Royal Project Foundation has many local agricultural research stations at highland area such

as Doi Inthanon, Doi Ang Klang etc. The main activity of each research station is to develop

winter plants experiment and also plant processing. There are many environment parameters

required to support the research. Because of highland area, the limited of electrical power and

communication facility effect to data management for research. To solve that problem, a

model of system support to agricultural data management is presented.

Model of system at each research station is set as head and child node. Each local site has

many small plants that produce different products. Each small plant is denoted as ―Site‖. This

system is designed in the hierarchical communication level. The top level consists of the

required information for all local sites of the Royal Project Foundation that we called ―Main

server‖. The lowest level is the ―Head node‖ of each site or ―Child node‖ for the case that

each site requires more than one node.

This research has 2 main parts, the data gathering and data center. The data gathering is an

embedded system that transmits information to the data center. The data center is a system

that provides information for user by web application. The embedded system has 2 functions,

the monitoring system for maintenance and the data logger. The monitoring system is a

tracking system for preventive maintenance in the rural area.

The data logger is a meteorological embedded system which consist sensors and

communication module. This data logger called ―Head node‖ is installed at a site. Some site

requires more than one data gathering device due to the different data. We call those other

devices are ―Child node‖. Therefore, a site has more than one node, it is necessary to have

only Head node. The Head nodes were connected to the local network for transmit data to the

main server. Head node from each site can send data to other Head nodes using packet radio

when Head node has no internet or network connection. At the Main server, there are 2

modes as online and offline modes. The online mode is the normal mode that the

communication between main server and lower level device is complete. The offline is the

opposite way. There is a server that located at the main office of Royal Project Foundation. It

will gather all information from local site Head nodes in form of database.

Finally, this project will manage real time local data from the sites as useful information for

further operations. We also develop embedded data logger system which appropriate for rural

environment. The lack of data communication also solve by package radio technique. The

user friendly web applications also support to researchers and office staffs.

Page 10: Abstract Proceedings of IEEE SCAP 2013

3

The 2013 IEEE Student Conference on Senior Capstone Project

Online shopping mall

Parita Doungngam, Chachanok Somchit, and Artit maneesri

Advisor: Sirapat Chiewchanwattana

Khon Kaen University, Khon Kaenn, Thailand.

e-mail [email protected] , [email protected] and [email protected]

Online shopping mall is an android application which is developed as an alternative of the

user. The application helps the market promotion. The system is divided into 2 parts which

are the web application and the mobile application. The web application, the main function of

this part used in communication between users and providers which also helping in sell

promotion of each online shop. The information about sell promotion which stored in

database center will be called by web service and display via web browser.

So it can display both in web browser from desktop or PC and the mobile application.

Another part is the mobile application. Facebook fanpage is another channel for shop

promoting. When the store owner has created a store webpage which working on PC, the

facebook fan page will be included. If the store is an android, there are 2 choices; use or not

use facebook fanpage. The user can download Online Shopping Mall application to install in

heir own mobile phone and browse for interested products. It is suitable for users who prefer

using internet from mobile device.

Which the online shopping malls will help improve the quality of life is better. Because

improving the quality of a better life, improve the accord of life and compatible with the

changes in the future. So it matches our project, Emphasize the use of technology is

becoming popular for example the facebook is used facebook fanpage and android on mobile

application help develop. Which modernize and increase the convenience to the user.

Page 11: Abstract Proceedings of IEEE SCAP 2013

4

The 2013 IEEE Student Conference on Senior Capstone Project

Comparison of indoor SLAM using ceiling images with different state representations

Piyapat Saranrittichai

Advisor: Nattee Niparnan, Attawith Sudsang

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected],

[email protected], [email protected]

In Robotics, Simultaneous Localization and Mapping (SLAM) is the problem whose

objective is to simultaneously estimate robot configuration and map of its environment from

sensor data. It is widely used in autonomous mobile robot systems to enable robust

navigation in given environment. Currently, many research works are conducted to solve

SLAM in several environment conditions. This work focuses on crowded indoor setting

where grid map SLAM methods using traditional laser range finder do not work properly. We

use fisheye camera. The camera is installed to look upward at the ceiling in order to perform

mapping of ceiling‘s visual features. This approach avoids the occlusion issues from the

crowd and other moving obstacles in the plane of navigation. Harris corner features is

extracted due to its ease of tracking using Lucas-Kanade optical flow.

To solve indoor SLAM, location‘s state vector is typically represented as a three-dimensional

state vector of the robot‘s position (x, y) and heading direction (θ). This is sufficient for a

robot using features on the two dimensional plane of navigation. Our case is different; our

method observes ceiling visual features. It appears that the fish-eye camera is not always

looking straight up. Vibration during robot‘s motion may cause slight variation of the

viewing angle. Robot at the same location may observe different ceiling images. As a result,

we propose to compare SLAM using EKF-SLAM with different state representations, i.e., the

three dimensional location‘s state vector of [x, y, θ] T and the six dimensions locational

state vector of [x, y, z, α, β, γ] T where z is the height of the camera from the floor and α, β

and γ are yaw, pitch and roll angle of the camera respectively.

The experiment is conducted using a physical robot in the real world. The accuracy of the

maps is calculated by performing scaling Iterative Closest Point (SICP) between the resulting

maps and the ground truth map. The result clearly shows that SLAM with six dimensional

location‘s state vector is more accurate.

Page 12: Abstract Proceedings of IEEE SCAP 2013

5

The 2013 IEEE Student Conference on Senior Capstone Project

Fabrication of Micropore on Silicon substrate

Krairat Mairin

Advisor: Mana Sriyudthsak

Chulalongkorn University, Bangkok 10330, Thailand, e-mail: [email protected]

Recently, Micro-Electro Mechanical Systems (MEMs) has involved in medical technology in

many aspects. Since MEMs has a small size, so there are many benefits of that. Example of

MEMs benefits are cost reduction, mass production, quantum effects and high potential to be

implantabled devices.

This project is about micropore fabrication on silicon wafer using photolithography process

and wet chemical etching. In this project the anisotropic etchant, KOH, is used to form the

pore. The procedure of this project is divided to 3 main steps; preparation, pore-fabrication

and evaluation and discussion. The first step is mainly on theoretical design and calculation,

thickness of silicon oxide layer from thermal oxidation as a mask was estimated according to

the the KOH-Silicon etching rate. Mask diameter of 650 μm was estimatated for the designed

pore size of 150 μm. For the pore formation, the process starting from cleaning the silicon

wafer, growing the SiO2 on silicon wafer, transferring the pattern to oxide layer and etching

the silicon by KOH (20%) at 80๐C. Scanning electron microscope (SEM ) was use to

observed and measure the pore size.

The result of this project was a micropore on silicon wafer with a diameter of 200 μm. It is

believed that the diameter of pore size can be formed smaller than this by using silicon nitride

or gold as a mask for KOH etching, because of their selectivity is better than Si/SiO2 in KOH

etching.

If the micropore can be repeatedly fabricated. It would have a lot of applications waiting for

it. For example, micropore can use to measure the deformability of a human RBC (red blood

cell) which can be used to study the factors that affect the deformability of a human RBC.

Page 13: Abstract Proceedings of IEEE SCAP 2013

6

The 2013 IEEE Student Conference on Senior Capstone Project

Development of an Embedded BLDC motor controller using RS485 standard

Teeratath Ariyachartphadungkit

Advisor: Kanjanapan Sukvichai

Kasetsart University, Bangkok, Thailand, e-mail: [email protected]

The development of robotic is more advanced in previous 50 years in many industries. The

most important component of the robot is an actuator. Robot can achieve very sophisticate

task by very precise joint motion. In this research, the brushless DC motor (BLDC) driver

controller is focused in order to make the motor has superior performance. BLDC motor is

widely used in industrial robot arm and other manufacturing machine because it has more

efficiency than regular DC or stepper motor. The main problem in order to use this kind of

motor is the driver circuit and control since this motor must be driven at precise phases and

sequences. Low cost embedded BLDC motor driver system is developed in this research.

ARM Cortex M3 from STMicroelectronics is used as the main controller. The FreeRTOS is

also implemented to the system. RS485 is used as the standard communication between Host

PC and the controller board. By using this technique, the controller can be easily replaced the

old stepper or DC servo motor controller in the production lines. Proportional-Integral-

Derivative (PID) speed and torque control laws are implemented as a built-in function in the

embedded system. All PID parameters can be set via RS485 protocol.

The embedded system in this research can be separated into 3 sub circuits. First is the BLDC

driver using Field Effect Transistor (FET). There are 3 phase coils in BLDC motor; thus, the

driving circuit must contain 6 FETs. Normally, there are two types of FETs must be used to

drive BLDC which are n-type and p-type but p-type is expensive and can be only use in

present of small driving current. In this paper, 6 n-type FETs are selected associated with

charge-pump and bootstrap to increase driving current. Bootstrap circuit will boot up the bias

voltage of source pin at n-type FETs. The second circuit is the controller circuit. Hall sensors

are attached to the BLDC motor in order to obtain phases. These hall sensors are not only use

to drive the motor coils at correct sequence but also they will be used to calculate the speed of

the motor. Hall sensors will be read by microcontroller and used in PID-Torque control

algorithm. The third part is the RS485 circuit. This communication is half duplex

communication. Each driving has its own id which can be set by initial setup. The command

packages are sent to the embedded system and it must response in specific of time. Moreover,

the microcontroller must handle communication packages and perform precise motor control

at the same time. Therefore, the FreeRTOS is implemented to this driver system in order to

schedule the sequence of tasks to achieve the hard real-time requirement.

Page 14: Abstract Proceedings of IEEE SCAP 2013

7

The 2013 IEEE Student Conference on Senior Capstone Project

An AU Golf Car: Autonomous Parking System

Hemmawat Yensuong*, Patompong Sulsaksakul, Passaporn Rattanasetyith,

Pattanan Detkrut and Pasawat Laokhonkha

Advisor: Narong Aphiratsakun

Assumption University, Bangkok, Thailand, e-mail: [email protected]

This paper presents part of the research work carries out at Assumption University to develop

technology to control and route the motor vehicle. Our objective is to endow the vehicle with

an Autonomous Parking System, which is a self-parking technology that assists motor vehicle

in parallel parking. The system is developed to be completely autonomous: steering, pedal,

brake and gear. The driver will no longer need to assist the vehicle by regulating the speed

like most systems on the market.

In the self-parking operation, the vehicle moves forward into the position beside the front

vehicle and the system will notify the driver when to stop if the parking space is large

enough. The driver then engages the system and the vehicle will goes into reverse mode

automatically and will move backward slowly. At the same instance, the steering system will

autonomously maneuver the vehicle seamlessly into the available parking space. The

microcontroller FIO board based on STM32 platform and dsPIC30F4011 are fundamental

elements in the operation of self-parking to the extent that they are the main processor that

monitors, manage, and control the entire system and operations. The FIO board will act as the

main controller and two dsPIC30F4011 will perform as joint controllers of the system. This

system utilizes ultrasonic sensors located at rear quarter panel and rear bumper of the vehicle

to detect the available space and calculate the area of the parking space. The EDM (distance

detection encoder) also simultaneously measures the distance of the available parking space

by measuring the rotation of the wheel, which will feed the data back into dsPIC30F4011.

The vehicle will be equipped with several preset-parking profiles to allow more capabilities

in parking with different size or conditions of parking. When there is a match to the profile,

the vehicle will perform parallel parking autonomously. During the operation, the ESAM

(steering angle measurement encoder) will measure the steering angle of the steering wheel

and make adjustments to the steering angle achieve a perfect parking.

This paper presents the method and process of parking detection, automated steering, and the

integration of the components on the AU Golf car. The result of the self-parking is very good

and accurate. AU Golf car could park at the appropriate parking slots with several

experiments.

Page 15: Abstract Proceedings of IEEE SCAP 2013

8

The 2013 IEEE Student Conference on Senior Capstone Project

Improved Material Removal Rate by Abrasive Water Jet Milling

Kasidet Wangvithayakul*, Ukrit Jampachon, and Sataphol Wattnapornmongkol

Advisor: Pairat Tangpornprasert

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

At the present, the conventional milling using in material removal has an amount of

restrictive weaknesses. For instance, it is time- consuming and costly for operation, in

addition, it is difficult to operate in complex area or the small size of the work-piece.

Furthermore, high-toughness materials; stainless steel, alloy steel and titanium, can shorten

the lifetime of cutting tools. As the consequence of this, Abrasive Water Jet (AWJ) cutting,

involving the entrainment of abrasive particles into a high pressure jet of water, is being

introduced. Due to the high competency in material removal, AWJ milling could be an

opportunity to minimize machining time and increase process flexibility.

However, AWJ is generally used in through cutting material. The crux of AWJ milling is the

controlled depth of cut. Therefore, the first investigation is to study the parameters which are

related to the depth of cut: feed rate, water pressure, the garnet size, abrasive flow rate and

stand-off distance. As the result, the relation between feed rate and depth of cut tends to be a

linear pattern. Feed rate was chosen to be the process parameter under the design of

experiments while other parameters were set at constant values. Providing feed rate is

maintained as constant speed along a jet path, the depth of cut can be controlled in a

singleline pattern. The simple technique of AWJ milling has been developed by moving jet

back and forth in rectangular shape and changing lower feed rate to increase material removal

rate (MRR). The size of the experimental area was 30x20 mm and the experiment was

conducted at 5-40 mm range of the depth.

Nonetheless, penetrating traces at both sides of pocket milling, caused by impact of water

hammer at the beginning and ending point, could be hindrance. Therefore, one of the

effective solutions is to use high-hardness material (high speed steel, HSS) as masks,

protecting both penetrated sides. By doing this, the penetrations were removed and the

surface of pocket milling was uniform pattern. Likewise, the experimental results mainly

seem to be contentable; for example, the sample of experiments, using the depth of pocket

milling at 20 mm, of AWJ milling spent 12 minutes in operation whereas conventional

milling by 8-mm-size tool spent 15 minutes. From this experimental observation, AWJ

milling has more 20 percent of MRR than conventional milling.

Page 16: Abstract Proceedings of IEEE SCAP 2013

Section 2

Science and Technology

Page 17: Abstract Proceedings of IEEE SCAP 2013

10

The 2013 IEEE Student Conference on Senior Capstone Project

Disruptions to Saccade Planning: Effects on Eye Movements

YueYinglong*,Yen Shih-Cheng, and Chua fook kee

Advisor: Shih-Cheng

National University of Singapore, Singapore, e-mail: [email protected]

The planning and execution of voluntary saccades to shift the eyes around and encode each

parts of the scene or movie is required when you performing a memory task after viewing a

scene or movie. Two experiments are deployed in this study to examine how eye movements

respond to sudden disruptions in saccade planning during natural scene and motion clips

viewing. The experiment to explore eye movements by using static scenes was intensively

studied previously. However, the research have been done under dynamic scene transition are

seldom studied. Moreover, previously studies used the time as the parameter to trigger

disruption, for example, the second scene as a disruption was displayed immediately after a

scene had been viewing for 4.5 seconds. Therefore, in this study, instead of using time, we

are going to use fixation numbers as the parameter. Participants viewed two consecutive

scenes or movie clips. Three distinctive groups of eye movement were observed when scene

or movie transitions were introduced. Fixation ended more than 100ms called End-­late

group, duration when the transition occurred were elongated, and the eyes move towards

center immediately, landing there in the first fixation after transition, Fixations in End-Early

group ended less than 100ms after transition, stayed at first fixation after transition for a very

short period, subsequently move towards to the center, landing there in the second fixation

after transition. When transition occurred during a saccade, this Sacc group exhibited the

similar behavior of eye movement with End-Early group. Besides, its fixation duration did

not vary before and after the transition. Our suggestion, for End-Late group, transition

occurred early enough that allowed the eyes to abort the original saccade planning and

reprogramming towards to the center, a salient vantage point to re-start scene or movie

exploration at the first fixation after transition. The fixations in End-Early group, there is not

enough time for the saccade to reprogram when the transitions occurred, so that the saccade

executed in the original plan and delayed the centering of eye movements. In terms of Sacc

group, saccade planning would not be affected when transition occurred during saccade,

hence, the saccade landed in the first fixation shortly and centered at the second fixation.

Moreover, By comparing the results from two experiments. The effect of saccade

reprograming is more salient in the Static case. It would probably because movie clips were

involving in many motion changes, people have adapted to these changes and become less

sensitive to the scene transition.

Page 18: Abstract Proceedings of IEEE SCAP 2013

11

The 2013 IEEE Student Conference on Senior Capstone Project

Parallel Additive Implementation for modified Flexible Interval Representation System

Phoonlarp Mekraksakit*, Thanisorn Thanapongsapak, Chawin Wongchindakhun,

Advisor: Athasit Surarerks

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

A major problem in a domain of computer arithmetic is to reduce the computational time. It

needs improving and developing for now, as we can see that many researchers try to

introduce an algorithm to solve this problem. However, some of these lead to the errors

during computation, because of constraints in the real world. Therefore, many researchers

spend their time to develop a number system representing intervals that be used for dealing

with round-off error problem. In the present, there are many interval representation systems

proposed for coping with this problem such as Flexible Interval Representation System

(FIRS), which is considered to be a system having a good performance in memory and

computational time. Therefore, we will focus on improving computational time for FIRS.

For generalizing the FIRS with binary base, we firstly develop a new interval representation

system for bases are any algebraic numbers by modifying the FIRS, that we call it modified

Flexible Interval Representation System (mFIRS). In order to improve mFIRS computational

time, our interest is to manipulate operations in parallel manner. As the addition is a

fundamental arithmetic operation for others, so we propose parallel addition algorithms for

mFIRS, which we classify into two parts. The first part is created for dealing with the

symmetrical digit set while another part concerns with the unsymmetrical digit set. Lastly, we

also introduce the way for implementing these algorithms in a practical way by representing a

function for cellular automaton.

With these solutions we have mentioned above, now we can be able to add two intervals in

parallel manner for any algebraic number bases which have no conjugate of modulus one.

Our algorithms do not need to do any non-parallel processes after doing these algorithms,

which can solve computational time problem indeed.

Page 19: Abstract Proceedings of IEEE SCAP 2013

12

The 2013 IEEE Student Conference on Senior Capstone Project

A Quadrature Current Controlled Oscillator with 250 – 700 MHz Tuning Range

Radit Smunyahirun

Advisor: Ekachai Leelarasmee

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Since the broadcast system of television in Thailand will change from analog to digital

system, there will be plenty of frequency spectrums available for other communication uses.

This spectrum occupies approximately from 200 MHz to 700 MHz and is called ―White

Space‖. As with most of communication system, quadrature oscillators are the key

component in generating career signals. This work presents a quadrature oscillator whose

frequency can be tuned within the white space spectrum by using currents, hence the so

called ― quadrature current controlled oscillator‖.

This circuit that we propose is shown in Fig.1 where currents are outputs. It uses

three identical current mirrors connected in a ring fashion. To obtain the quadrature phase, an

extra capacitor is added to the middle current mirror. Its value is very large with respect to

parasitic capacitors within current mirrors. As dictated by Barkhausen criteria, the sum of

phase shift between input and output current signal of three current mirrors must equal to 360

degrees. Due to the added capacitor, phase shift between of the middle current mirror is

almost 90 degrees or quadrature related. From analysis, we get simple equations for design as

follows.

(1)

(2)

and

(3)

where is phase shift deviation from 90 degrees between , is gain of a single current

mirror, is the sum of all parasitic capacitors of a single current mirror, is

transconductance of N1, N3 and N5, and is frequency of oscillation. Since depends on

bias current, we can adjust by tuning the DC bias of all PMOS to change DC current

bias of all current mirrors.

The simulation result of the circuit by using Multisim with , , and

is shown in Figure 2. From this simulation result, we obtain

and .

.

Page 20: Abstract Proceedings of IEEE SCAP 2013

13

The 2013 IEEE Student Conference on Senior Capstone Project

Real Time People Tracking and Collision Avoidance using Sensors Fusion for an Indoor

Omni-directional Wheels Mobile Robot

Bhirawich Pholpoke

Advisor: Kanjanapan Sukvichai

Kasetsart University, Bangkok, Thailand, e-mail: [email protected]

Indoor robot applications fascinate robot researchers because they can be implemented and

have potential to be used in the real world applications. A person tracking is one of the most

important parts of mobile robot applications because it can be used to take care and help

people in everyday life. The robot not only tracks a person but must also run in a real

environment without colliding with obstacles. Therefore, the robot has to know its state such

as velocity and position in the environment while sensing the obstacles surrounding it. In this

paper, we divide the topics into two parts, i.e. the robot odometry estimation and a person

tracking trajectory generation algorithm.

First, the omni-directional wheel mobile robot though has a good maneuverability for many

kinds of environment but it always faces wheel slippage due to its mechanical characteristics.

This makes the wheel encoders which sense the rotation of wheels do not provide accurate

robot‘s velocity. This inaccuracy in odometry information, it is important for further robot

localization and control processes. Normally, an inertial measurement unit (IMU) is usually

used to fuse with wheel encoder data to increase the overall accuracy but this technique is not

sufficient applied to the high slip environment. By adding a laser range finder sensor, laser

scan matching technique can be implemented in this paper to improve system accuracy. All

of information from encoder, IMU and laser scan matching method are combined by using

the Kalman filter technique.

Second, a person tracking algorithm is an essential purpose of this paper in order to help

people in many activities, such as following and carrying items for people or carrying food

for elderly in a house. In this paper, tracking module is developed by using a laser range

finder sensor to track the initial object which is assumed to be a target person. Then, the

maximum likelihood algorithm is used to decide and perform as a data association algorithm

for probability filtering in the next process. Other important issue in this research is a safety

issue. The robot needs to be used in the real environment such as houses, offices and malls

which have wide varieties of obstacles. Therefore, the robot must track the person and avoid

these obstacles simultaneously. Safety must be obtained from the environment and the stop

command must be sent to robot before collisions against obstacles occur. Besides, the robot‘s

dynamic constraints are necessary to assign the safe trajectory for a desired path to the robot.

In conclusion, the real implemented robot was built and experimented. Robot consists of four

Swedish onmi-directional wheels driven by dc motors. Two URG-04LX laser scanners from

Hokuyo Automatic Co., Ltd are used as robot laser range finder devices; one for tracking and

another for avoidance. The example of an experiment result shows that the overall system

works pleasantly. The robot can maneuvers through the messy environment and it can track a

person while avoid the obstacles. All of these can be the fundamental for developing the

robot that can help elderly or handicap and improve human quality of life in the future

Page 21: Abstract Proceedings of IEEE SCAP 2013

14

The 2013 IEEE Student Conference on Senior Capstone Project

Design of a Phasor Measurement Unit (PMU) using FPGA

Porjed Sakunjareanpornchai

Advisor: Thirawut Fueangfu

King Mongkut‘s University of Technology North Bankok, KMUTNB

Bangkok, Thailand

e-mail: [email protected]

An imbalance of power generation and load in a power system causes the fluctuation of the

power system frequency. Changes in active and reactive power flow due to disturbances

affect the power system stability. The system operators need to know the transient behavior

of the system in order to improve the operations. Normally the operators use the control and

monitoring system called SCADA. However, the SCADA system updates every 2 – 6

seconds. This is not fast enough for the operator to remedy the effects of the disturbances in

real time. This also makes the recorded values useless for later transient analysis.

A phasor measurement unit (PMU) is a monitoring system that monitors bus voltages and a

frequency in real time. It sends the monitored values every 100 milliseconds to the server.

The differences in voltage angles of the system buses indicate active power flow. The

differences in voltage amplitudes show the reactive power flow. The changes in the

frequency (df/dt) can predict a system behavior.

The project designed the phasor measurement unit using FPGA. First the simulations were

done in EMTDC/PSCAD to obtain the power swing signals. Then the signals were imported

to MATLAB to test the algorithms. Next the algorithms were written on FPGA board to

make the PMU hardware. Finally the hardware was tested with OMICRON.

Page 22: Abstract Proceedings of IEEE SCAP 2013

15

The 2013 IEEE Student Conference on Senior Capstone Project

Fire Infiltrating Robot with Image Acquisition: A Microcontroller based Urban Search

and Rescue (USAR) Robot

Ruth A. Silvallana, Russel L. Amado, Kristoni R. Aquino, Carmen Fe C. de Guzman,

Moreno, Kendrick Kent L.

Electronics Engineering Department, Technological Institute of the Philippines, Quezon City

Philippines, e-mail: tip.edu.ph

The principal objective of this design is to lessen the risk of the job of fire fighters in

responding trapped victims inside fire incident area. A fire infiltrating robot was designed

and constructed as an Urban and Search Rescue (USAR) robot that has a real time image

acquisition utilizing wireless fidelity (Wi-Fi) technology to see the current situation inside the

emergency area which can be view in a customized controller, and wireless intercom to allow

fireman-victim communication. Further, it contains first aid materials at the back portion

compartment of the robot to increase the survival rate of trapped persons. The robot can also

send back ambient temperature and battery capacity to the controller. This robot is overall in

two parts: electronics and mechanical components. Electronic components such as the

PIC16F877A for the microcontroller, transceiver module, motor relay circuits and voltage

regulators. The mechanical components are Honda wiper motors, Stone Wool for thermal

insulation, aluminum and galvanized metal.

Different testing procedures were conducted to assess the robot‘s functionality. These

procedures are the test for maneuverability of the conveyor wheels at flat surface which will

test the robot‘s forward, backward and sideward movement and climbing downstairs/upstairs

capability at standard staircases. The material heat resistance was also tested specifically the

chassis that protects the electronic parts of the robot. Speed response and image acquisition

display quality were also included at the testing procedures. These tests result verified the

USAR application of the robot as a useful tool during fire emergency operation.

Page 23: Abstract Proceedings of IEEE SCAP 2013

16

The 2013 IEEE Student Conference on Senior Capstone Project

A New Algebraically Simple Chaotic Jerk Circuit

and System using Memristor Nonlinearity

Jirayu Peetakul and Sawitee Wangthawal

Advisor: Wimol Sanum

Intelligent Electronic Systems (IES) Research Laboratory

Faculty of Engineering, ThaiNichi Institute of Technology (TNI)

Patthanakarn, Suanlaung, Bangkok, Thailand, 10250. Fax :(+662)7632700,

Tel :(+662)7632600

e-mail addresses: [email protected] and [email protected]

Chaotic systems have been characterized as a system that offers a sensitive dependence on

initial conditions, i.e. small perturbations ultimately results in a dramatic change in system

states, and attracted great attention in recent years due to many possible applications in

various fields such as in secured communications and nonlinear control systems.

Considerable research interests have been made in searching for new chaotic systems with

minimal algebraic models as well as simple circuit implementations. The memristor, which is

postulated by Leon O. Chua in 1971 and successfully fabricated in 2008, is a two terminal

element in which magnetic flux between the terminals is a function of the electric charge that

passes through the device. Such a memristor has been realized as an addition to existing basic

circuit components including resistor, capacitor, and inductor. In particular, the memristor

can be modeled by third order polynomial and exhibits PieceWise Linear (PWL)

nonlinearity, which can be utilized in designing nonlinear chaotic circuits. Recently, chaotic

circuits implemented based on memristor have been suggested as a promising technology in

nonlinear dynamic circuits. However, algebraically simple and feasibility in dynamical

behavior variations have not much been investigated. This paper suggests the design of

algebraically simple chaotic jerk circuit and system through the use of memristor nonlinearity

where the term ‗jerk‘ comes from the fact that successive time derivatives of displacement,

velocity, and acceleration. The proposed system is simple in terms of mathematical model

with wide range of complex dynamical behaviors. The simulations have been performed in

MATLAB whilst the circuit implementation on board has realized a multipliers no. AD633

for building the memristor. Dynamical properties are demonstrated in terms of Equilibria,

Jacobian matrix, chaotic attractors, Bifurcations, Lyapunov exponents, KaplanYorke

dimension, and Poincaré maps. The proposed circuit offers a potential alternative to nonlinear

oscillators in communications and controls applications.

Page 24: Abstract Proceedings of IEEE SCAP 2013

17

The 2013 IEEE Student Conference on Senior Capstone Project

An AU Golf Car: Obstacles Avoidance System

Ankur Agrawal*, Krittamate Jirundorn, Tidarat Punyachai and Samuel Hussain Mogakolodi

Advisor: Narong Aphiratsakun

Assumption University, Bangkok, Thailand, e-mail: [email protected]

This paper presents part of the research work carries out at Assumption University to develop

an autonomous obstacles avoiding car. The car uses a digital camera as its primary sensor to

detect obstacles and calculate the optimum route to avoid the obstacle and use as little energy

as possible while doing so. This system could be a building block in a completely

autonomous car. Self-driving cars could bring along many benefits such as increased road

safety by taking humans out of the equation, and a decrease in road congestion due to poor

driver habits to name a few.

We are using Golf car as a test plant. This golf car uses a camera, in conjunction with an

image recognition program written in openCV to recognize objects that need to be avoided.

Once the obstacle has been detected, the software calculates it position and its area. This

information is used to send instructions to two separate dsPIC30F4011 microcontrollers,

which then send the data to a FIO board, which is based on the STM32. The FIO board then

adjusts the steering wheel, accelerator pedal, brake and gear in order to avoid the obstacle

with as little deviation from the original path as possible. This is demonstrated by the block

diagram, where arrows labeled ‗D‘ represent digital signals while those labeled ‗A‘ represent

analog signals.

AU Golf car could avoid placement of obstacles on the road with several experiments. Future

additions to this system include a more accurate detection and tracking system, possibly

through more complex software and/or through the inclusion of ultrasonic sensors.

Page 25: Abstract Proceedings of IEEE SCAP 2013

18

The 2013 IEEE Student Conference on Senior Capstone Project

Stability Control of a Self-Driving Bicycle

Vittavat Kaewmaneekul *, Sompong Jarujitjumlern and Whicha Thaitavon

Advisor: Viboon Sangveraphunsiri

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

This project, we propose a control strategy to stabilize an autonomous electric bicycle based

on the well-known state variable feedback. The simplified dynamic model of the bicycle is

the most important for the design of the controller. The necessary parameters of the dynamic

model are verified through 3D computer models and experimentation. The stabilizing control

of an autonomous bicycle is derived independently based on the simplified model. The

steering for stabilizing the bicycle is derived based on the output-zeroing controller or

regulator. A programming tool, LabVIEW, is used for developing an embedded control and

implemented on the National instrument hardware, the Compact RIO. A velocity controlled

conveyor is designed for verifying the total controlled system. The numerical solutions based

on the simplified model with the parameters of the model obtained from experiments are

shown and are confirmed by experimental results to verify the effectiveness of the proposed

control strategy.

Page 26: Abstract Proceedings of IEEE SCAP 2013

19

The 2013 IEEE Student Conference on Senior Capstone Project

Investigating on Damping Property of Laminated Composite Materials

Thanyarat Singhanart*, Kritsadaporn Thongsawang, Nattapon Kaewchalam

Nutthasan Srikwanma and Worapol Phraechinda

Department of Mechanical Engineering, Chulalongkorn University,

Phayathai, Pathumwan, Bangkok, 10330

* [email protected], Telephone Number: 02 218 6619, Fax. Number: 02 252 2889

An aluminium laminate – a wall structure made of polyethylene plate attached with

aluminium sheets on both sides by glue - is studied in this paper. As the vibration of the

structure can be controlled by the damping properties, the objective of this paper is to

determine the basic damping properties of aluminium, glue and polyethylene by an

experiment that is based on the determination of the logarithm decrement under free bending

vibration of beam with ended mass. The effect of frequency is considered by varying beam‘s

length and ended mass. In the experiment, the cantilever beam is excited by the cyclic

magnetic force at the fundamental natural frequency. Then, the force is removed to produce

the free vibration during which the deflection is detected by the laser displacement sensor.

The logarithm decrement can be determined from the experimental results and the basic

damping property as the function of frequency is presented in terms of specific damping

capacity (Fig. 1). As the specific damping capacity of the laminate can be determined using

the basic damping properties of the components, the specific damping capacity of aluminium

laminate is also predicted and verified by using the finite element method as shown in fig. 2.

Two finite element models is only considered; one is the model including aluminium and

polyethylene without glue and the other one is the model with glue. The results show that the

specific damping capacity can be predicted but the predicted value is smaller due to the effect

of shear strain in polyethylene and glue. In conclusion, this paper presents the experiments

for finding the basic damping properties of the materials and the specific damping capacity of

aluminium, glue and polyethylene as well as the aluminium laminate.

Page 27: Abstract Proceedings of IEEE SCAP 2013

Section 3

Health and Medicine

Page 28: Abstract Proceedings of IEEE SCAP 2013

21

The 2013 IEEE Student Conference on Senior Capstone Project

Design of an appointment system and bed management for Chemotherapy department

Thanakrit Thanathipanont* and Achiraya Eurakarawong

Advisor: Wipawee Tharmmaphornphilas

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Daycare or Chemotherapy department of a studied hospital receives patients from 4 main

upstream departmentsconsisting of Radiology department, Hematology department, Medical

Oncology department and Surgery department. The number of patients arriving at the daycare

department depends on the diagnosis schedulesof these 4 departments. Since the upstream and

daycare departments never collaborate in capacity planning, demand at the daycare department is

fluctuated resulting in demand over capacity on peak days. Currently, the daycare department

does not have an appointment system. Patients receive their queuing numbers and need to wait at

the department without knowing their expected time to service.

We develop patient information and drug formula databases that are linked to an appointment and

bed management application to help schedule patients and assign patients to beds. Two

scheduling rules are offered in the application, which are FCFS (the current rule used in the

hospital) and FCFSconsidering bed utilization. This application can provide information such as

queuing number, expected time to service to patients and help nurse in arranging beds.

Simulation is applied to analyze system performance. We found that the FCFS considering bed

utilization rule provides 5% increased in bed utilization, reduces the number of postponed

patients from 5 to 3 over peak days; however, increases 30 minutes of waiting time comparing to

FCFS rule.

We also propose a plan to smooth the demand by rescheduling diagnosis at the 4 upstream

departments. With simulation, we found that the new diagnosis schedule combining with FCFC

rule decreasesthe waiting time and number of postponed patients approximately 30 minutes and 4

out of 5 persons, respectively. Also, the new diagnosis schedule and FCFC with considering bed

utilization decreases the waiting time approximately 20 minutes and results in no postponed

patients.

Page 29: Abstract Proceedings of IEEE SCAP 2013

22

The 2013 IEEE Student Conference on Senior Capstone Project

Individual Speech Synthesis for ChulaDAISY

Sivaporn Homvanish*, Thiti Watansrimongkol, and Natchanon Phachongkitphiphat

Advisor: Proadpran P.Punyabukkana and Atiwong Suchato

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Individual Speech Synthesis for ChulaDAISY is the system that is implemented to help

create audio books for blind people who have no chance to reach knowledge freedom, so it is

hard for them to deal with daily knowledge form books. ChulaDAISY program is the

program that can create DAISY typed book (an audio book with full text). However, when a

volunteer who want to record their voice for making an audio book. They need to read and

record all sentence, which is a very time-consuming process. This project, Individual Speech

Synthesis for ChulaDAISY, will solve this problem. When users who are volunteer creating

audio books import content of books in formatted in document (.doc and .docx) and use this

mode. Program will look for all necessary phones that use in the project and store the

sentences which have those phones in the project. Then, program will show all sentences

which users have to record on the screen. This system helps users to reduce time to spend

while they are recording their voices for creating audio books because they do not have to

record all of sentences that have in books but program has chosen some sentences that cover

all of phones used in project. When users have finished recording, voice synthesis button will

appear and if users click this button, program will generate personal voice model to fulfill the

other sentences that have not recorded yet automatically. Moreover, this system will store

user's voice in the program. This feature will help when the same user want to use their voice

for making a new book.

Page 30: Abstract Proceedings of IEEE SCAP 2013

23

The 2013 IEEE Student Conference on Senior Capstone Project

Real-Time Gait Analysis Using Kinect-based Motion Capture System

Pattaravut Maleehuan

Advisor: Pizzanu Kanongchaiyos

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

The human gait has traditionally been studied subjectively through visual observations. By

combining advanced measurement technology and biomechanical modeling, human gait can

now be measured objectively. Today, there exists commercial systems can both capture

the end- user data, consisting of several important parameters in gait cycle, and analyze

the data for clinical diagnosis. However, the cost of such motion capture system is still too

high, hence only a few hospitals in the city can afford while patients cannot access to these

services at rural hospitals and clinics. This project proposes a real time motion capture system

based on low-cost Microsoft Kinects, each consists of infrared depth sensors and cameras, for

gait motion capture. Without marker configuration, kinematics and kinetic calculations can

be performed and standard temporal and spatial parameters such as joint angles can be

reported. Other typical gait parameters are speed, stride length, step length and time, stance

and swing phase can be user-defined or automatically computed. Finally, both proposed

system and commercial motion capture products at Chulalongkorn Comprehensive

Movement Disorders Center‖ are used to capture gait motions in parallel. Euclidean distance

between each joint parameter of both captured motions shows that the proposed system gives

the similar capture results to the commercial systems at the hospital. Besides its lower cost

and ease of set-up and configuration, the proposed system also provide the suggestion

algorithm for evaluating a Parkinson‘s patient‘s status which may be useful for hospitals and

clinics in rural areas where no gait specialists such as physiotherapists, orthopedists or

neurologists.

Page 31: Abstract Proceedings of IEEE SCAP 2013

24

The 2013 IEEE Student Conference on Senior Capstone Project

Gait Phase Detection using Wireless Smart Shoe with

Optimized Threshold by Genetic Algorithm

Nantawat Pinkam

Advisor: Itthisek Nilkhamhang

Sirindhorn International Institute of Technology, Thammasat University, Bangkok, Thailand,

e-mail: [email protected]

Nowadays, there are many kinds of gait rehabilitation technology to assist people who have

problem walking to attain the normal gait. Motion capture technology, such as VICON, can

be used to acquire body movement and provide three dimensional analyses.

Electromyography (EMG) is a biomedical sensor that can also be used to verify gait

abnormality, such as Lokomat developed Hocoma. However, these gait analysis devices lack

mobility due to the need for large equipments.

This paper proposes a novel decision system for segregation of five normal gait phases

(stance, heel-off, swing 1, swing 2 and heel-strike) by using a real-time wireless smart shoe.

The classification uses four force sensitive resistors (FSR) to measure force underneath the

foot together with an inertia measurement unit (IMU) that is attached at the back of the shoe.

In addition, IMU gives magnitude of acceleration and inclination angle of the foot with

respect to the ground. Data acquisition is collected through XBee wireless network protocol

in order to be processed serially by a computer. Threshold-based state transition theorem is

used to distinguish gait phases from received data. Video recording with real time data

embedded interface is used to verify the output of the proposed gait phase detection

algorithm. Experimental testing is conducted indoors on a treadmill. The verification process

uses human inspection to classify the gait phases from the recorded video based on the gait

phase definition. The thresholds of state transition are optimized by genetic algorithm by

comparing sets of thresholds with the result from the video. This method was experimented

and verified by a person who has a normal walking gait cycle.

This smart shoe can be used for gait rehabilitation of people who have abnormal gait, such as

the elderly and people who suffer diseases effecting their ambulation cycle. It can also be

applied in sportwear to help improve the performance of athletes. Most significantly, normal

people can use smart shoe to evaluate their own walking style. Moreover, the advantage in

mobility allows smart shoe to be used in other equipments, such as a prosthetic knee.

Page 32: Abstract Proceedings of IEEE SCAP 2013

25

The 2013 IEEE Student Conference on Senior Capstone Project

Ultrasound B-scans Image Denoising via

Expectation Maximization-based Unsharp Masking

Theerawit Wilaiprasitporn

Advisor: Chedsada Chinrungrueng, Widhyakorn Asdornwised

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

In this paper, we present an unsharp masking-based approach with subsequent bilateral

filtering stage to noise smoothing of ultrasound (US) image. At our first processing stage,

we propose image segmentation via EM to segregate two pixels populations instead of

separating original image into the low- and high-frequency components. Our proposed

method then enhances the edge by shifting the mean of the two pixels populations away from

each other. This is similar to the conventional unsharp masking structure, except that the

concept is reformulated and worked in probabilistic setting. At our second stage, we use

bilateral filtering to attenuate the retained noise in the flat areas. Performance of synthetic

and real clinical B-scan US images based on several dominant image quality measures, e.g.,

signal-to-noise ratio (SNR) and contrast-to-noise-ratio (CNR). The performance is improved

over the conventional US image de-speckling methods. The CNR-SNR performance tradeoff

is also addressed here for the first time.

Our experimental results in particular show that our proposed method with bilateral filtering

gives the best performance. One of the essential contributions in this work is that we address

a new image quality tradeoff, also known as SNR-CNR dilemma, for the first time. We found

that reconstructed US images cannot share good results of both the above mentioned

measures. This tradeoff is similar to of the tradeoff mentioned earlier in W. K. Heisenberg's

Uncertainty Principle, which stated that we cannot accurately measure displacement and

momentum at the same time.

Page 33: Abstract Proceedings of IEEE SCAP 2013

26

The 2013 IEEE Student Conference on Senior Capstone Project

An Automatic System for Measuring Electrode-Skin Impedance

Prawsiri Veeranarapanich*

Advisor: Arporn Teeramongkonramee and Apiwat Lek-uthai

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

It is well accepted that the quality of biopotential signals such as ECG, EEG, EMG etc. is

strongly affected by the impedance at electrode-skin interface. The power-line interference

which is common mode signal in nature, could be converted into differential mode signal via

the electrode-skin impedance and electrode unbalance. One possible solution in practice is to

minimize the electrode-skin impedance by a proper skin preparation such as skin abrasion.

Thus, the study of the electrode-skin interface is one of the key factors to obtain high quality

of biopotential signals. In this article, we present an automatic system which is capable to

measure electrode-skin impedance as a function of applied frequency. Our system consists of

a programmable function generator, an isolated DAQ module, a measuring circuit and a

laptop. In our measurement, the electrode-skin interface is applied with a constant amplitude

of a sinusoidal wave and the current information is provided by our measuring circuit. A

LabVIEW program has been developed to control both amplitude and frequency of sine wave

applied to the electrode-skin interface and also to acquire impedance voltage and current

from DAQ module. Our developed system is able to vary the applied frequency ranging from

1 Hz to 10 kHz. The results indicated that the changes of electrode-skin impedance in terms

of magnitude and phase versus applied frequency agree quite well with a simple lumped-

circuit model. We also observed a relatively large data scattering of impedance in the low

frequency (~ 1 to 200 Hz range). Since the current flow in the measuring circuit is limited by

relatively high values of impedance at low frequency, this results in a low signal signal-to-

noise ratio.

Page 34: Abstract Proceedings of IEEE SCAP 2013

27

The 2013 IEEE Student Conference on Senior Capstone Project

Computer Application using ARIMA and Simulation techniques for managing

inventory in hospital

Pattranit Khobkun, Watsa Tantiyuth, and Wassachon Veerakul

Advisor: Surapong Sirikulvadhana

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

This paper aims to develop a computer application for inventory management in order to

balance uncertain demand and supply effectively in hospital (Medical supply unit). Our

procedure can be divided into three main steps. First, we analyze data which consist of

historical demands and necessary parameters such as lead time and on hand which are

collected by the studied hospital. There are two different inventory policies, order when reach

s and replenish to S; (s, S) and order Q in every T period; (T, Q). Second, we forecast non-

stationary demand using Autoregressive integrated moving average (ARIMA). However,

ARIMA is an effective method for demand prediction only in short and medium periods.

Therefore, tracking signal is used to remind for keeping the accuracy of forecasting model.

Third, we find reorder point and max level for (s, S) policy as well as period and quantity for

(T, Q) policy by using the computer application based on criteria which is an acceptable

shortage level (Service level) with the lowest average inventory units. To develop the

computer application, we use MATLAB to calculate and also model ARIMA. Moreover, C#

language is used to construct the application and its interface. After that, we compare the

results with the existing planning method to show how effective of our study are. Finally, the

developed computer application cannot only be used in hospital industry but also other

industries which have non-stationary demand to improve inventory management.

Page 35: Abstract Proceedings of IEEE SCAP 2013

28

The 2013 IEEE Student Conference on Senior Capstone Project

Admission and Discharge Processes Improvement

Nopparuth lurkittikul, Tanawut Supparerkapa and Tanawat Jarusiripipat*

Advisor: Wipawee Tharmmaphornphilas

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

We study admission and discharge processes in one of the biggest hospital in Thailand. An

admission process is the process to examine whether patients are ready to stay at a hospital

for at least one night to get treatments. Normally, it consists of the following sub-processes:

interview, lung x-ray, blood test, electromagnetic scan, diagnosis, and room reservation.

Since patients need to halt food intake in order to do blood test, blood test and its prior

processes should be seriously considered. Currently, 5 nurses are assigned to the interview

process based upon diseases, 3 technicians are assigned to lung x-ray, and 1 nurse is assigned

to blood test. This results in 2.5 hours to complete blood test and its prior processes. It is also

found that blood test is the current bottleneck station. We develop a simulation model for the

admission process. This helps in assigning proper resources to each process. We found that

using the current staff but transferring a nurse from the interview process to blood test can

reduce 50 minutes of the completion time. Moreover, if an extra nurse can be assigned to the

blood test station, it will reduce the completion time by 1.25 hour. However, adding other

additional nurses is not significantly reduced the completion time.

Discharge process occurs after doctors order discharge. Currently, about 50% of the patients

cannot leave a hospital by noon of the discharge day. Consequently, they need to pay for an

extra day, as well as new patients cannot check in. We study discharge processes of 4 wards

to determine the best practice. Delayed discharge is expected to be 20-30% after

implementing the standard process.

Page 36: Abstract Proceedings of IEEE SCAP 2013

29

The 2013 IEEE Student Conference on Senior Capstone Project

Design and Manufacture Argumentative Mobility Device For Osteoarthritis patients in

Thailand

Vorapraj Chutintaranond, Wongsatorn Sathitsuksanoh, Pitchaya Sovachinda and Teerinth

Changsawang

Advisor: Pairat Tangpornprasert and Chanyaphan Virulsri.

Chulalongkorn University, Bangkok, Thailand, e-mail:[email protected]

Osteoarthritis (OA) is one of the most common chronic diseases founded in aging society. It

not only directly affects patients‘ mobility, but also causes depression due to commuting

restriction. During the twentieth century in Thailand, the proportion of elderly people

significantly increased. The proportion of men and women population over 60-year-old was

4.4% - 5.2% in 1970, 6.2% - 8.4% in 1995, 14.6% - 18% in 2025 and 22.7% and 26.9% in

2050[1]. To extend independent living and promoted their health, assistive mobility devices

have been essentially created and developed.

For the time being; there are several types of walker being exposed to commercial market as

argumentative mobility device. However, they are still lack of some significant functions

because of different environment between designed countries and Thailand where roughness

and step on the walk way are normally found. In addition; inconvenient carriage associating

with improper weight and obstructed architecture is required to be improved. Finding the best

solution on solving the mentioned situation became our project objectives.

A Hubless wheel walker provides advantageous weight and carriage compactness with

modern architecture. The wheel comprises aluminum wheel with solid tire and rotatable inner

hoops, which were fabricated by Nylon 6. The maximum capability of dynamic system

crossing step height and floor roughness is 10 centimeters. High grade aluminum was not

only used for supporting highly stressed parts on its structure but also appropriated for light

weight consideration expected lesser than a common use walker. For Folding in hubless

wheel walker was designed under concept aligning bicycle wheel circumference, so it

required radial slot mechanism on rear wheel structure and handles. Furthermore; middle

column adhered to body structure could be stretched and moved through to train entrance

way moreover it‘s also favorably utilized as seat. Consequently, a hubless wheel walker will

resolve the different environment for support and aid elderly in their daily life with high

stability and safety; besides, offer the new value on easy carriage in limited area.

Page 37: Abstract Proceedings of IEEE SCAP 2013

30

The 2013 IEEE Student Conference on Senior Capstone Project

Second Ear: A Hearing Tool for People with Hearing Disability

Pattara Sukprasert and Nuttapong Phaisarn

Advisor: Atiwong Suchato, Proadpran Punyabukkana

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

With more than 10% of the population in Thailand are people with disabilities, it is necessary

that new technologies are designed and developed to help with their daily life. Although

there are some technologies available, we found that most are targeted for people with visual

disabilities or the motor handicapped. For people with hearing disabilities, their difficulties

may be invisible to most. Often time, they need to have an assistant to constantly translate

spoken to sign language. Therefore, this project, ―Second Ear‖, is particularly devoted to

exploit the power of mobile technology to automatically translate spoken words into sign

language and pictures in the hope to help the deaf to communicate more easily.

This ―Second Ear‖ is an innovative solution that combine the use of Automatic Speech

Recognition (ASR), with the text-to-Sign Language algorithm, and Search algorithm. It

receives three words or syllables in Thai from any speaker through a microphone connected

to a mobile phone, then transmit it to an engine that performs all necessary functions and

returns the resulting sign language and pictures back to the phone. Naturally, the limitation of

three words will help improving the accuracy of the ASR. Though the users may have to limit

each utterance to three words or syllables, it is rather common to do so in the Thai

environment, especially among the younger generation.

This paper describes techniques employed in our research and offers results from our

experiments. The outcome of this project is quite promising and can further be developed to

reach a more natural way to help the deaf communicate better.

Page 38: Abstract Proceedings of IEEE SCAP 2013

31

The 2013 IEEE Student Conference on Senior Capstone Project

Design and Implementation of Surveillance Robot for Disabilities and Elderly

Surainan Kahmimah

Advisors: Jakapan Suaboot and Apichat Heednacram

Prince of Songkla University, Phuket, Thailand, e-mail: [email protected]

Robots have played important role in assisting mankind, especially surveillance tasks for

partial disabled and elderly people. These people are unable to move conveniently around

their spaces because of their physical limitations. Sometimes it is difficult for them to

visualize the environment around them without having to walk to the scene physically. If

there is a portable device that brings the real-time surveillance images to the partial disabled

users, they will respond to the near-by environment more effectively. Our proposed system

brings the live video to disabilities and elderly using low-cost surveillance robot. We apply

existing technologies and add more values by attaching the integrated camera that is

controlled via smart phone.

The system consists of two parts: a surveillance robot with rotatable spy camera; and a

remote control device for a user. The robot is made up of a radio-controlled truck available to

be purchased from a normal market. The truck carries an Arduino board, two DC motors, a

servomotor, a set of two batteries, and a surveillance camera that is used to send video signals

to the display (in this case we use a television). The communication between the robot and

the remote control is vital. Wireless LAN and mobile network were used in. Although, these

communication types offer high bandwidth and coverage, the system is complicated and the

price is high. A tiny and low batter consume radio transceiver was also used; however the

transmitter and receiver in this type of communication must use the same hardware and is not

suitable in our context. Therefore, our design is based on Bluetooth BlueDrone model that

allows many forms of inputs such as touch screen, accelerometer. The implementation uses a

smart phone with installed Android operating system and Bluetooth capability. The smart

phone acts as a remote control for a truck to move forwards, backwards, turning left and

turning right. A camera on the truck can rotate left and right which operates through the

software interface on the phone in a limited distance of 40 metres. The camera sends the live-

video images to the television, a display terminal. Our first prototype is considered low cost,

given that we can apply the devices used in a normal life (a smart phone and a television).

The RF camera, however, needs to be purchased (usually it is cheaper than IP camera).

Page 39: Abstract Proceedings of IEEE SCAP 2013

32

The 2013 IEEE Student Conference on Senior Capstone Project

Learning Causal Structures of Brain Connectivity in fMRI time series

Arnan Pongrattanakul* and Puttichai Lertkultanon*

Advisor: Jitkomut Songsiri

Chulalongkorn University, Bangkok, Thailand

e-mail: [email protected], [email protected], [email protected]

This project presents a method for learning causal structures in a time series. Learning causal

structure is a problem of finding directional relationships between any pair of variables in a

system. In this work, we consider the brain system whose variables are functional activities

occurred in each region. The brain activities can be measured via blood oxygen levels using

the technique of functional Magnetic Resonance Imaging (fMRI). Hence, discovering causal

structures in fMRI time series becomes an important topic in brain connectivity analysis.

Our goal is to construct a multivariate autoregressive (AR) model that describes the dynamics

of blood oxygen levels in the brain by solving an estimation problem with Granger causality.

The Granger causality principle states that for any pair of variables that has no causal

relationship between them (one variable does not cause another to occur, or vice versa), the

parameters in the model corresponding to those two variables are zero. Therefore, if we

assume that each brain region has strong connectivities with only some other regions, we can

expect parameter matrices in the model to be sparse (having many zero elements). Using the

least-squares method, we will show that the problem of estimating AR models with sparse

coefficient matrices can then be formulated as a convex optimization problem.

The process of learning causal structures in fMRI time series can be described as follows:

1. Formulating the model estimation problem using a framework of convex optimization. An

l1-norm penalty function is added into cost objective function to promote sparsity in the

solution.

2. Reducing the number of variables in the system. Since the number of variables in the brain

system is in the order of ten thousands, we apply the method of principal component analysis

(PCA) to reduce the problem dimension.

3. Solving the convex optimization problem via an iterative method, namely, the alternating

direction method of multipliers (ADMM). This method is suitable for large-scale problems

and it has a desirable convergence rate in practice.

4. Selecting the best model from a candidate set containing models with various orders and

Granger causality patterns. These models are constructed by varying parameters in the

estimation problem. Our criterion is to choose a stable model that yields the lowest fitting

error.

5. Constructing a graphical model from the estimated Granger causality pattern. The

graphical model will visually depict the relationship pattern between brain regions. Hence, it

can serve as a guideline or as prior knowledge for further brain studies in neurosciences.

We verify the effectiveness of this approach on a synthetic data set and then apply the method

to fMRI time series. The estimated causal structure of brain connectivity will be discussed in

the full paper.

Page 40: Abstract Proceedings of IEEE SCAP 2013

33

The 2013 IEEE Student Conference on Senior Capstone Project

Patient and treatment scheduling in Dermatological Outpatient Department queuing

system

Hataipat Juthasri*, and Suwat Tansiriyakul

Advisor: Oran Kittithreerapronchai

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Because of lack of medical resources and universal access to health care institutes, Thailand

has faced the quality of health service problem. Patients have experienced a long waiting time

and a heavy congestion, especially in outpatient departments (OPD) as they arrive at the same

time. Many researchers have viewed this problem as an operational issue and applied

principles of Operations Management to reduce the average time that a patient spends while

maintaining high utilization of medical resources. This article studies the effects of patient

appointment system and centralized treatment scheduling to Dermatological OPD queue at a

private hospital using commercial simulation software. We notice that the medical care

service has unique characteristics. Medical services have 2 independent criterion to separate

patients into the system which is the medical necessity and continued therapy. As a result,

patients are separated into 3 categories, particularly emergency patient, appointed patient and

walk-in patient. Moreover, each medical resource can perform more than one purpose that

provides greater complexity in scheduling. Having analyzed the OPD, we observed that the

diagnostic time depends on doctors and language barrier. The patient who requires a

translator spends long diagnostic time. Therefore, we observed that the queue is similar to a

job shop scheduling problem with precedent constraints. As a result, we compared job-shop

queuing policies which can provide guideline on medical care queuing system.

Page 41: Abstract Proceedings of IEEE SCAP 2013

Section 4

Network and Communications

Page 42: Abstract Proceedings of IEEE SCAP 2013

35

The 2013 IEEE Student Conference on Senior Capstone Project

An Alarm System via Social Network Services

Theeradet Khamperawat*, and Gorarit Chandhanimi

Advisor: Korporn Panyim

Department of Electrical Engineering, Faculty of Engineering, Mahidol University, Thailand,

e-mail: [email protected]

Alarm system has been widely used in wide-range of applications for security purposes (e.g.,

home security, fire and smoke detection system, and video surveillance system). When an

event is detected, the alarm system uses various methods to send an alarm message in order

to notify the property owner (or emergency personnel). One of the popular methods includes

sending the alarm message as a voice call to the property owner‘s mobile phone. Alternative

method includes sending a text message via SMS (Short Message Service). The limitation of

these methods is that the alarm message may not reach the property owner if his mobile

phone is turned off or he is failed to pick up the phone. As a result, it may be too late to take

an appropriate response to the incident.

The goal of this project is to look for an alternative method to deliver the alarm message to

the property owner for fast action. Recently, the social network services (e.g., Facebook,

Twitter, Instagram) have become one of the most popular tools for communications. One

interesting feature of the online social network is that the information posted is not available

only for page‘s owner, but also all the people on the page of owner‘s friend list. With some

adjustments in user‘s privacy setting, the post can also be seen publicly. The benefit is that

even though the property owner did not see the alarm message posted, there is a high chance

that Facebook friends of the property owner will see the alarm message (on the owner‘s

wall). They can inform the property owner via other channels or take an appropriate response

(e.g., call the emergency authorities) by themselves.

In this project, we develop the social network alarm system that uses the social network

services to deliver the alarm messages. The system will send the alarm signal to the property

owner‘s Facebook page. Our prototype uses a smoke detector that will detect an event of fire.

The alarm signal will be sent through Wi-Fi channel to the server that will automatically post

the alarm message on the property owner‘s Facebook wall. Note that this method can be used

as an alternative delivery mode in cooperation with voice calls or SMS messages.

Page 43: Abstract Proceedings of IEEE SCAP 2013

36

The 2013 IEEE Student Conference on Senior Capstone Project

An Improvement of 3G Network Coverage with Picocell in Mahidol University

Sanhapit Phatratipakorn*, and Thanate Srieakpanit

Advisor: Korporn Panyim

Department of Electrical Engineering, Faculty of Engineering, Mahidol University, Thailand,

e-mail: [email protected]

Up to 2013, the 3G cellular networks from various operators have been widely deployed in

all area of Mahidol University, Salaya campus. However, our measurements indicate that

there are number of locations that are experiencing weak signal strength. Those include areas

that are inside the building where offices and classrooms are located. This forces users to

move outside their offices or classrooms in order to receive stronger signal.

The goal of this project is to perform a measurement study to identify the 3G networks

coverage hole (an area where there is no signal or the signal strength is lower than the

acceptable threshold), and prepare a guideline for solutions to improve 3G networks signal

strength and coverage.

In this project, we use the field test mode of the iPhone as our measurement tools. We

perform our study on the 3G networks of the Advance Info Service Plc. (AIS). Based on our

preliminary results, we select the building that has low signal strength on average to perform

signal measurements in greater detail. The engineering building (a 5-floors building) is

chosen as the results show that there are number of the areas experiencing low signal level

(less than -85dBm) especially on the 1st

floor to the 3rd

floor. We perform a detailed

measurement and create a signal strength map of the engineering building. Then we perform

an analysis of using picocells in order to improve the signal strength of the network. We

study from both technical and financial view. We carefully select positions of the picocell

and its antennas. We enhance our deployment plan with the information for the building map

that shows the locations of offices and classrooms. For example, deploying an additional

antenna close to an office or a classroom would be more benefit place it to the area that is not

crowded (e.g., hallway or stairway). We use log-distance path loss model and wall

attenuation model to study the signal level after deployment. We also optimize our

deployment by adjusting the transmitted power and antenna directions. We also calculate the

return of investment after the picocell is deployed.

Our results show that the return of investment is close to 24 months. Our results provide a

guideline with insight information that can assist the university administrator when making

decisions on improving the network coverage of the 3G networks in Mahidol University

Page 44: Abstract Proceedings of IEEE SCAP 2013

37

The 2013 IEEE Student Conference on Senior Capstone Project

Performance Improvement of Wi-Fi Network in Faculty of Engineering, Chulalongkorn

University

Nipoon Sintoowong*, Nonthawat Ungtrakul, and Thanach Chotruangprasert

Advisors: Chaodit Aswakul and Chaya Limchitti

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected], Tel: 662-215-0871-3,

Fax: 662-215-4804, Corresponding author e-mail: [email protected]*,

[email protected], [email protected]

Faculty of Engineering, Chulalongkorn University, has provided an internet Wi-Fi for

students and staffs under the name of ―ENG‖. The purpose of the Wi-Fi Network is to

support the education activities of the faculty. However, the performance of ―ENG‖ Wi-Fi

has been dropped down respectively in these recent years. The echo from many users

complaining is perceivable. It‘s also significant for engineer to seek and examine the cause of

the problem. This senior project examines the in-depth detail of the engineering Wi-Fi. Look

deeply into the connection for successfully establish.

The connection problem can be classified into four main parts; the interference of the Wi-Fi

channel in 2.4 GHz is extremely dense which cause the connection inaccessible, the layout

planning of access points need to be well-planned to cover all the area of the faculty, in order

to insure the congestion of the traffic in the network, the insufficient power over Ethernet

distributed to access points is able to cause the connection unstable and the packet loss during

establishing the connection of the internet can also cause unsuccessful connection. Cisco

Spectrum Expert is used to help classifying the traffic intensity. The result is analyzed in

order to identify the cause of channel interference. Wireshark , the network sniffer tools, is

included in the project to capture the packets that run across the network. Auditing the

engineering network system log files is involved in the process of the project to observe the

packet that flow along the server. Also, checking on the network infrastructure design, both

by observing on the network blueprint and by perambulate around, to be able to suggest the

way to avoid the insufficient of the power over Ethernet at each access point.

The conclusion can also be drawn that the quality of life should be improved, in the term of

data accessibility. The student and the staffs of engineering, Chulalongkorn University should

be able to stay connected without any problem. The problem in establishing the connectivity

should be clarified. And the user‘s satisfaction should be increased.

Page 45: Abstract Proceedings of IEEE SCAP 2013

38

The 2013 IEEE Student Conference on Senior Capstone Project

Microwave Sensor with Artificial Neural Network Application

Narong Borijindargoon* and Nuttapon Muangmor

Advisor: Chuwong Phongcharoenpanich and Sorawat Chivapreecha

Faculty of Engineering, King Mongkut's Institute of Technology Ladkrabang,

Bangkok 10520, Thailand, e-mail: [email protected]

This project presents a concept of using the microstrip antenna as a microwave sensor for

testing solution with different sugar concentrations. The basic properties of relative

permittivity of different concentrations are discussed since the variation of sugar

concentration in such solution is directly related to the variation of the media characteristics,

such as dielectric constant and loss factor. Dielectric constant, which describes how the

media store the energy, is corresponded to the amount of the polar molecules in media itself,

while loss factor, which describes how the media dissipate the energy into heat, is

corresponded to the amount of bonding between molecules in such media. A microwave

sensor, which mainly relies on the distinctive of relative permittivity of the media, is

determined. The structure of the microwave sensor consists of two element of microstrip

antennas; one is for transmitting and the other is for receiving. The electromagnetic radiation

from the transmitting antenna is partially reflected and partially transmitted as a result of the

discontinuity in media. The standing wave, which influenced by the transmitted and reflected

signal, must be detected. Some partially transmitted signal is then absorbed by such media

and after that received by the receiving antenna. Signal from the microwave sensor is then

sent to process by an algorithm based on Artificial Neural Network (ANN). Artificial Neural

Network computational modules have recently gained recognition as an unconventional and

useful tool for RF and microwave modeling and design. Neural Network can be trained to

learn the relevant behavior between the detected signal and the characterist ics of the media,

which is the concentration of sugar in the solution.

Page 46: Abstract Proceedings of IEEE SCAP 2013

39

The 2013 IEEE Student Conference on Senior Capstone Project

Development of Radio Resource Planning Program for 2G and 3G

Mobile Cellular Networks

Jiradchaya Kanjanakunchorn and Jidapa Hansawangkit

Advisor: Pongsatorn Sedtheetorn

Department of Electrical Engineering, Faculty of Engineering, Mahidol University, Thailand,

e-mail: [email protected]

Recently, 3G or Third generation cellular networks have been launched in Thailand. Thanks

to WCDMA (Wide-band Code Division Multiple Access) and HSPA (High Speed Packet

Access) technology, operators are able to offer internet services to their subscribers, such as

internet browsing, emailing, video streaming etc. However, the radio resource planning is still

the major bottle neck of the system. The rising of radio resource usage from both newly

emerging 3G and existing 2G makes the planning process far more complicated.

In this project, we develop a radio resource planning program for both 2G and 3G cellular

networks. Here we apply theoretical backgrounds on cell site planning with the

considerations of frequency interferences and signal orthogonality. As a result, a powerful

resource planning program with user-friendly graphic interface is introduced. The program is

developed on MapInfo platform in which the source code is written by Map basic. With the

program, network engineers can automatically perform an efficient radio resource planning

with quality of service satisfaction. Furthermore, there are enhanced functions-- such as

checking neighbors and resources, searching cell sites, alerting improper planning, displaying

signal quality and monitoring network performance-- which allow engineers to accurately

allocate the resources. Then, this ubiquitous program can replace the time-consuming manual

resource planning which is easily mistaken by human error.

It is worth saying that this project is rewarded and granted by True Innovation Sector, True

Cooperation, as well as the program will be used in their business units such as 2G/3G RF

Planning Sector and Network Operation and Optimization Sector. Moreover, the project can

be further developed for radio resource planning on LTE (Long Term Evolution) networks in

which OFDMA (Orthogonal Frequency Division Multiple Access) is complied.

Page 47: Abstract Proceedings of IEEE SCAP 2013

40

The 2013 IEEE Student Conference on Senior Capstone Project

Communication Emulation using DMSL/COSEM on PC

Raivin Srisamrit

Advisor: Wanchalerm Pora

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Nowadays natural resources are decreasing continuously, resulting from the increasing of

population growth in the past. It causes over consumption of natural resourc. In additional,

most people consume the resource without realization that how much it has been taken. A

solution to slow down such consumption is to make people realize how much and when it

was taken. A smart meter will allow consumers to know the amount of electricity they

consume at what time. It is possible because the smart meter has an ability to store the

characteristics of electricity consumption in details for a month or more. Information from

smart meters allows electricity producers to keep a stability of its power system in a safer

zone. But to achieve this, ones need communication and IT infrastructure.

This project concerns about an emulation of communication of smart meter. By developing

Device Language Message Specification and Companion Specification For Energy Metering

(DLMS/COSEM) protocols on a personal computer, communication among meters can be

emulated. The protocols employ the technique of object modeling. The object has the

information of the meter inside. The communication channel is simulated by a software

circular buffer. Since the objects has various sizes and structures, they all must be encoded

into bitstreams. Moreover, most practical physical channels are bit-oriented. In this project,

Basic Encoding Rule is exploited, before an object is sent out. Two or more meters are

emulated by creating tasks in two or more threads.

The purpose of this project is to provide effective communication between the receiver and

sender. It also needs to control sequence of input and output of data for effective

communication.

Page 48: Abstract Proceedings of IEEE SCAP 2013

41

The 2013 IEEE Student Conference on Senior Capstone Project

Service Quality Improvement in Telecommunication Service Encounter

Konlawat Luengrat and Kritsana Settapittayakul

Advisor: Natcha Thawesaengskulthai

Industrial Engineering Department, Chulalongkorn University, Bangkok, Thailand,

e-mail: [email protected], [email protected], [email protected]

This project focuses on improving the service quality which is a key competitive advantage

of a telecommunication service encounter company. Providing first impression and

satisfaction to customers, the company must continuously enhance their service quality under

the limited resources. Questionnaire by SERVQUAL in five-likert scale which evaluates the

gap between customers‘ expectations and perceptions in five dimensions was used to analyze

the factors that impact to customer satisfaction. The important factor that shows the biggest

gap is the service speed. Hence, this project focuses on reducing the waiting time of the

service encounter to improve the customers‘ satisfaction level. For short-term improvement,

Value Engineering and brainstorming tools were implemented such as using color label

to indicate types of service counters, implementing fast track lane for the service with

short service time and assigning an available administrator to help the front office that

has customer with long service time. For long-term improvement, simulation with Arena13.0

was adopted to scheduled operators‘ lunch break period between 12.00 p.m. to 5.00 p.m.

The approach was matching the interarrival time of customer to the number of counter in

each period, and allowing few operators break in short interarrival time. The schedule

was set as no operator breaks in the first hour, one operator breaks in the second hour and

other three periods two operator break in each hour. The result of short term improvement is

that waiting time of customer payment reduced from 251 seconds to 105 seconds or

approximately 58.09% time savings and the waiting time of after sales service‘s customer

reduced from 669 seconds to 267 seconds or approximately 60.04% reduction. Finally, the

result of long term improvement is that waiting time of payment‘s customer during Monday

to Friday at the begin and end of month reduced from 131 seconds to 67 seconds or

approximately 48.8% reduction.

Page 49: Abstract Proceedings of IEEE SCAP 2013

42

The 2013 IEEE Student Conference on Senior Capstone Project

Target Location Searching Application using Ad-hoc Network

Kornkanok Khaoampai and Kamonluk Suksen

Advisor: Kultida Rojviboonchai

Department of Computer Engineering, Faculty of Engineering Chulalongkorn University,

Bangkok, Thailand, e-mail: [email protected], [email protected],

[email protected]

Nowadays, the number of elderly people is continuously increasing in every country in the

world including Thailand. Due to advanced public health and medication, the expectancy of

life has extended. Therefore, elderly people need help because of the declination of body

which naturally occurs when people get older. In addition, chronic diseases usually develop

at elderly age so it requires special care unlike people with other ages. Therefore, the goal of

this project is to create an application for tracking positions of the elderly people in the

hospital in order to provide them more intensive and quicker care. Consequently, it will

generate good quality of life and it will benefit such a small unit of society like the family as

well as the bigger unit like nation.

The developed application uses ad-hoc network to implement on smartphones. By using the

DECA protocol, data transfer is fully covered, transmission time is low, and resource usage

of devices is low. These advantages make the DECA protocol outperform other existing

protocols.

The smartphone running this application can determine its own location by retrieving list of

nearby wireless access points. This method has an advantage that it can be used to determine

the indoor location. After the location is determined, the application will broadcast its own

location via ad-hoc network. The smartphones in the network can broadcast their own

location or searching other smartphones‘ location. Not only is this application useful for

searching a position of the elderly people in the hospital but also useful for easily and quickly

finding a position of people in the buildings without any cost of Internet connection.

Page 50: Abstract Proceedings of IEEE SCAP 2013

43

The 2013 IEEE Student Conference on Senior Capstone Project

Traffic sensing application using ad-hoc network

Korakot Sirinam and Krittin Intharawijitr

Advisor: Kultida Rojviboonchai

Department of Computer Engineering, Faculty of Engineering,

Chulalongkorn University, Bangkok, Thailand,

e-mail: {korakot.si, krittin.i}@student.chula.ac.th

In the present, the lifestyle of people is confronted with a lot of rush hours. If people can

know traffic information of each road, it will help planning the trip, avoid the traffic jam, and

save their travel time. When people can choose the best route for the travel, they will have

more comfortable life. Therefore, we developed a traffic monitoring application using ad-hoc

network.

Nowadays, there are several techniques for traffic monitoring. Most of them are based on a

centralized structure with sensors along the roadside monitoring traffic density and

transmitting the results to a central unit where traffic information is analyzed and distributed

to users.

In this project, our traffic monitoring application named DecaTraffic is completely different.

Our application is based on a decentralized structure and implemented on the android

platform. Data processing, data transfer and storage are done in a self-organization way via a

smart phone in a vehicle. The smart phone in the vehicle can be considered as a good sensor

since it has a GPS, a physical wireless card, and a display screen. No sensors, no central units

and no broadcast stations are needed to be installed along the highways. This can reduce

infrastructure cost. Traffic data is collected and processed by the smart phone. The vehicles

exchange their traffic data when they come close to each other. The data is transferred via ad-

hoc network to all surrounding vehicles in the transmission range. To transfer the data, in this

project, we use the DECA protocol, which has been tested to be suitable for ad-hoc network

on the vehicles.

Page 51: Abstract Proceedings of IEEE SCAP 2013

44

The 2013 IEEE Student Conference on Senior Capstone Project

A Centralized Online Network-based Intrusion Prevention System Control via

Web Application

Ekgapark Wonghirunsombat*, Teewalee Asawaniwed, and Vassapon Hanchana

Advisor: Naruemon Wattanapongsakorn

King Mongkut‘s University of Technology Thonburi, Bangkok, Thailand,

e-mail: [email protected]

Nowadays, Internet and technology become an important part of all businesses and

organizations while attackers still find network vulnerabilities in order to exploit them.

Damages caused by malware tend to be more dangerous and expandable. There are several

types of Internet attacks with different objectives. For example, Denial of Service (DoS)

attack can essentially disable a computer or network. Port scanning (Probe) attack allows

attackers to know which ports are available to execute various attacks. Moreover, there are

Internet worms which are self-propagating and fast spreading in the Internet. Each type of

worms has its own behavior which is quite completely different from the other. So we need

an efficient way to protect our system from these malicious attacks.

In this paper, we present a Network-based Intrusion Detection and Prevention System

(IDPS), which can detect both network attacks and Internet Worms in online environment.

The proposed system can immediately classify various network attack types which are DoS

(e.g., HTTP flood, UDP flood), Probe (e.g., ACK Scan, FIN Stealth) and Internet worm (e.g.,

Witty, CodeRed II, Slammer, Blaster, Sasser). First, we extracted essential features from

Internet packet header that represent main characteristics of each network attack types (i.e.

number of distinct source/destination IP addresses and ports). Then, we provide many well-

known machine learning algorithms which are Decision TreeC4.5, Random Forest, Ripple

Rule, Bayesian Network and Back Propagation Neural Network in order to classify the traffic

into four classes; Normal, DoS, Probe and Worm. User can select any of these algorithms to

work with.

Our proposed IDPS is designed to work at a gateway as well as at various places in the

network for detecting attacks and worms inside the network. We have both stand-alone model

and distributed model. For stand-alone model, the system can detect and protect the network

attacks at a gateway by itself. For distributed mode, the packet capturing can be set up at

various locations inside the network while the classification part is installed at a separated

server which is implemented as a web application. The server allows system administrator to

monitor all of network traffic data captured from each places in the network, and also update

existing rule sets or learn new network datasets with a user-friendly graphic user interface.

Our detection tool is also integrated with iptables which is a packet filtering technique to

protect the network by blocking or dropping the malicious connection immediately.

In our experiments, we set up the experimental online network environment and captured

various network traffic data from the Computer Engineering Department at King Mongkut‘s

University of Technology Thonburi (KMUTT) for six days. We generated network attack to

the environment in order to evaluate the performance of the IDPS. From the experimental

results, we found that our IDPS can efficiently detect various network attack types with

detection rate higher than 98% as well as protect organization networks instantly if network

attacks are detected. In addition, our IDPS can be updated with new detection rules in order

to detect new attack types.

Page 52: Abstract Proceedings of IEEE SCAP 2013

45

The 2013 IEEE Student Conference on Senior Capstone Project

Voice On Demand Application over Ad-hoc Network

Jittapon Busarakum

Advisor: Kultida Rojviboonchai

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Nowadays, a mobile phone, particularly smartphone, has become one of the world most

widely used popular devices, including Thailand. However, mobile signal failure problem

still happens frequently in crowded places. Therefore, in this project, an application to

provide call service for users even though mobile signal fails is developed. It can be

improved the quality of communication.

Presently, people are able to communicate not only via mobile base station, but also via

personal wireless network which does not rely on a preexisting infrastructure. One of such

wireless network is Ad-hoc Network. The communication through Ad-hoc Network needs a

routing protocol so, in this project, Ad-hoc on-demand distance vector protocol (AODV)

which provides free-loop routes and the shortest path is used to minimize transmission cost.

In addition, the AODV protocol also has an ability to control a variation of routes and create

a new route when the path is broken. This application is implemented on the Google Nexus

One smartphone and the software used for implementation is Android API version 16.

This application uses the AODV protocol to send and receive voice data. The function of this

application is similar to the use of mobile phone with the range of 0-50 m. However, there is

a long delay for the range which is longer than 50 m. Therefore, this application is suitable

for users who are nearby and frequently communicate. It can be concluded that this

application would be a great choice for voice communication while mobile signal fails and

two communicating end points stay close.

Page 53: Abstract Proceedings of IEEE SCAP 2013

46

The 2013 IEEE Student Conference on Senior Capstone Project

A Study on Chaos in Digital Filter and Its Application to Secure Communication

RattaponThassanasateankit*, WittayaJantaramatsakarn, and WisarutPhompung

Advisor: SorawatChivapreecha

Department of Telecommunication Engineering, Faculty of Engineering,

King Mongkut‘s Institute of Technology Ladkrabang, Bangkok, Thailand,

e-mail: [email protected]

One important requirement for communication systems is data security. Data send from

source to destination generally does not want a third party triggered. Many algorithms

involving cryptography technology has been implemented and presented in many studies.

This paper is a study on chaos in digital filter and its application to secure communication.

Chaos in digital filter can occur due to finite wordlength effect in arithmetic operation using

two‘s complement number format. Generally, digital filter is linear time-invariant (LTI)

system. When arithmetic operation is overflowed, nonlinear phenomena will be happened in

digital filter and leads to chaos in digital filter. The chaotic encoder can be implemented by

infinite impulse response (IIR) filter while chaotic decoder which is inverse system can be

implemented by finite impulse response (FIR) filter. Overflow characteristic of two‘s

complement arithmetic is modeled by modulo-2 function, as f(x) = [(x+1) mod 2]-1. For

second order system, at least one of filter coefficient is chosen outside the stability triangle.

The chaotic encryption-decryption can be increase complexity by increasing order , the high

order chaotic cryptography is also proposed using basic idea of cascading of second order

sections. The filter coefficients in IIR filter and FIR filter can be compared to the keys that

used in any crypto systems. For the decryption, the process is inversion of encryption by

using the same filter coefficients. Moreover, the original data can be in form of sound and

image.

Page 54: Abstract Proceedings of IEEE SCAP 2013

Section 5

Computer and Information

Technology

Page 55: Abstract Proceedings of IEEE SCAP 2013

48

The 2013 IEEE Student Conference on Senior Capstone Project

Data Partitioning to Improve Similarity Join by Prefix Filtering Algorithm

Methus Bhirakit

Advisor: Jaruloj Chongstitvatana

Chulalongkom University, Bangkok, Thailand, e-mail: [email protected]

String data appears in most real world applications such as bioinformatics, data integration,

data cleaning etc. String data from multiple sources can vary because of the forms of words

or misspelling. Inconsistent data can lead to inaccurate match that can result in wrong

decision and near duplicated records. As a result, similarity match and similarity join, which

find similar string pairs from different sources, become necessary. The challenge of this

problem arises when real-world data sets are large. The filter and verify method has been

proposed to reduce the running time of the brute force method. Prefix filtering filters out

some dissimilar strings by examining only the prefix of strings, instead of the whole strings.

In this project, we propose prefix filtering with data partitioning for similarity join. For our

approach, the database is divided into partitions and prefix filtering is performed for each

partition of data. The proposed algorithm supports parallelism because filtering can be done

independently on each partition. Furthermore, a proper prefix length is determined for each

data partition in order to reduce the filter time while improve the selection of candidate

strings. An experiment is performed to compare the proposed algorithm to state-of-the-art

algorithms. The experiment shows that our method achieves higher performance as a result of

parallel execution. It is also found that the proposed algorithm increases the CPU utilization

when the number of partitions is increased. In addition, it is also shown that the number of

candidates retrieved in the filter step differs with different numbers of data partitions.

Page 56: Abstract Proceedings of IEEE SCAP 2013

49

The 2013 IEEE Student Conference on Senior Capstone Project

Automatic Screening Angle-Closure Glaucoma using Genetic

Algorithm and Genetic Kernel SVM

Paopat Ratpunpairoj

Advisor: Waree Kongprawechnon, Pished Bunnun, Toshiaki Kondo

Sirindhorn International Institute of Technology, Thailand, e-mail: [email protected]

In this study, an algorithm that contained the genetic algorithm (GA) and the genetic kernel

support vector machines (GK-SVM) is generated to screen the angle closure glaucoma. The

algorithm deals with the combination of the feature extraction and classification from slit

lamp images. The algorithm consists of four processes. At first, White Top-hat transform is

used as a pre-processing. The next process is feature extraction, which consists of channel

subtraction, normalization and power law transformation. Then, genetic algorithm (GA), it is

used for finding mathematical models representing slit lights in slit lamp images. Finally,

classification, GK-SVM is introduced in order to diagnose angle closure glaucoma from the

features based on mathematical models. The features of the classification are selected by

using the genetic algorithm. The test results obtained as GK-SVM algorithm is effective to

detect the angle closure glaucoma. The rate of the correct classification was 97 percent.

Page 57: Abstract Proceedings of IEEE SCAP 2013

50

The 2013 IEEE Student Conference on Senior Capstone Project

Application of Combinatorial Optimization with Coincidence Algorithm

for Flow shop Scheduling problem

Ornrumpha Srimongkolkul

Advisor: Prabhas Chongstitvatana

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

This research studies the problem of flow shop scheduling. As the market expands very fast,

the customers want more variety of products. Therefore, manufacturers must refine the

production in order to reduce both cost and time such as making many products by using the

same production line. The arrangement of machines to match the sequence of tasks is called

Flow shop. This problem consists of N jobs and M machines. Each job has to be processed on

every machine in the same order. As the result, there are vastly many possible solutions. The

main objective of this research is to find the work sequence which has minimum total

flowtime. This has challenged many researchers for many years. Many effective algorithms

have been proposed and new best result has been introduced rapidly. This work applies a

method called combinatorial optimization with coincidence algorithm to the problem. It is an

algorithm that makes use of both positive and negative samples to improve the solutions. The

algorithm starts with a set of the work sequences which is called population. Then, the total

flowtime of each solution is determined. Next, the algorithm selects some solutions and uses

them to update a probabilistic model that represents the distribution of the solutions. These

steps are iterated until a good solution is found. To test the proposed method, Taillard

benchmark instances are used. The results are compared to a number of well-known

algorithms such as Local Search and Estimation of Distribution Algorithm. The comparison

of the performance between these algorithms is discussed in details.

Page 58: Abstract Proceedings of IEEE SCAP 2013

51

The 2013 IEEE Student Conference on Senior Capstone Project

3D Game Framework for Android Platform

Wirune Kaewjai

Advisor: Thiti Rungcharoenpaisal

Department of Digital Media and Game System Engineering

Dhurakij Pundit University, Bangkok, Thailand,

e-mail: [email protected], [email protected]

Games on android platform often use game frameworks for the development process because

game frameworks reduce the amount of time for managing resource, handling various 3d file

format, vertex skinning, texturing, displacement mapping, lighting, controlling user interface

and etc.

Almost tasks in games can be exploited the power of parallel processing in GPU. However,

only commercial game frameworks in a market currently implement tasks supporting GPU

especially the tasks using an Autodesk standard file format call FBX.

FBX file format is the file containing resources being use for the whole scene in the game.

These resources can be implemented to use them by CPU or GPU but, nowadays, open

source game frameworks always use CPU for using these resources. Hence, when each frame

has a lot of objects and resources from FBX, this will lead to inefficient

Therefore, we have developed the games framework focusing in 3D graphics rendering

system using GPU and supporting FBX file format. This framework has main features such

as creating scene graph from FBX, resource management systems for reducing redundancy,

GPU vertex skinning for animated bone-characters, GPU displacement mapping for terrain

and GPU object instancing for many objects in each frame.

In addition, we also develop the features for games such as collision system, AI system via

standard state-machine and path following, graphics user interface system and controlling

system via accelerator motor and finger touching

As a result, we create a game using our framework call ―Wingkwai Huylehuy‖ and the demo

scenes to demonstrate and evaluate the efficiency of this framework. The results show that

our game framework can contain forty animated bone-characters using forty-eight frames per

second in average on the first model of Samsung Galaxy Tab 7 and can be used for creating

playable 3D racing game with AI systems and full 3D terrain by using FBX file format to be

the resources.

Page 59: Abstract Proceedings of IEEE SCAP 2013

52

The 2013 IEEE Student Conference on Senior Capstone Project

Development of Ground Planar Segmentation algorithm using 3D Point Clouds

Information from Kenect Depth Image Camera

Thanarat Chaveekolakit

Advisor: Kanjanapan Sukvichai

Kasetsart University, Bangkok, Thailand, 10900, e-Mail: [email protected]

A depth image camera such as 3D laser range finders or the Microsoft Kinect provides dense

3D measurements that typically consist of millions of points called Point Clouds. In robotics

applications like object manipulation or tele-operation, it is often crucial to interpret this

massive amount of data in real-time. For one application which is presented in this paper, is

an efficient method to segment 3D point clouds data into planar segments for adaptive using

in robot vision application such as robot visual simultaneous localization and mapping. In

consequence of using a depth image camera to be a sensor or a vision of robot, this sensor

will scan everything in the field-of- view (FOV) of its specification. This means the robot

vision is depended on the direction of the sensor. Therefore, if the sensor directly point to the

ground, a robot will define the ground plane as an obstacle and then it cannot move to other

places or destination by using any regular planning trajectory generators. This is the reason

why extraction of the ground plane first is must be performed before make robot does other

task. The ground plan extraction can be done by using planar segmentation before making

Virtual 2D Scan to be a vision of a robot.

In this work, a method for extracting a ground plane segmentation from 3D point clouds data

is presented for making a robotic map (Virtual 2D Scan Map), which including four main

steps to approach. First step is down sampling point clouds using a Voxel Grid filter. Next

step is setting a vision boundary of robot using Pass Through filter, third step is detection of

ground plane using Planar Segmentation (Random Sample Consensus, RANSAC) algorithm.

Last step is extraction ground plane using Extract Indices filter.

Page 60: Abstract Proceedings of IEEE SCAP 2013

53

The 2013 IEEE Student Conference on Senior Capstone Project

Adaptive strategic decision point extraction from Influence maps in games

Namo Podee

Advisor: Vishnu Kotrajaras, Boonserm Kijsirikul

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

One of the main challenges facing artificial intelligence (AI) is the vast amount of data,

which contains both helpful and unhelpful information from the perceived environment

mixed together. It is often difficult for an agent to understand relationships amongst data, and

this can lead to ineffective learning and decision-making. Many studies have proposed

solutions to this problem by using machine learning. However, recent results in applying

machine learning to AI are insufficient to encourage uptake outside academia. Its

unpredictable behavior combined with its unanalyzable structure make it difficult to use in

practical applications.

Our research tackles how the agent learning process in games can be improved so that it is

more robust and faster when retrieving information from its surrounding. The influence map,

which is a representation of the agents‘ influences, will help and support the AI gain a greater

understanding of the environment, without requiring detailed information of the

map.Therefore the vast amount of data from the map is transformed into information that can

be efficiently processed. Moreover, using the influence map as the input to a machine

learning system leads to improvement in the efficiency of learning process and allows the

system to generate the map that indicates the strategic value of each area. Our AI then uses

such strategic values to decide its action. This allows higher-level analysis of game situations,

leading to more accurate and faster decision-making. In addition, since programmers have

more direct control over the AI‘s decision, the AI behavior becomes more predictable.

Page 61: Abstract Proceedings of IEEE SCAP 2013

54

The 2013 IEEE Student Conference on Senior Capstone Project

Tower Defence Game Project

Nattapong Rattasamut and Weerapong Siriyindee

Advisor: Natasha Dejdumrong

King Mongkut's University of Technology Thonburi, Bangkok, Thailand

e-mail: [email protected]

The Tower Defence Game is about defence from enemy raiders. The player can build tower

for prevent the enemy. The enemy out to be decided based on a player's style. The towers and

enemies are different abilities and powers. Such as the attack power and the attack speed.

The game is by using HTML5 and JavaScript to write a part of the engine. They are divided

into sections as follows.

Tower System. There are many different types of tower for different enemies. The ability of

tower can be identified clearly.

Enemy System. Enemies will come out at the start and will continue to along the route and go

to the end of the scene. The players must build the tower for defence enemy. Enemies in each

level will have different difficulty.

Scene System. The scene divided by the level of the player. The scene can be changed if the

player through 50 levels. In the scene have the start point, end point and the barrier.

Level System. The levels increased when player defence enemies in the level. The ability of

the enemy to be different.

Money System. The money calculated from the level of the enemy multiple levels of the

game. The money can be used to purchase and upgrade towers. And the end of the game

money is calculated as the score.

Score System. The score calculated from the level of the enemy multiple levels of the game

and minus the number of all towers. If the number is negative, then the value is 0.

Page 62: Abstract Proceedings of IEEE SCAP 2013

55

The 2013 IEEE Student Conference on Senior Capstone Project

Application for Publicity of promotional products online.

Natthaporn Singmanee and Chadaporn Chiansri

Advisor: Sirapat Chiewchanwattana

Khon Kaen University, Khon Kaenn, Thailand,

e-mail [email protected],[email protected]

Application for Publicity of promotional products online is the system was developed on a

mobile phone with an Android operating system, and can connect to the Internet. The purpose

of Application for Publicity of promotional products online to support and promote products

and services to users. The system is divided in to 2 parts which are the web application and

the mobile application, and which both of them use data from database

In part of the web application, the traders can store the information of products in the

database. For use publicity of promotional products. As for a mobile application, users can

receive information of products and the latest news about promotional products from traders.

The highlight of this work is Promotion is offered through wireless technology that can

attract users to share the best deals and, as well as, enhance the quality of life in terms of

access to discounts on products. And a new way of the goods delivering for trader. Also,

enhance the quality of life for trader in terms of promotional products and, as well as, reduce

the expenses in the media rates increasing ahead of inflation. The result of the system, the

system is included are promotions in a single system. Can be used with the trader and many

types of products. And has its own social network, can offer promotions, scoring of the user

interest are categories and can show the path to the user store form a location based service

system. Compared with Mister Donut with a unique product offering that only.

Page 63: Abstract Proceedings of IEEE SCAP 2013

56

The 2013 IEEE Student Conference on Senior Capstone Project

Study on a National SMS HUB Solution for Bulk SMS and National Policies on

SMS Regulator in Sri Lanka

M.W. P. Maduranga*, M. Gamage

Advisor: Manodha Gamage

Asian Institute of Technology, Bangkok, Thailand, e-mail: [email protected]

This research was aimed to design a national hub to overcome some difficulties in bulk SMS

business in Sri Lanka. As of now in Sri Lanka Bulk SMS usage (advertising, voting in reality

shows etc.) is not regulated. Since this could lead to national cyber security issues and some

social issues, it is an urgent requirement in Sri Lanka to decide on the policies to handle Bulk

SMS. Also there must be a technical solution to safely operate bulk SMS by all mobile

operators.

In this solution, it is proposed to issue a new license to a National SMS HUB operator (or

two) and all mobile network operators and bulk SMS service providers should provide their

services through the National SMS Hub. The network architecture in which these connections

are to be made also proposed in this paper.

Mobile network operators connect their Short Message Service Center (SMSC‘s) to the

National SMS Hub and bulk SMS service providers connects their SMS gateways to the hub.

As a part of this research a National policy to manage bulk SMS and the license to be given

to Hub operator(s) by Telecommunications Regulatory Commission (TRCSL) also proposed.

Mobile network operators, bulk SMS service providers and National Hub operator should be

responsible for adhering to the policy.

A survey was conducted and the responses to the survey were used in formulating the policy.

Also discussed with mobile network operators, mobile equipment vendors, as well as bulk

SMS service providers in Sri Lanka. Their ideas were useful to propose an architecture with

SMS Hub and decide on several policies.

Page 64: Abstract Proceedings of IEEE SCAP 2013

57

The 2013 IEEE Student Conference on Senior Capstone Project

Indoor Environment 3D-Image Reconstruction using Microsoft Kinect for Rescue

Robot

Piyaphat Phukphan*, Vasitphon Pawankiattikun

Advisor: Toshiaki Kondo, Itthisek Nilkhamhang

Sirindhorn International Institute of Technology, Thammasat University, Thailand

e-mail: [email protected], [email protected]

Nowadays, robotic technology has received widespread attention and implementation

throughout the world. Current trends of robotic research favors humanoids that have

mechanical systems, external appearances and programmable actions similar to a human

being. One important component of a robot is the vision system that mimics the behaviors of

human eyes. It allows the robot to determine its location, identify environmental objects,

select a path to reach its destination and avoid collision with obstacles. This project therefore

focuses on the vision system of rescue robots that can scan and map the unknown

environment.

A major part of this project is 3D reconstruction of the environment from captured camera

images. The camera used in this project is the Microsoft Kinect, which is a special camera

that can detect and store depth values of any object it sees. In one image frame, each pixel has

its own RGB value and depth value that is measured from the Kinect. The image is displayed

as a 3D depth map and color representation corresponding to RGB and depth values. A

matching algorithm is then applied to merge different frames together. This process is done

by dividing the right region of previous image and the left region of current image into

multiple blocks and then finding the sum of absolute difference to determine matches. Both

RGB and depth values are used for matching condition to increase efficiency. As the Kinect

moves, a 3D reconstruction of the environment is obtained.

The results of the project show that a 3D model of the experiment room can be constructed

completely by rotating the Kinect camera placed at the center of the room. The optimal

threshold values are obtained by trial-and-error method and compared using sum of absolute

difference. The Kinect was also mounted onto a robot arm to maintain a constant angular

speed. Lighting conditions were found to significantly affect the image matching process as

the color of the objects change depending on the intensity of the light reflecting from some

surfaces. The experimental room was therefore constructed to control ambient light and

objects in terms of color and depth.

The proposed algorithm provides a complete and clear view of the environment around the

robot, mimicking the behaviors of human vision. Additionally, path finding and object-

tracking algorithm can be implemented together with this project to further enhance the

system. This will allow a rescue robot to see collapsed surroundings, find and identify victims

and avoid obstacles more easily and precisely.

Page 65: Abstract Proceedings of IEEE SCAP 2013

58

The 2013 IEEE Student Conference on Senior Capstone Project

Multiple-CCTV monitoring system for iOS-based mobile device

Rakneung Pichitwitaya-ake

Advisor: Supavadee Aramvith

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Nowadays, Closed Circuit Television (CCTV) cameras are installed in both indoor and public

places for surveillance purpose. However, there are some limitations such as the access to the

video can be done only through computer monitor. As the mobile devices have been

progressively developed and are widely used, this project has an objective to extend the

capability of CCTV system for detecting the intrusion. The video from CCTV cameras can be

monitored via mobile devices. In 2011, multiple-CCTV monitoring system has been

developed for Android-based mobile devices. This year, the extension has been proposed for

monitor system based iOS mobile devices such as iPhone and iPad. The monitoring system

using iPad with iOS version 5 and 6 is tested with four Pan-Tilt-Zoom (PTZ) IP cameras

installed at floor 12 and 13, Engineering 4 Building, Chulalongkorn University.

Programs developed in this project consists of two parts: on the server and on iOS-based

mobile device. The program on the server consists of motion detection using frame

differencing in a given area chosen by users, and the process of sending alert messages to

users' iOS-based mobile device. If there is some motion in a given area, the server will record

an image and send alert messages to users immediately by sending push notification via

wireless internet. Another program is developed on the iOS-based mobile device to display

monitoring video and to send control signal for PTZ camera via wireless internet.

In summary, the system can monitor and control up to four PTZ cameras via iPad. As simple

motion detection algorithm is used, sufficient light is needed for accurate detection.

Page 66: Abstract Proceedings of IEEE SCAP 2013

59

The 2013 IEEE Student Conference on Senior Capstone Project

Development of a real-time object tracking camera

*Krittin Pachtrachai, Toshiaki Kondo

Advisor: Toshiaki Kondo

Sirindhorn International Institute of Technology, Thammasat University, Thailand,

e-mail: [email protected]

One of the most significant information in artificial intelligence is image information.

Undoubtedly, the image must contain some information that has benefits on the application, it

can be some interested object that is captured by the camera. In practical situation, the

interested object moves all the time with an unpredictable movement, the situation that the

interested object goes out of frame may occur, while the image is demanded to yield clearly

image of the interested object. It would be more convenient if the targeted object is focused at

the center of the frame all the time of the process. Object tracking can be applied to solve this

kind of problem, it is implemented on several applications nowadays such as traffic

management, or medical therapy. In order to sustain the targeted object to be focused, i.e. be

at the center of the frame, there are two major functions, object tracking and camera motion

control. In the object tracking part, several tracking methods provided by Computer Vision

technique can be applied to track a desired object. For instance, the sum of absolute

differences (SAD) matching, the sum of square differences (SSD) matching, correlation

matching, and normalized cross-correlation method. The matching technique will locate the

pattern that seems to be perfect match of the template in the image obtained from the camera.

On the other hand, micro-controller and two servo motors play a vital role in the camera

motion. Camera motion tries to maintain the targeted object to be at the center with less

oscillation according to the error in motion tracking. This paper proposes an object tracking

camera system, that can provide real-time object tracking camera with insensitivity with other

environment.

Page 67: Abstract Proceedings of IEEE SCAP 2013

60

The 2013 IEEE Student Conference on Senior Capstone Project

Development of Unification Filesystem for Cloud Storage

Worachate Apichanukul* and Thanawat Suknim

Advisor: Krerk Piromsopa

Faculty of Engineering, Chulalongkorn University, Bangkok, Thailand,

e-mail: [email protected]*, [email protected]

Tel: +66-(0)2-218-6956, Fax: +66-(0)2-218-6955

This engineering project represents a Unification Filesystem for Cloud Storage. The

distinguished feature of this filesystem is the integration of files and folders from three cloud

storage providers: Google drive, Sky drive, and Dropbox, merging them in such way that user

can see as if files and folders from three cloud storages are together at the mount point on a

user‘s PC. At the mount point, a unification filesystem will be created in order to show files

and folders. There are two important mechanisms behind the scene to communicate, control

and manage all real files and folders on three cloud storages. The first mechanism, main

mechanism of this filesystem, is a File and Folder Names Management mechanism. This is a

mechanism that manages file and folder names in order to allow a user to identify and

recognize the real address of these files and folders. Moreover, this mechanism can deal with

the duplication name of files and folders which come from different source of cloud storages.

And another mechanism with minor importance is a Suitable Cloud Storage Selection for

Uploading mechanism. This mechanism has a special protocol for a user to allow the

filesystem to choose the suitable cloud storage source for uploading a file. The technique that

is used for selecting the suitable cloud storage is based on an ability to support the type of

data of each cloud storage provider. In an architecture design, this filesystem is designed in a

modular way which is easy to add more cloud storage source in the future. ―UFCS‖

(Unification Filesystem for Cloud Storage), a prototype of this filesystem, has been

developed under the FUSE framework which is a standard for developing the Linux

Filesystem. About the efficiency of this prototype, it has normal file and folder control

functions like the Linux Filesystem but it has only one disadvantage that can happen in some

situation which is high response time of the filesystem due to the communication delay.

Page 68: Abstract Proceedings of IEEE SCAP 2013

61

The 2013 IEEE Student Conference on Senior Capstone Project

Personalized Assessment, Review and Learning System Project

Tawan Junhom, Sawaruk Sirimungklanurak and Siriwat Saiwiwat

Advisor: Natasha Dejdumrong

King Mongkut's University of Technology Thonburi, Bangkok, Thailand,

e-mail: [email protected], Tel: 024709390

The project aims to develop a customized e-learning system called PEARL (Personalized

Assessment Review and Learning) based on HTML5, PHP, and JavaScript. This personalized

learning system is designed to assess and recognize each learner‘s abilities so that the content

can be delivered based on the exact needs and skills of an individual. It helps save time by

creating an efficient learning process that matches each person‘s goals, skills, and

availability.

By following a few simple steps, a personalized content and schedule can be generated by the

system. Initially, each user has to define their study goals and availability. Based on that

information, an assessment is then created to evaluate the learner‘s strengths and weaknesses

down to each skill level. The system subsequently proceeds to recommend customized

lessons and schedule that suit the learning needs of that particular learner. The teacher can

also play a role by giving advice and interacting with the student directly.

The system basically consists of two main websites: the main and the administrative sites.

The main website is for the learner to perform the assessment and take lessons. It is

composed of the authentication, assessment, testing and schedule management systems. The

administrative website is used by a superuser to administer and manage all the data in the

main website including the courses, the content, and the tests. This is comprised of the

authentication, course management, lesson management, test management and member

management systems.

This project is based upon the notion of personalized education employed at iFAST English

Language Institute.

Page 69: Abstract Proceedings of IEEE SCAP 2013

62

The 2013 IEEE Student Conference on Senior Capstone Project

Proximity Information Exchange based on NFC

Boripach Khuhaprema, and Win Eiwwongcharoen

Advisor: Supachai Vorapojpisut

Thammasat University, Pathumtani, Thailand, e-mail: [email protected]

This project considers a scenario in which users must be at a specific location in order to

acquire information from a publisher. Based on the concept of proximity marketing,

advertisement of multimedia contents is considered as a case study. We have selected Near

Field Communication (NFC) technology as an information container to distribute sample

multimedia contents to user via smartphones. NFC is a short range wireless communication

that is similar to Bluetooth or Wi-Fi. The typical range is from 1 to 4 centimeters. Due to the

read-write capability of NFC tags, information exchange is implemented as two-way

communication with respect to the concept of bulletin boards. That is, a store owner registers

an NFC Data Exchange Format (NDEF) record as a content URL into an NFC tag, then users

can view and write comments about the viewed content back into the NFC tag. Due to

limitations of tag storage, multimedia contents and also user comments are stored in Internet -

based servers in which URL contents. The store owner will be able to upload their contents to

the servers. The demonstration is implemented as a client mobile application for Android

smartphones and a collection of cloud computing-based web applications/services on Google

App Engine. This project is aim to help publishers to advertise their contents easier and help

consumers to take their interested contents with them in their smartphones. It also show how

useful NFC can be and how it can be used in the real world situation.

Page 70: Abstract Proceedings of IEEE SCAP 2013

Section 6

Sufficiency Economy

Page 71: Abstract Proceedings of IEEE SCAP 2013

64

The 2013 IEEE Student Conference on Senior Capstone Project

Markerless Fruit Augmented Reality

Songpon Sastrawaha, Sirinad Kapoung

Advisor: Kumron Sunat

Khon Kaen University, Khon Kaen, Thailand, e-mail: [email protected]

Climate change has already occurred. Proof is the melting of glaciers and ice sheets in the

Polar Regions, Bleaching of coral reefs, Sea level rise and Ecosystem changes and a long

drought and deteriorating. The World Health Organization reports that more than 150,000

people die each year from climate change. For this reason, a new generation has started

sustainability development conservation and natural resources are increasing. Growing the

tree is a well-known activity for world taking care and adds the green space, which is like the

lung absorption of industrial plants waste or smoke caused from the combustion of the engine

on the road. Therefore the creation of a sustainable development is the development approach

of the nation to meet the needs of the present generation. Regard to future generations.

Maintain the quality of natural resources and the environment to continue to exist. By using

resources more efficiently and control the limit. While the world's still progress to be the

relentless. So an application is developed to support the activity for world taking care. Details

of the application are including permitting growing and taking care of the tree. The

Augmented Reality (AR) technology is applied for the amused. The AR is a technology that

integrates the real world with the virtual world via mobile camera and display 3D object. The

application is advanced from the existing AR applications. They detected the marker plate

which causes of non-flexible. Our application is more flexible by detecting a real object; real

fruit, using AR and showing plant of fruit in 3D object. The plant of fruit uses the reference

location from GPS system and display growing of the tree like real tree. The application can

exchange the real product in store who participated with their promotion via stamp code

which generated by server. The store can access this application via web application for

managing their information. The development go far along ,the environment problem

increasingly incubated and intensified .So , our project request to be a part of promoting the

new generation people to take care of the world. The resource would alive with this world a

long time.

Page 72: Abstract Proceedings of IEEE SCAP 2013

65

The 2013 IEEE Student Conference on Senior Capstone Project

Design and Implementation of Artificial Prediction Markets Using

Logarithmic Market Scoring Rules

Pannate Jongpanichkultorn

Advisor: Prabhas Chongstitvattana and Daricha Sutivong

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

―Prediction markets‖ is a technique developed for predicting an outcome of future events by

trading contracts in the markets. Many well-known international firms, such as Google,

British Petroleum, and Hewlett-Packard, use prediction markets for their internal decisions.

Graefe and Armstrong have also shown through various evidences that the accuracy of this

technique is at least equivalent to other techniques with similar pool of participants. There

are two main approaches to study prediction markets. One is to experiment using controlled

real prediction markets. The other is to construct artificial prediction markets-virtual

prediction markets implemented on a computer where trading is done by computer-generated

agents. The first option faces numerous problems including uncontrollable factors and

various financial and law issues. Hence, this paper proposes the model of artificial prediction

markets that capture the characteristics of real prediction markets. Our work includes the

proposed belief model, the information distribution approach, and the market mechanisms.

Specifically, the agents in these markets can update their belief towards the specified event

according to their prior belief, received information, and market price. Using the proposed

market design; the artificial markets are implemented and analyses are performed on key

parameters, such as the impact of the number of traders, the mean value of initial belief, the

memory pool size, the transmission rate, and the ratio of innovators on the proposed

prediction markets.

In the experiments, the markets performance is measured using the number of steps to

convergence, which is the minimum number of steps it takes until the market price changes

within one percent of the real price for ten consecutive steps. This measure reflects the time

needed to obtain the markets answer in the real prediction markets. Our experimental results

have shown that the number of traders and the mean value of initial belief have no significant

impact on the convergence speed. However, the memory pool size has a negative impact on

the convergence because of its delay in adjusting to the true value. Finally, the transmission

rate and the ratio of innovators have positive impacts on the convergence of the prediction

markets.

Page 73: Abstract Proceedings of IEEE SCAP 2013

66

The 2013 IEEE Student Conference on Senior Capstone Project

Internal Rate of Return for Probabilistic Cash Flows Project

Phatharaphan Chamnannarongsak

Advisor: Daricha Sutivong

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Internal rate of return is one of the major indices used in financial analysis and project

evaluation. In computing the internal rate of return, cash flows are usually considered as

deterministic variables. However, in practice, future cash flows are sometimes uncertain.

Probabilistic cash flows lead to a non-deterministic rate of return and different risk

characteristics.

Past research mostly focuses on practical approaches for computing the internal rate of return

of probabilistic cash flows by numerical approximation. For example, Hillier‘s method is

originally the solution for the probabilistic net present worth, which is converted to the

probabilistic internal rate of return by conditional probability. Another numerical

approximation method applies the second moment of Taylor Series in the computation. The

past numerical approximation methods are limited to normally distributed cash flows and

extensive calculations are needed. Some attempts have been made to solve for the exact

probability density function of the internal rate of return for short-term projects by limiting

some of the cash flows as constant variables. However, assuming some constant cash flows

does not completely represent the real applications.

This paper presents the exact solution of the probability density function of the internal rate

of return in the case of a two-period problem, which consists of an uncertain current

investment and two periods of cash flows. Moreover, the presented exact solution is

compared with simulation result for verification. The random cash flows are modeled to be

uniformly and independently distributed. Due to the complexity of the internal rate of return‘s

equation and the exponentially growth in the number of solution intervals, deriving for an

exact solution beyond three periods becomes impractical. Therefore, the next section

provides the study of the probabilistic internal rate of return by simulation method for higher

number of periods and various probability distributions of cash flows. Specifically, the paper

focuses on the relationship between the internal rate of return and four key factors, i.e. the

number of periods, input probability distributions, standard deviation and mean of cash flows.

This study emphasizes on understanding of how the change in parameters of cash flows will

affect the internal rate of return on investment and its risk characteristics.

Page 74: Abstract Proceedings of IEEE SCAP 2013

67

The 2013 IEEE Student Conference on Senior Capstone Project

Factors Analysis of Bayesian-based Pricing Prediction Markets

Nattanont Olarnthatchanant

Advisor: Daricha Sutivong

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Prediction markets are methodologies created for making predictions on the outcome of a

specific event of interest of the market maker. The market aggregates the information from

the crowd by means of trading contracts in the market. Trading patterns or actions of traders

(buy, hold, or sell) are mechanisms that move the market price of the contract and the current

market price, therefore, can illustrate the probability of the event. There are many sound

underlying reasons that support the validity of this technique: (1) at each time, each trader in

the market can adjust their thoughts toward the outcome of the event by taking the market

price which reflects the overall belief of all the traders into consideration; (2) traders who

possess proper information and wisely reveal it, in a form of actions, to the market will be

incentivized; thus, it encourages the informed traders to give out information to the market;

(3) every trader has to be responsible for their decisions more than they do in traditional

information aggregation methods such as surveys or polls so that they will be prudent with

their actions. Evidence so far suggests that prediction markets generally outperform the

forecast from expert groups, polls and many other conventional methods of forecasting.

Due to their effectiveness, prediction markets have become attractive to many people in

recent years across various fields. In general, there are two main ways to study the behavior

of prediction markets: (1) by experimenting in real markets and (2) through the use of

artificial prediction markets, virtual markets in which all the trades are done on a computer

and the traders are replaced with trading agents, computer-generated traders. Typically, the

pricing algorithm widely used in prediction markets is Logarithmic Market Scoring Rule

(LMSR) of which its cost and price functions are based on logarithmic functions. Despite the

wide utilization, LMSR markets still have a serious limitation. Specifically, LMSR markets

are ordinarily loss-making. This paper constructs an artificial prediction market as our

platform to study the characteristics of a newly proposed pricing algorithm named Bayesian

Market Maker (BMM). BMM markets apply the concept of conditional probability in

updating market price of the contract. This Bayesian-pricing prediction market can address

the significant drawback of LMSR market makers – it significantly lowers expected market

maker‘s loss at the same level of liquidity. It can also quickly converge when a shift in the

underlying true value of the contract occurs while its former version, zero-profit market

maker, fails to achieve this property.

In our model, we study effects of various parameters on the performance of BMM such as

how the mini-order size of each trade and the window size affect the speed of convergence.

In addition, this paper also studies how the number of traders and the traders‘ behavior of

accepting signals influence the convergence behavior of BMM. Finally, we propose new

simple means of adjusting the value of the market maker‘s variance in the jump case.

Page 75: Abstract Proceedings of IEEE SCAP 2013

68

The 2013 IEEE Student Conference on Senior Capstone Project

Optimizing the Liquidity Parameter of Logarithmic Market Scoring Rules Prediction

Markets

Suparerk Lekwijit

Advisor: Daricha Sutivong

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Prediction markets are widely-adopted techniques to aggregate dispersed public opinions via

market mechanisms and contract trading in order to obtain the wisdom of crowds and to

predict uncertain future event‘s outcome. This approach has been internally adopted by many

global companies and was externally implemented in various major events such as the

national elections. Many experiments have shown that prediction market‘s results are more

accurate than those of other traditional forecasting methods such as polling.

Various market mechanisms are used in prediction markets. One of the most commonly-used

mechanisms is Logarithmic Market Scoring Rules (LMSR). LMSR mechanism is relatively

simple and easy to implement; however, there are several design decisions to be made when

one wants to implement the LMSR market. These dilemmas include the setting of the

parameter ‗b‘ or ‗liquidity parameter‘, which plays a vital role in the market price calculation

and behaviors of the market including the market liquidity, the price adaptability, and the cost

to market makers. Since this parameter has significant effects on the market performance,

setting the value of the liquidity parameter in order to manage these tradeoffs is considered

more of an art than a science and, thus, few studies focus on how to set this parameter.

This paper aims to provide some comprehensible basis for the setting of the liquidity

parameter for those who want to implement LMSR prediction markets. We explore two

measurable factors that tend to be related to the value of the liquidity parameter. First, we

explore the effect of the liquidity parameter on the forecast standard error and the cost to

market makers, which will then be compared to the forecast standard error and the cost of

polling. Second, we study the relationship between the liquidity parameter and the

convergence step, i.e. the number of steps or the amount of time needed to make the market

price converge to the true value. Our experiments have been set up using artificial prediction

markets, the proposed simulation models that mimic real prediction markets. The simulation

results indicate that prediction market‘s forecast standard error decreases as the value of the

liquidity parameter increases. Moreover, we found that, for any given initial market settings,

there is always an optimal value of the liquidity parameter which yields the appropriate price

adaptability and makes the price converge to the true value most quickly. This optimal value

of the liquidity parameter enables the market price to quickly adapt to the true value without

causing excessive price swings when the price reaches the true value. From these findings,

market makers can effectively set the value of the liquidity parameter provided that the

standard error and the convergence step are their primary concerns.

Page 76: Abstract Proceedings of IEEE SCAP 2013

69

The 2013 IEEE Student Conference on Senior Capstone Project

A Comparison of Moving Average and Support Vector Machine

for predicting SET index

Poom Saowalukpan*, and Tanpitcha Chattrakoonpong

Advisor: Oran Kittithreerapronchai

Chulalongkorn University, Bangkok, Thailand

e-mail: [email protected], [email protected]

Because of inexpensive digital storage costs and insufficient resources, many organizations

have utilized large quantities of data to provide a better understanding of their businesses

using Machine learning, a self-learning computer algorithm that enables users to explore

underlying pattern from data and incorporates new data to achieve a better result. One of

interesting and recent developed Machine learning techniques is called Support Vector

Machine (SVM). Originally, it was developed as a classification technique that uses

mathematical equations to separate data according to their attributes or to predict outcome. It

was later improved to handle regression. As the technique becomes popular, practitioners

have applied SVM to various fields including financial area.

The purpose of this study is to compare SVM to Moving Average (MA), a popular and ease

of use technique, and to a generalization of MA using other relevant financial factors to

predict weekly averages index of Stock Exchange of Thailand (SET). The stock exchange

forecasting is complex and difficult because many factors influence the market. In this

research, we selected multiples factors which were reported by various studies to have

significant impact on SET index. Then we verified the influence of factors by statistical

process. After that, we proceeded by predicting SET index by MA and SVM. Prediction

results showed that SVM was slightly more accurate in predicting SET index provided that

SET index value in the prediction period does not vary greatly from the training period. It is

also shown that tuning and adjusting learning period of SVM can lead to substantially

improved results.

Page 77: Abstract Proceedings of IEEE SCAP 2013

70

The 2013 IEEE Student Conference on Senior Capstone Project

Thai Sign Language Recognition using Data Gloves and Motion Trackers

Ditsarin Vongasemjit*, and Jenwit Amonpongitsara

Advisor: Chotirat Ratanamahatana

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Sign language was invented as a means of communication between normal people and the

deafs. However, it still posts itself as a major obstacle since most people do not understand

sign languages. There have been some approaches that try to recognize and translate sign

language to help people understanding them. Most of the works are based on images

collected from a camera or a webcam. Nonetheless, the recognition results are not very good

and still impractical for actual communication since Thai sign languages are more complex

than many other sign languages; many of the Thai hand signs are very similar and very

difficult to be recognized. This has become our motivation to improve the Thai sign language

recognition by utilizing special devices—data gloves and motion trackers—which can record

the data much more precisely.

This project focuses on the recognition of Thai characters. Our team develops software which

will read data from both devices, find keyframes, and classify those keyframes into Thai

characters. For the keyframe finding, we use the mixed approach between threshold of the

change in the data received and the rate of change itself, which results in the accuracy of

75.67%. And for the recognition, we use K-Nearest-Neighbor (KNN) algorithm to classify

the class, which results in the accuracy of 86% when using K=3. The features used on KNN

depends on the hand sign's type as there are right-hand-only signs and both-hand signs. After

classification, we use Longest Common Subsequence (LCSS) to compare the output

sequence to the known dictionary words to help improve overall accuracy. Finally, the

program will display the classification result on the screen. The program is also capable of

displaying and recording the device's data for further usage. Our future improvement includes

the speed of the software, classification algorithm, and user interface.

Page 78: Abstract Proceedings of IEEE SCAP 2013

71

The 2013 IEEE Student Conference on Senior Capstone Project

Vision-Based Thai Sign Language Recognition

Jakapan Narkbuakaew*, Chanaporn Pikulnarkwong, and Chawintorn Pathomkasikul

Advisor: Chotirat Ratanamahatana

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Communication with hearing impaired people in Thailand is still be a big and challenging

problem. Most of them use sign languages as their major communication tool. However, it

can hardly be understood by normal people. Therefore, to enable better communication for

everyone, we are implementing a Vision-Based Thai Sign Language Recognition Project

which can help to resolve this problem.

Vision-Based Thai Sign Language Recognition Project is aiming to develop recognition

model in order to recognize Thai-sign language in a form of Thai characters, Thai vowels and

numbers via camera. In this project, the main image features is its contour. The contour can

be transformed into Time-Series which can be used to find the similarities among them. The

time series were separated into two groups. First is the time series generated from new input

data, and second is the time series in our database. In order to find the distance between two

time series, we use two main algorithms to do so. First is Euclidean Distance and second is

Dynamic Time Warping using with other optimization techniques to enhance the model‘s

performance. Finally, in our experiment, we found that the best algorithm for recognizing the

data is Dynamic Time Warping using 5% Sakoe-Chiba band constraint, and applying Wedge

and Lower Bound to speed up the method. The result of this model is 72.34043% correct, but

this model is not the fastest model to recognize the data. The fastest model we built is the

model using Euclidean Distance with Wedge. It use only 4.684 second to complete the cycle,

and the result is also high which is 63.82979%. However, these two models cannot recognize

some of the signs which is need to do more than one steps to represent the language. The

Longest Common Subsequence (LCSS) algorithm is then used to solve this problem, as it

could enhance our model and to perform a word-level recognition as well.

Page 79: Abstract Proceedings of IEEE SCAP 2013

72

The 2013 IEEE Student Conference on Senior Capstone Project

An Improved Traffic Incident Detection System Using Fuzzy Logic

Jaraspat La-inchua*1

Advisor: Sorawat Chivapreecha1 and Suttipong Thajchayapong

2

1Department of Telecommunication Engineering, Faculty of Engineering,

King Mongkut‘s Institute of Technology Ladkrabang, Bangkok, Thailand,

e-mail: [email protected] 2National Electronic and Computer Technology Center (NECTEC),

National Science and Technology Development Agency, Phatumtani, Thailand,

e-mail: [email protected]

Traffic incidents are usually not predictable in nature and also are the main causes of traffic

jam, they can increase travel delay and consume more fuel without producing any benefit. If

these traffic incidents can be early detected, the traffic operation and management personnel

will clear the roadway and return traffic condition to normal as fast as possible. Therefore, it

is a challenge to develop an intelligent transportation system (ITS) that can detect these

traffic incidents.

An algorithm to detect and classify traffic anomalies using only microscopic traffic variables

is proposed. This algorithm uses statistical calculations with Bayesian change detection

which are rather complicated for real-time application. Likewise, artificial intelligence has

been applied for anomalies detection. Artificial neural network (ANN) is used for traffic

incident detection but a lot of traffic data is required for training this algorithm. Fuzzy logic-

based traffic incident detection algorithm which uses four traffic variables as input is

proposed.

This paper also proposes traffic incident detection system using fuzzy logic-based but with

less computational complexity as only two traffic variables are used as the inputs to the fuzzy

inference system (FIS) that consist of mean speed and standard deviation of inter-arrival time,

called hybrid macroscopic-microscopic model. The proposed system uses fuzzy logic to

identify traffic status as normal and abnormal. Furthermore, majority voting with adjustable

window size is applied after the results are obtained from fuzzy logic to improve detection

rate (DR) and mean time to detection (MTTD). Performance evaluation results based on

simulation are shown that the proposed system has lower mean time to detection than the

previous system, while giving comparable detection rate to.

Page 80: Abstract Proceedings of IEEE SCAP 2013

73

The 2013 IEEE Student Conference on Senior Capstone Project

Load Consolidation Management System

Ratsuda Amatyakul*, Manavut Sunthonpaobvong,

Rapeepat Sukprasert and Sukritta Viryasiri

Advisor: Seeronk Prichanont

Chulalongkorn University, Bangkok, Thailand

e-mail: [email protected]

In this study, we consider a loading and routing problem for a logistics company in a less-

than-truck-load environment. The company receives orders from customers all over the

country, each consisted of various products with arbitrary demand quantity. Orders then are

divided into groups with respect to the regions of the country. For each group, with help from

transportation management software, a logistics planner has to make the loading decision, i.e.

which item has to be delivered by which truck. The loading decision then automatically

determines the routing of the truck. It is however found that the solutions obtained from the

software are usually not practical due to the fact that the software is meant for a full-truck

load environment. The study‘s objective therefore is to develop a decision tool that helps the

planner make better loading decision under the less-than-truck-load condition.

To achieve the goal, we develop a new transportation planning application that integrates

three components together. The first component is the data input module where user can

transfer the transportation orders from the company‘s legacy system to the transportation

planning application. The decisions are then made through the decision module. Appropriate

mathematical models are formulated and solved using CPLEX Optimization Studio. The

objective of the model is to minimize the transportation cost under conditions specified by the

company. The output is then fed to the output module by which the solutions are presented to

the user in a predetermined fashion. The input and output modules are written as a combined

GUI in C# language. The mathematical models have been validated by the company. The

results show that, with the new transportation planning application developed in this study,

the company can make better loading decisions in terms of transportation cost with

significantly less planning time.

Page 81: Abstract Proceedings of IEEE SCAP 2013

74

The 2013 IEEE Student Conference on Senior Capstone Project

Design, Analysis and Testing of Semi-Monocoque Frame for Student Formula

Kulanun Chutisemachai, Kasemphan Siriployngam, Kiatnathee Dilokthonsakun and Jintasarn

Sanchai

Advisor: Thanyarat Singhanart

Department of Mechanical Engineering, Faculty of Engineering, Chulalongkorn University

e-mail: [email protected]

Frame is one of the most important parts that provided the attachment points for the different

systems such as front and rear suspension and engine, etc. It can be separated into three major

types; a truss-like frame constructed by steel tubes, a monocoque frame supported load by

exterior skin, and a semi-monocoque frame with the combination of truss-like and

monocoque parts. The performance of the frame can be evaluated by its mass and torsional

stiffness. Team from Chulalongkorn University has participated in ―TSAE Auto Challenge

Student Formula Competition‖ for many years. Based on previous frame development, the

semi-monocoque frame with weight less than 30kg, torsional stiffness of more than 1200

Nm/deg, and good strength is re-designed in this paper. Therefore, the connection between

steel and monocoque part become more important and has to be studied before designing the

frame. From 2012 SAE rules & regulations, connection between steel and composite has to

support the tensile load of 30kN without fail. Firstly, many joints are designed and analyzed

under this load by Finite Element Method. One of them, double lab joint with two holes and

no clearance, is chosen to be tested and verified by the experiment. The experiment shows

that the carbon-fiber and KEVLAR fabric composite laminate has the Young‘s modulus of

113 GPa and strength of 285 MPa; the joint fails at 25 kN applied load with tension or tearing

failure mode.

The different between experimental and numerical results is due to the clearance between

composite and steel plate from manufacturing process. That is, the connection strength comes

from the strength of composite with holes and has no effect of clamped force. The connection

design can be improved by reducing the thickness of iron sheet, changing the alignment of

the bolts holes axis, or increasing the cross section of the composite plate. The changing the

alignment of the bolts holes axis with load capacity of 35 kN is used to design the semi-

monocoque frame by finite element method. From an analysis at maximum applied torque,

the maximum stress at connection is 147 MPa which is less than the yield stress of the

composite, the mass is 29.6kg, and the torsional stiffness of the frame is1760 Nm/degree. It

can be seen that the frame does not fail under maximum load and can achieved the

requirements of mass and torsional stiffness. Anyway, the frame should be tested to verify the

results for further research. It can be concluded that the connection for semi-monocoque and

the semi-monocoque frame can be designed to meet the objectives.

Page 82: Abstract Proceedings of IEEE SCAP 2013

Section 7

Energy and Transportation

Page 83: Abstract Proceedings of IEEE SCAP 2013

76

The 2013 IEEE Student Conference on Senior Capstone Project

Design of a Pole-Slipping Protection for Loss of Synchronizing of a Generator

Supachai Sabua, Veerayut Pojui

King Mongkut‘s University of Technology North Bangkok, KMUTNB

Bangkok, Thailand

E-mail: [email protected]

Transmission line faults in power systems are common. They result in electromechanical

oscillations called power swing. If the power swing is severe, it can lead to loss of

synchronism among the interconnected generators. This is referred to as pole slipping

condition. The voltage and current swing during a pole slipping condition can damage

electrical components in the power system. Unwanted operations of various protective

devices may occur. These could affect reliability of the power system causing power failures

in wide areas.

Relays require effective algorithms for fast and accurate detection of pole slipping condition.

This project was focused on the development of a relay that contained simple and effective

pole slipping techniques. The relay was capable of detecting loss synchronism condition in a

generator. To achieve this, the algorithms were first developed in MATLAB and then tested

with the signals derived from PSCAD/EMTDC simulation. The pole slipping algorithms

proposed were based on impedance detection in R-X diagram and equal energy area in time

domain. The algorithms were then implemented on a hardware based on DSPIC

microcontroller. Finally, the relay hardware was tested with analogue signals from

OMICRON.

Page 84: Abstract Proceedings of IEEE SCAP 2013

77

The 2013 IEEE Student Conference on Senior Capstone Project

Mathematical modeling of radiant cooling room with radiation

Jarudon Pattarabancha, Jetarin Chareonpornpoj

Advisor: Kuntinee Maneeratana

Chulalongkorn University, Bangkok, Thailand, e-mail : [email protected]

Lately, the air conditioning system by radiation cooling [1], which has lower energy

consumption, has been used in some modern buildings, such as Suvarnabhumi Airport . The

objective of this project is to simulate the radiation in a radiant cooling room using

commercial computational fluid dynamics software Ansys Fluent. The model combines heat

transfer with natural convection and radiation in a rectangular enclosure-like room. The

boundaries are walls with various condition such as isothermal, constant heat flux or

adiabatic. If the air is cooled by the radiant cooling panel, generally under isothermal

condition, the air density changes due to the change of temperature. The air flow is generated

by the buoyancy force, causing natural convection in the room. Therefore, natural convection

has to be included into the model as well. Since the boundaries are large with the Rayleigh

number exceeds 109, the air flow is considered as turbulent.

The model must be verified with available experiment or numerical data. The results are

compared with experimental and numerical study of laminar and turbulent natural convection

and radiation in rectangular enclosures with aspect ratio of 1, 0.5 and 2. Then, the case

studies with and without radiation are simulated with three aspect ratios of 0.5, 1 and 2. The

room dimension H × L is 3 × 6 m2, 3 × 3 m

2 and 6 × 3 m

2 respectively. The 10°C isothermal

cooling panel is placed at ceiling while other walls gain 25 W/m2 constant heat flux. The

result clearly shows that rooms with radiation have lower temperatures. The average

temperature differences are 10°C, 5°C and 2°C for the rooms with aspect ratios of 1, 0.5 and

2, respectively.

Page 85: Abstract Proceedings of IEEE SCAP 2013

78

The 2013 IEEE Student Conference on Senior Capstone Project

Line current time shifting behavior when fault occurred on branch circuit of radial

distribution system by using ATP program

Swanya Sittiprasertphol

Advisor: Weerapun Rungseevijitprapa

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Distribution system is served to send energy for home, industry and commercial building.

The activities of energy users are interrupted during the fault. Nowadays, single line to

ground fault and down conductor fault with high impedance is hardly detected in

conventional protection system. This project is studied on the line current behavior at

substation when fault occurred on branch circuit of radial distribution system. From thesis

proposed time shifting method to detect broken distribution conductor and touching ground

on the load side. Therefore, this project analysis line current behavior by using time shifting

method. Distribution system model is designed for these 2-type of faults simulation in 22 kV

systems, with 14 kilometers long feeder line and branch circuit distributed along the feeder

line. These 2-type of faults are simulated by ATP program and then analyzed period time by

MATLAB program.

According to the result of both types, the period of line current is shifted from the ordinary of

20 ms during fault. For single line to ground fault, the period in faulty phase is the highest

hange in longer than 20 ms, whereas the other two phases are slightly shorter and longer.

For down conductor fault, show the 3-phase line current when down conductor fault with

high impedance occurred at phase A and the characteristic of period of line current at the first

wave when fault occurred in each phase is shown.

Page 86: Abstract Proceedings of IEEE SCAP 2013

79

The 2013 IEEE Student Conference on Senior Capstone Project

Analysis of Voltage Fluctuation based on Thai Utility Regulation

Paranee Chailungka

Advisor: Thavatchai Tayjasanant

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Voltage fluctuation is one of power quality problems in a power system. Voltage fluctuations

are caused by varying loads in a power system. Any load with significant current variations,

especially reactive component, can result in voltage fluctuations. Examples of varying loads

are arc furnaces, electric welders and starting-up large induction motors. Wind turbines are

also considered as voltage fluctuation sources in a power system. Voltage fluctuations can

cause flicker which can be measured using a flickermeter. IEC 61000-3-7, Edition 2.0

Electromagnetic compatibility (EMC)–part 3 discusses an assessment of the emission limits

for fluctuating loads in MV and HV power system. If voltage fluctuations are very high,

voltage regulation problems can happen. Therefore, mitigation methods are required for

voltage fluctuations.

This project develops a computer program using Graphic User Interface (GUI) feature in

MATLAB environment to analyze voltage fluctuations based on Thai utility regulation for

commercial and industrial customers. Relative maximum voltage change (dmax) is primarily

used in Thai regulation and is calculated for the analysis by the developed program.

Evaluation status of voltage fluctuation level is shown on the screen if varying loads are

allowed to connect to the power system or not. An example in the Thai utility regulation is

used to verify the correctness of the develop program. The developed program can also

calculate relative maximum voltage change from voltage waveforms obtained from actual

field measurements. For the mitigation technique, the developed program can suggest the

modified operation characteristic of varying loads in order to be able to comply with Thai

utility regulation. The developed program is user-friendly, flexible and useful for customers

to evaluate voltage fluctuation level required by Thai utility regulation.

Page 87: Abstract Proceedings of IEEE SCAP 2013

80

The 2013 IEEE Student Conference on Senior Capstone Project

Reducing Electricity Costs in the Tire Manufacturing by Load Shifting Technique and

Using High Efficiency Motors

Aleena Sakulsanchartthai

Advisor: Hadsakoon Boriphonmongkol

Chulalongkorn University, Bangkok, Thailand

This project focuses on the reducing electricity costs in the manufactory by load shifting

technique and using high efficiency motors. The manufactory used as the case study in this

project is Otani Radial Limited. This manufactory calculates electricity costs by Time of Use

Rate. The study analyzed and evaluated the results by Energy Index and the worth of

economic investment. The analytical results show that load shifting together with using high

efficiency motors can actually reduce electricity costs and Energy Index. The worth of

economic investment determined by Payback Period, Net Present Value and Internal Rate of

Return suggests that the investment is worth. The proposed approach makes it possible to

reduce the electricity costs, leading to lower production cost. It can be applied for use in other

manufactories whose the production procedure is similar to Otani Radial Limited.

Page 88: Abstract Proceedings of IEEE SCAP 2013

81

The 2013 IEEE Student Conference on Senior Capstone Project

Design and Construction of a Test Rig for the Cable-Pulley

Power Transmission System

Pao Wichayaketsophorn, Poompat Pornwissanukul,

Songkran Taepaisitpong, Suksit Wongwanich

Advisor: Phongsaen Pitakwatchara

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Mechanical transmission system allows actuators to transmit power to other structures. Each

type of mechanism has its own limitations, such as backlashes, stiffness, and friction, which

affect the system‘s performance. A cable-pulley transmission has very high transmission

efficiency with zero backlash motion and high stiffness.

The objective of this project is to design, construct, and control an experimental apparatus

which consists of 3 main parts, a driving pulley, a driven pulley, and cables. The driving

pulley is attached to a motor and the position of the driven pulley is measured and compared

with those of the driving pulley using an encoder for each pulley. This one degree-of-freedom

apparatus has the diameter ratio of 3:1 between the driven and driving pulleys. For this

equipme

the home position.

In the testing, we design a controller based on a step input for the desired angle of the driving

pulley, aiming to get a zero steady state error and a small settling time with an acceptable

overshoot. From the dynamic model of this system and the expected response conditions, a

lead compensator is used as a controller and directly applied to the motor in order to drive the

pulleys. The driving pulley is rotated according to the input from the motor to the desired

angles; the driven pulley thus rotates to a certain angle relative to the driving pulley‘s angle,

depending on tension of the cables. Data of actual angles from both encoders are collected

and plotted as degree-time graphs with varying tension. From the graphs, it was found that

higher tension, the higher stiffness, allowing two pulleys to rotate without slipping. At low

tension, there are some backlashes.

Page 89: Abstract Proceedings of IEEE SCAP 2013

82

The 2013 IEEE Student Conference on Senior Capstone Project

Design of a stator ground fault protection for a generator directly connected to a busbar

Sathichai Padungsiripanich, Kailerk Ruangsakul

King Mongkut‘s University of Technology North Bangkok, KMUTNB

Bangkok, Thailand, e-mail: [email protected]

Generators are one of the most important components in power systems. They produce

electric power to all components in the networks. Faults in generators cause severe damage to

the generators‘ windings. This may result a wide spread black out and long supply

interruption time.

A differential relay is used to protect stator windings. If there is a neutral grounding reactance

or resistance installed at the neutral point of the generator, the relay can protect only 90 –

95% of the generator winding when there is a ground fault. This cause 5 – 10% of the stator

winding unprotected. The winding can be damaged because the relay is unable to detect and

separate the fault.

The objective of the project is to extend the protection coverage to 100% of the stator

winding. By using a phasor comparison technique, the protection zone starts at the busbar

that the generator is connected and ends at the neutral point of the stator winding. The

technique was simulated using PSCAD/EMTDC. Then the signals from PSCAD were

imported to MATLAB in order to test the algorithm. The algorithm was then transfer to a

microcontroller board (dsPIC33FJ128GP708) to make a relay. The ground fault relay that

covered 100% of the stator winding was tested using the signals generated from OMICRON

Page 90: Abstract Proceedings of IEEE SCAP 2013

83

The 2013 IEEE Student Conference on Senior Capstone Project

Feasibility Study and Human Machine Interface (HMI) for 'Automatic Control and

Monitoring of a Micro Hydro Power Plant'

Isuru Mudannayake*, and Anuradha Colombathanthri

Advisor: Erik Bohez

Asian Institute of Technology (AIT), Pathumthani, Thailand,

e-mail: [email protected]

Micro Hydro Power Plants (hereafter mentioned as MHPP) are small systems that are used to

produce energy from falling water such as rivers and streams. These types of power plants are

ideal to use in remote areas which do not have access to the main power grid to use electricity

and also even at the accessible areas, these can be used since MHPP can supply the excess

power back to the grid. When excess power is supplied back, the grid acts as infinite power

storage and during an electricity shortage grid can supply power to the MHPP and then

distribute around the village.

Most of the times, controlling and monitoring are jobs done manually by the experienced and

skilled personals. This may be impractical when considering the cost. To monitor 50 MHPPs,

50 skilled labor is required, so that there will not be a profit overall. To eliminate this

problem and also to make the operation more efficient, implementation of a monitoring and

automatic controlling system will be required. So that all these 50 MHPPs can be operated by

one centralized monitoring system, which reduces the need of 50 skilled personals to just one

skilled person. This person can monitor the state of function of all the MHPPs at once and if

some error occurred, that particular MHPP can be stopped functioning till someone go to the

site and fix the error. That means this system can be really efficient and the safety of the

MHPP is also assured.

Doing a feasibility study to automate a manually operating MHPP and developing a

prototype monitoring and control system for a MHPP using PC controlled Human Machine

Interface (HMI) are the objectives of the project. This prototype will be done on the model

MHPP at the Water Lab in AIT. The targeted MHPP for the feasibility study is Ban Khun Pae

MHPP in Chiang- Mai, which is capable to produce 90kW and which is currently in the non-

working state.

Page 91: Abstract Proceedings of IEEE SCAP 2013

84

The 2013 IEEE Student Conference on Senior Capstone Project

Analysis of Electric Power Quantities under Sinusoidal, Nonsinusoidal, Balanced or

Unbalanced Conditions

Phonlawat Wannasiwaporn

Advisor: Thavatchai Tayjasanant

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Non-linear loads generate harmonics which can cause power quality problems in electrical power

systems. Harmonics, which are components with integer multiple of fundamental 50-Hz

frequency, can cause malfunction in protective devices, damages in electrical devices, lower

electrical machine efficiency, and insulation life-time reduction. In some cases, high magnitudes

of voltages and currents can happen due to harmonic resonance phenomena. Definitions of

electric power quantities have been under discussion for many years when harmonics exist in a

power system. Conventional definitions may not apply for various cases. Recently, IEEE Std.

1459-2010, Standard Definitions for the Measurement of Electric Power Quantities under

Sinusoidal, Nonsinusoidal, Balanced or Unbalanced Conditions, has been approved by IEEE.

Electric power quantities are separated into fundamental and non-fundamental components with

the help of effective current and voltage concept. Harmonic pollution and load unbalance are also

proposed in this standard. One-phase and three-phase cases are discussed.

This project develops a computer program using Graphic User Interface (GUI) feature in

MATLAB environment to calculate electric power quantities based on IEEE Std.1459-2010 and

evaluate harmonic level based on Thai harmonic regulation. The developed program can also

calculate electric power quantities from voltage and current waveforms obtained from actual field

measurements. Evaluation status of harmonic level is shown on the screen if non-linear loads are

allowed to connect to the power system or not. Analysis in various stages are carried out and

displayed. Results from the developed program are verified with the standard and journal paper.

Result analysis can be used as a guideline to improve harmonic level.

Page 92: Abstract Proceedings of IEEE SCAP 2013

85

The 2013 IEEE Student Conference on Senior Capstone Project

Impact of the Distributed Generator to Protection on Distribution Power System

Tanyaluck Anusin

Advisor: Hadsakoon Boriphonmongkol

Chulalongkorn University, Bangkok, Thailand

This project presents the impact of the distributed generator (DG) on the coordination

between protective devices in distribution systems, which emphasizes on the impact to the

magnitude of short-circuit currents when symmetrical three-phase faults occur in the power

system. It uses the Matpower 4.0 to computes short-circuit currents, assumes sizes and

positions of DG and coordinates between protective devices before faults occur, which

consist of circuit breaker, recloser and fuse. The connection of DG affect to the reduction of

reach and miscoordination for some case. In case of fault occurs behind recloser and connects

DG behind recloser, fuse has the short-circuit current more than recloser, make the fuse

works before fast recloser and causes the temporary fault become permanent fault. When the

size of DG is large, the short-circuit current at recloser is down but the short-circuit current at

fuse increase. The result for each cases can change depend on the system.

Page 93: Abstract Proceedings of IEEE SCAP 2013

86

The 2013 IEEE Student Conference on Senior Capstone Project

Application of Phasor measurement Unit for Power System Voltage Stability

Monitoring

Puriwat Suttitham

Advisor: Naebboon Hoonchareon

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Voltage stability problem can lead to blackout. Generally, the analysis of voltage stability

problem is rather complex. This project proposes application of phasor measurement unit for

power system voltage stability monitoring, intently to be a real-time monitoring tool to assist the

system operators. The application employs Maximum Power Transfer principle to determine

Impedance Stability Index. Additionally the Graphics User Interface to update the plot of

Impedance Stability Index as well as V-P characteristic with respect to time at every 0.1 second is

included. Performance of this application has been verified using 2 test systems, that is the single

machine infinite bus and the 4-generator 2-aera test system by applying a continuous load

increasing. The simulated test results show that Impedance Stability Index can indicate voltage

stability problem and hence confirm the usefulness for on-line voltage stability monitoring.

Page 94: Abstract Proceedings of IEEE SCAP 2013

87

The 2013 IEEE Student Conference on Senior Capstone Project

Forecasting and Hedging Natural Gas Price for Hedging Electricity Price in Thailand

Sanyapong Petchrompo

Advisors: Chonawee Supatgiat, Surapong Sirikulvadhana

Chulalongkorn University, Bangkok, Thailand, e-mail: [email protected]

Due to the great strides along the path to industrial development in Thailand, it is undeniably

true that electricity has been playing a vital role as a main source of energy for Thai industry,

accounted for not less than 60000 GWh a year. Electricity price for Thai industrial consumers

is very volatile that can dramatically change by 13% in 4 months. Such electricity price risk

cannot be managed or controlled because Thai electricity market is regulated with a single

seller. Hence, it is very difficult for Thai industry to maintain competitiveness in the global

arena where its major competitors in other countries can manage their electricity price risk

via deregulated electricity systems. To allow factory owners to mitigate electricity price risk,

Electricity Generating Authority of Thailand (EGAT), Metropolitan Electricity Authority

(MEA), and Provincial Electricity Authority (PEA) may provide fixed-price electricity

contracts as an alternative choice. To do so, EGAT need to forecast and hedge their fuel costs

in order to avoid fuel price risk exposure.

This paper, therefore, illustrates how to forecast and hedge natural gas prices, which represent

approximately 70% of fuel consumed in Thai electricity generation. This study is undertaken

based primarily on their historical expenditure on both fuel consumption and electricity

purchasing from private power producers between 2008 and 2011. In Thailand, PTT Group is

the sole provider of natural gas to EGAT. The purchasing contracts between the two entities

require different types of external variables, e.g. crude oil prices, exchange rates, and

consumer price indexes, to determine the purchasing price. To deliver the solutions, multiple

linear regression is proposed to determine the relationship of the related variables to natural

gas prices.

Having conducted this research, it can be readily seen that the results of which can accurately

forecast natural gas prices and appropriately calculate the amounts of crude oil and currency

futures contracts, which significantly reduce EGAT‘s fuel price risk when offering fixed-

price contracts. The implication is that EGAT can use this research as a guideline, integrated

with their confidential data, to effectively forecast and manage their natural gas expenditure.

Last but not least, the results of this study will be used as a major component of the extended

fixed-price project, which will demonstrate how to strategically develop the fixed-price

structure of Thai electricity supply chain.

Page 95: Abstract Proceedings of IEEE SCAP 2013

88

The 2013 IEEE Student Conference on Senior Capstone Project

A New Single Solar Tracking System for Concentrating Linear Fresnel Reflectors

Nopharut Laopornpichayanuwat, Sunhaporn Sribanditmongkol, and Sukrit Chuntaravisuid

Advisor: Supachart Chungpaibulpatana

Sirindhorn International Institute of Technology, Thammasat University, Pathumthani,

Thailand.Tel. 086-800-6114 e-mail: [email protected]

Solar thermal energy is a sustainable source for power generation especially in the region

where sun radiation is high. A type of solar collectors called Concentrating Linear Fresnel

Reflector (CLFR) is used for power plants in many countries. CLFR is suitable for power

generation due to its simplicity, sturdiness, efficient use of land, and relatively low

construction cost. Thailand, being situated in a tropical region, has a condition that can

sustainably utilize solar energy for power generation. However, the cost of foreign CLFR

system is still very high for Thai market. Therefore, lowering the construction cost can make

CLFR system becomes practical for Thailand and encourages the use of this sustainable

technology for power generation in long-term.

CLFR operates by using reflecting mirrors panels to focus and ‗concentrate‘ the sunlight onto

a linear receiver absorber located above the panels. As with all concentrating collectors,

CLFR requires a tracking system in order to operate efficiently. The tracker is typically

installed to each individual reflecting panel resulting in high cost of the whole tracking

system. This project aims to reduce the cost of tracking for CLFR by developing an automatic

system that is able to control the tilt angles of all reflecting panels by a single control unit.

The experimental CLFR model used in this project consists of fourteen 1-m long and 0.1-m

wide reflecting mirror panels. The tracking system for CLFR developed in this project is

composed of three parts: 1) a sensory device 2) an Arduino UNO R3 microcontroller and 3) a

mechanical linkage system with a servo motor.

The sensory device made of four LDRs (photo-resistors) is used to determine the solar

incident angle. Each LDR senses the difference in light intensity and sends the signal to

microcontroller. The relationship between solar incident angle and panels‘ tilt angles is a

basis for the processing of a microcontroller. A mechanical linkage system links all reflecting

panels together and rotates them simultaneously once the servo motor rotates. A servo motor

receives signal from the microcontroller. The system keeps sending and receiving signal

repetitively until all reflecting panels can reflect the sunlight properly onto the absorber.

The CLFR tracking system developed in this project is able to reflect sunlight into the

absorber area automatically using only one motor. The accuracy of the system is measured in

terms of a percentage of sunlight reflected within a desired boundary which represents the

absorber area. This project keeps the cost of the tracking system low by utilizing only

one motor to control all reflecting panels. Consequently, the CLFR becomes more practical

and suitable to be a sustainable technology for future.