faculty of engineering and architecturestaff.aub.edu.lb/~webfea/feasc/previous/proceedings.pdf ·...

Faculty of Engineering and Architecture American University of Beirut

Proceedings

June 6-7, 2002

Bechtel Engineering Building Beirut, Lebanon

Ibrahim N. Hajj Samer M. Abdallah (Eds.)

1

©2002 American University of Beirut Press

Published by American University of Beirut Press American University of Beirut Beirut, Lebanon

All rights reserved. This publication may not be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, or otherwise, without t he prior permission of American University of Beirut Press.

The findings, interpretations, and conclusions expressed in this proceeding are entirely those of the authors and should not be attributed to the American University of Beirut, to its affiliated organizations, or to members of its board of trustees.

For permission to reprint a page or more of this publication, submit a Permission Reprint request form available by fax at 961-1-361-091 or e-mail [email protected]

Printed in Beirut, Lebanon

2

FEASC 2002 Organizing Committee

Samer M. Abdallah Riyad Chedid

Ibrahim N. Hajj (Chair) Mounir Mabsout

Fuad Mrad Mohamad Rawas

Alan Shihadeh Maher Salameh

FEA 50th Anniversary Committee

Samer M. Abdallah Samir Abou Samra

Sami Alamuddin George Ayoub

Ibrahim N. Hajj (Chair) Raja Iliya

Samir Kadi Ayman Kayssi

Ibrahim Khoury Leila Musfy

Maher Salameh

FEA 50 th Anniversary Website: http://www.aub.edu.lb/fea/50

FEASC 2002 Website: http://www.aub.edu.lb/fea/feasc

3

Welcome to FEASC 2002

This academic year the Faculty of Engineering and Architecture at AUB is celebrating its 50th Anniversary. As part of this celebration, the 1st FEA Student Conference was initiated. One of the aims of the Conference is to provide a forum at which students present and display their projects, research results, and artwork. Another aim is to invite some of the former students, or alumni (we'll use the term alumni to include both male and female former students), to share their professional experiences with us. A third aim is to bring together students, alumni, faculty, family, and friends to learn about what FEA and some students and alumni have be en doing, and to strengthen the bonds that connect us all.

One of the statements in the AUB mission states that AUB "strives to educate those who will be leaders in public affairs, academia, and the private sector." This has been proven true time and time again, especially in Engineering and Architecture.

Over the past 50 years, FEA has graduated more than 6000 alumni. A large number of them have distinguished themselves through their careers and through their contributions to the profession and to the society. Five of the alumni, out of the many of the alumni who distinguished themselves, have been selected this year to receive the FEA Distinguished Alumnus Award and to give plenary talks at the Conference. Other alumni who have been selected were not able to attend because of other commitments. The selection task has not been easy, especially with the large number of eligible distinguished alumni that grew during the past 50 years. But we plan to catch up by having this Conference become an annual event.

I do hope that you will join us in making the FEASC 2002 a great success and a truly memorable experience for all.

Ibrahim N. Hajj, Dean Faculty of Engineering and Architecture American University of Beirut Beirut, Lebanon

4

Table of Contents

FEASC 2002 Organizing Committee.…………………………………………………………………….2 FEA 50th Anniversary Committee................................................................2 Welcome to FEASC 2002...........................................................................3 Table of Contents.....................................................................................4 50 YEARS OF EDUCATION.........................................................................6 Raja Iliya, George Ayoub, Albert Kuran and Ibrahim N. Hajj The FEA Distinguished Alumnus Awards.....................................................11

Nabil F. Azar ................................................................................................................................11 Milhim Joseph Ghulmiyyah ....................................................................................................12 Adib Kanafani..............................................................................................................................13 John Makhoul ..............................................................................................................................14 Hussein Rifaï................................................................................................................................15

The FEA Distinguished Scholar Awards……………………………………………………………….16 Thomas Kailath………………………………………………………………………………………………….16 Breaking the 0.1 Barrier in Optical Microlithography via Signal Processing…..18 Technical Program – Schedule ..................................................................19

Thursday, June 6, 2002 ................................................................................................................19 Friday, June 7, 2002 .....................................................................................................................19

Technical Session I – Computer Vision, Signal Processing ............................................................20 Technical Session II – Computer and Communication ...................................................................20 Technical Session III – Electrical, Computer, Communication ......................................................21 Poster & Demonstration Session .....................................................................................................21 Technical Session IV – Mechatronics, Bioengineering...................................................................21 Technical Session V – Mechanical, Electrical.................................................................................22 Technical Session VI – Civil, Transportation, Education................................................................22

Civil engineering and novel technology: Where do they meet? ..................................................22 Cultered Engineers: A Necessity or a Privilege...........................................................................22

Graphic Art Exhibition of Fine Art Prints.....................................................23 Human Powered Vehicles.........................................................................23 The 50th – Structures Contest 2002...........................................................24

DISCORD ....................................................................................................................................24 ENLIGHTENMENT....................................................................................................................24 IL CONSTRUCTURA.................................................................................................................24 MAC CIVIL .................................................................................................................................24 MACSI.........................................................................................................................................24 THE CIVILIANS.........................................................................................................................24

Final Year Projects (FYP)..........................................................................25 FYP’s – Department of Electrical & Computer Engineering ..........................................................25 FYP’s – Department of Mechanical Engineering ............................................................................29

Presented Papers Iris Recognition......................................................................................32 Carine Daouk, Lina El-Esber and Fouad Kammoun Beat The Traffic......................................................................................36 Houda Karaki, Jihad Ibrahim and Hassane Kamal Perspective Image Formation Using Omnidirectional Vision...........................40

5

Adel T. Zouheir y and Mounir M.C. Doumani Speech Recognition for Voice Based Control...............................................44 Youssef Naous, Ghinwa Choueiter and Nesrob Ohannessian Mobile.Music..........................................................................................48 Joe Abou-Rjeily, Jade Bissat and Mario Harik Mobile Jammer.......................................................................................50 Paula Atallah, Najwa Hamzeh and Elias Nahra “SMS Shell”...........................................................................................52 Rima Daoud, Jean Ghanem and Nayla Hamadeh MSIS: Mobile Student Information System.................................................55 Hicham Fadel, Habib Haddad and Hady Moussa Above and Beyond the Wires W.A.M..........................................................57 Ayman Chalhoub, Mark Ghibril and Wissam Haddad GIS Based Computerized Traveling Guide..................................................59 Chucri Abi Haidar, Mossine Koulakezian and Fadi Abou Ghantous Common Sense Controller (CSC) using LabVIEW........................................61 Sima Azar and Sandra Dandach Linearization of a D/A Converter...............................................................65 Michel M. Azarian, Alain M. Hani and Roland N. Saad The Design of a Prosthetic Glove for a Child with Cerebral Palsy....................67 Tarek O. El-Chidiac and Jawad T. Sabbah Development of a Computer- Based Measurement Station............................71 Marianne Sarkis, Michel Sarkis and Joseph Malkoun CNC Retrofit for Milling Machines..............................................................74 Imad Doumit, Georges El-Mir and Makram Nehme Hoover Saucer “Defying Balance”..............................................................76 Selim Hobeika, Nael Itayem and Marc Tabchy A Comparative Assessment within a Multi-grid Environment Of the Performance of Segregated Pressure-Based Algorithms for Fluid flow at All Speeds..................................................................................................79 Daniel Asmar CHOCOOLER©........................................................................................84 Fadi Hachache, Ibrahim Geha and Abdul Karim Younes Human Electric Locomotive Project “HELP”.................................................87 Raya Abdel Baki, Antoine Abboud and Chadi Chaker Capacitor Placement Using Genetic Algorithm and Distributed Computing...........89 Francois Layoun, Maher Salameh and Raji Sayegh Geophysics and Engineering Hand in Hand.................................................91 Nadim Baalbaki and Jouanna Doummar Potential for Rail Freight in Lebanon..........................................................93 Rawad Hani Civil Engineering and Novel Technology: Where Do They Meet?....................95 Loai Naamani Cultured Engineers: A Necessity or a Privilege............................................97 Loai Naamani Author Index..........................................................................................99

6

President Chamoun's & Dean Weidner Inauguration Ceremony

Bechtel Building 1952

50 YEARS OF EDUCATION

Dr. Raja Iliya , Professor Emeritus, Dr. George Ayoub , Professor of Civil and Environmental

Engineering, Prof. Albert Kuran, Associate Professor of Mechanical Engineering, and Dr. Ibrahim N. Hajj , Dean of the Faculty of Engineering and Architecture

This academic year the Faculty of Engineering and Architecture (FEA) at the American University of Beirut is celebrating its 50th anniversary. The Faculty was established in 1951 and a new building donated by Stephen D. Bechtel was constructed to house it. However, as early as 1913 the American University of Beirut recognized the need for engineering education in the Arab World and, consequently, a program leading to the degree of Bachelor of Arts in Engineering was established within the School of Arts and Sciences. The courses offered were in the field of Civil Engineering, and were covered during the Junior and Senior years. The Department of Engineering was located in Bliss Hall, room 108, where all of the engineering courses were given and, by 1944, sufficient additional courses had been added to permit conferment of the degree of Bachelor of Science in Civil Engineering.

The first class in this program graduated in June 1945, and the last class in June 1954. By that time a separate School of Engineering had been established in 1951 under the Deanship of C. Ken Weidner, and a new building donated by Stephen D. Bechtel was constructed to house it. This building, which was named the Bechtel Engineering Building, after its donor, was inaugurated on April 16, 1954, under the auspices of his Excellency Mr. Camille N. Chamoun, President of the Lebanese Republic, and in the presence of Dr. Constantine Zurayk, Acting President of AUB, and Mr. Stephen Bechtel. Four years curricula were initiated in civil engineering, mechanical engineering, electrical engineering, and architectural engineering but the years between 1951 and 1954 were a transitional period of continuous development toward the new curricula that were completed in 1954. In 1962 Raymond S. Ghosn, Professor of Architecture, was appointed to succeed Dean C. Ken Weidner. In 1963, a five-year program leading to the degree of Bachelor of Architecture was introduced, replacing the Bachelor of Architectural Engineering program, the last class of which graduated in June 1966. In that same year the School was renamed the Faculty of Engineering and Architecture and was reorganized to comprise four departments: the Department of Architecture, and

7

Surveying Camp - Bhamdoun Village 1953

the Departments of Civil, Electrical, and Mechanical Engineering. Since then the curricula have been under constant review with changes introduced whenever deemed necessary. The first programs leading to a Master's degree in Civil, Electrical and Mechanical Engineering were introduced in 1962 and other programs have been added to help meet the growing demand for advanced engineering education and to promote physical planning to serve social ends. After the untimely death of Dean Raymond Ghosn on February 17, 1976, Professor Robert Sloane was appointed Acting Dean for a period of six months followed by six month rotating acting deanships from among the chairmen of departments namely, the late Professors Thomas Bridgewood, Khosrof Yeramian and Edward Hope. In 1977 Professor Kanaan Kano was appointed Dean of the Faculty and served till 1982 when the late Associate Professor Henri Madani took over as Acting Dean until 1986, the year Professor Nassir Sabah was appointed Dean of the Faculty of Engineering and Architecture. Dean Sabah served until the year 1999 when he was succeeded by Professor Mohamed Harajli as Acting Dean of the Faculty. In the year 2000 Professor Ibrahim Hajj was appointed as Dean of the Faculty and has been serving in this capacity up to the present. During the tenure of Dean C. Ken Weidner, the Faculty of Engineering and Architecture grew tremendously in enrollment, laboratory facilities, and physical plant. The present University power plant was initially built for use by the students in the Departments of Mechanical and Electrical Engineering, as well as to provide electric power and hot water to the University as a standby unit. The diesel generator sets, switchgears, and boilers were donated through the efforts of Dean Weidner. During the same period three additional buildings were constructed to house mainly the shops (wood, metal, and welding), and laboratories of the Electrical and Mechanical Engineering. Also Sanitary Engineering and Hydraulic laboratories were built and fully equipped. It is worth noting that most of the original laboratory and shop equipment were acquired through the persistent effort of Dean Weidner and the generosity of American and European donors. In 1962 a graduate program in Sanitary Engineering was introduced. In keeping with international trends, the program was renamed in 1973 as Environmental Engineering. The year 1978 witnessed a freezing of the program because of the devastating events in the country. The program was unfrozen at the conclusion of the civil strife in 1992.

An important aspect introduced into the program as early as 1952 was the establishment of a summer camp where all students of the Faculty, after completing the first Year, spent ten weeks studying surveying and the use and operation of building construction equipment such as traxcavators, bulldozers, and road scrapers. Besides the academic benefits, the students had the opportunity to experience the conditions of community camp living by sleeping, eating, and studying under the same tents as their professors. Undoubtedly, this life created and strengthened the school spirit and exposed the students to the hardships of real professional life. The first real camp site was started in 1954 in Zgharta, and was moved in 1958 to the grounds of Shwayfat National High School (Mr.

8

Using plane table - 1952

Using total station - 2001

Bechtel Building - 1992

Charles Saad). In 1960 a permanent camp set-up was established in Mazboud in the Chouf District. Due to a change in the curriculum the surveying course became restricted as a requirement to Civil Engineering majors only. This resulted in the closure of the camp in 1970, and the surveying course has since been taught on the Main Campus.

Since the early 90's an ambitious program of renovation and construction has been undertaken. The fourth floor in Bechtel building was expanded and remodeled to house the engineering library and a new examination hall. A fifth floor was added to provide additional faculty offices and large classrooms. An existing structure in the shop area was remodeled to house the Environmental Engineering Research Center and a section was added to the Fluid Mechanics laboratory to house the Water Resources Center. An additional third floor was added on the western side of Wing B to house laboratories for control systems, communications and microprocessors. The Architecture Building was completely renovated with the addition of an extra floor to provide design studios and faculty offices. Plans are underway to replace Wings B and C with a new engineering building to house state-of-the-art laboratories, lecture halls, and offices.

In 2001 a new building dedicated to the late Dean Raymond Ghosn was completed and inaugurated under the auspices of Mr. Rafic Hariri, Prime Minister of Lebanon represented by Mrs. Bahia Hariri, Member of Parliament and in the presence of AUB President John Waterbury and other dignitaries. In 1986 the Electrical Engineering Department introduced a new program leading to the degree of Bachelor of Engineering - major, Computer and Communications Engineering. In 1992 the Department of Architecture introduced a new four-year undergraduate degree program in Graphic Design. At the graduate level, each of the three engineering departments offered a program leading to the Degree of Master of Engineering. The Master's degree program in Urban Planning and Urban Design was initiated within the Department of Architecture. A new interdepartmental graduate program leading to the Degree of Master of Engineering Management was started in 1990. The feasibility of introducing a Ph.D. program in engineering is underway.

9

Dean Raymond Ghosn Building - 2001

1948 Model (Still in Use)

2001 Model State (state-o f-the-art) Material Testing Machine

The number of full-time faculty members increased from 14 in 1952 to 49 in 2001. Over the same period the student numbers increased from about 100 undergraduate students in 1951 to about 1200 undergraduate and 150 graduate students in 2001. The need for extra faculty as a result of the increase in the number of programs and students over the past ten years resulted in an increase in part-time faculty numbers from 23 to 60 in 2001. Every effort has been and continues to be made towards the enhancement of academic standards, whether through student admission, faculty development, recruitment of highly qualified faculty, emphasis on research activity and effectiveness of teaching, updating and improvement of curricula, strict evaluations and monitoring of student performance, and upgrading of laboratory and physical facilities. Faculty were encouraged to avail themselves of opportunities to attend conferences, workshops, seminars, and short courses abroad, and to spend the summer months, on a rotating basis, on working visits to universities in the USA and Europe.

Social, cultural, and athletic activities were encouraged because of their importance for personality development of engineering and architecture students, as is the fostering of ethical conduct, positive attitudes, and traits of hard work and self discipline. The primary goal of the Faculty of Engineering and Architecture has always been to graduate engineers and architects who are renowned as much for their spirit, personality, conduct and attitudes as for their professional skills and knowledge. Alumni of the Faculty have contributed significantly to the efforts and plans of the Faculty. A Council of Alumni Support of the Faculty of Engineering and Architecture was established and has met regularly to plan and supervise Alumni activities. Relations and ties with Alumni have always been a source of pride to the Faculty and should be continually nurtured, strengthened and enhanced to the benefit of both Faculty and Alumni. The Faculty of Engineering and Architecture continuously strives to maintain its national and regional leadership in education and research, and its commitment to the service of the country and

10

the region and in keeping pace with international developments. In October 2001 all engineering curricula were restructured to conform to the US Accreditation Board for Engineering and Technology (ABET). Interdepartmental research groups were formed that focus on energy, information technology and mechatronics. Other research groups are being formed. Research funds were received from local and international agencies and companies. In 2001 LibanCell donated $200,000 to the FEA, $100,000 for scholarships and $100,000 for equipment and research support on wireless communications. AUB trustee Ray Irani donated $100,000 to support research projects in the Energy Research Group. Dar Al-Handassah (Shair and Partners) donated $250,000 towards a $2,000,000 endowment fund to support research in engineering. ASHA donated $600,000 towards new laboratory equipment. Also in 2001, other grants, contracts, gifts and donations were received from the AUB research Board, the Lebanese National Council for Scientific Research, other local and international government agencies, businesses, and individuals, including a $50,000 annual donation for scholarships from an anonymous alumnus, an annual $10,000 Fawzi W. Azar award to a student in Architecture, an annual $5,000 donation from the Charles Kettanneh Foundation, and an annual $10,000 donation from the Lakeside Foundation. The AUB Engineering & Architecture Alumni Chapter has a Scholarship Fund earmarked for the School of Engineering & Architecture. In 2001 the Chapter donated $42,000 to the Scholarship Fund. The Chapter also helped provide 70 jobs for our graduating students through their annual Job Fair. A number of activities have been planned for this academic year to celebrate the 50th Anniversary of FEA. The Raymond S. Ghosn Building that was dedicated on October 31, 2001, was part of the 50th anniversary celebration. On November 3rd, 2001, the Engineering and Architecture Alumni Chapter organized a gala dinner at the Phoenicia Hotel to honor a number of distinguished alumni and friends. This Spring 2002 celebration events include students’ sports activities, art and science exhibitions, lecture series in all departments, a musical concert, job fair and gala by the AUB Engineering Alumni Association, class reunions, a conference, and final year projects exhibitions by students.

11

The FEA Distinguished Alumnus Awards

FEA Distinguished Alumnus Award To Nabil F. Azar in recognition of his innovative and

extensive architectural designs that integrate an interpretation of regional tradition, contemporary

functionalism, and future vision.

Nabil F. Azar BE 1970

Chief Executive Officer

Builders Design Consultants Beirut, Lebanon

Nabil F. Azar is a graduate of the American University of Beirut with a Bachelor Degree in Architecture (1970) and a holder of the Penrose Award. He then traveled to Japan where he studied at Chiba University (1971).

Nabil Azar established his firm "Builders Design Consultants" in 1973. The firm offers consultancy services in architecture, interior architecture, restoration, landscaping, street beautification, and supervision of projects in both the private and the public sectors.

B.D.C has completed over 200 projects with diverse building types and assignments: religious, educational, cultural, touristic and recreational centers, commercial centers, hospitals, housing complexes, private residences and apartment buildings. His projects are located in Lebanon, Kingdom of Saudi Arabia, the Gulf countries, Cyprus and Zanzibar.

The technical staff of B.D.C. includes qualified architects, planners, structural and site engineers operating as a team with a vision of an architecture concerned with people, context, climate and culture implementing the most recent quality requirements which conform to ISO 9001 standards.

Plenary talk at FEASC 2001: “Architecture: Theory and Practice”

12

FEA Distinguished Alumnus Award To Milhim Ghulmiyyah in recognition of his pioneering professional achievements and outstanding leadership in the area of thermal design and construction of power plant facilities and pipeline systems.

Milhim Joseph Ghulmiyyah BE 1982

General Manager AGAP Arabia Ltd. Saudi Arabia

Milhim Joseph Ghulmiyyah is a graduate of the FEA Mechanical Department class of 1982 and currently is the General Manager of AGAP Arabia Ltd., which is a Mechanical Industrial construction company in the field of Oil, Gas and Petrochemical Plants in Saudi Arabia.

Upon graduation, Mr. Ghulmiyyah joined C.A.T. Co (Contracting and Trading Company) in Saudi Arabia as a site Engineer where he supervised many projects in various locations in the Kingdom and in UAE from 1982 till 1994. During that period he designed and supervised the construction of thousands of pipelines for various functions (drinking water pipelines, water injection lines, crude oil pipe lines, gas pipelines and multi-product pipelines) and designed and supervised as well the construction of pumping stations, oil plants, gas plants, water environmental plant and a gas turbine driven power plant.

In June 1994, Mr. Ghulmiyyah joined PETROFAC Int. as a Construction Manager for a gas plant project in Syria, where he later headed their Damascus office as a Resident Manager. The constructed gas plant project at Dier Al Zour in Syria had the capacity to process and compress 150 million scfd of natural gas.

In 1997, Mr. Ghulmiyyah joined the Nesma Group of Companies as the General Manager of AGAP Arabia ltd. (The industrial construction company in the Nesma Group). He has restructured the company and rebuilt its resources from 125 employees to about 800 by now, while multiplying the company assets.

Mr. Ghulmiyyah professional contributions are demonstrated by the so many prestigious projects he has been responsible for.

Plenary talk at FEASC 2001: “Engineers in Construction and Construction Management”

13

FEA Distinguished Alumnus Award To Adib Kanafani in recognition of his significant

contributions to air transportation, intelligent transportation systems, and education.

Adib Kanafani BE 1964 MS 1967

PhD 1969

Edward G. and John R. Cahill Professor of Civil Engineering Chairman

Department of Civil and Environmental Engineering University of California at Berkeley, USA

Adib Kanafani graduated from AUB in 1964 with a B.E. degree in Civil Engineering (with distinction). He obtained his M.S. and Ph.D. degrees from the University of California at Berkeley in 1967 and 1969. Since joining the faculty at Berkeley in 1970 he has taught and conducted research on transportation systems, transportation engineering, airport planning and design, and air transportation economics. He has served on a number of national and international advisory panels to Government and industry. He is currently Chairman of the Department of Civil and Environmental Engineering at Berkeley.

Kanafani’s made contributions to air transportation including demand analysis, airport capacity analysis methods, and airline network analysis. His research on airline hubbing and on the relation between aircraft technology and airline network structure laid the ground for much of the work aimed at understanding the implications of airline deregulation in the late 1970’s. He was a member of the research team that developed airport capacity analysis methods that are in widespread application in airport planning and design. In 1997 he was founding Co-D irector of the National Center of Excellence in Aviation Operations Research, NEXTOR, a University/Industry partnership funded by the Federal Aviation Administration and headquartered at Berkeley. Kanafani served as Director of Berkeley’s Institute of Transportation Studies from 1983 to 1998, during that period he played an important role in establishing the Intelligent Transportation research effort in the U.S. He was a founding member of Mobility 2000, the precursor organization to today’s national program in ITS, and he founded Berkeley’s PATH program which is one of the premier research programs in ITS.

Professor Kanafani has authored over 170 publications on transportation, including a book on Transportation Demand Analysis, and one on National Transportation Planning. He is a recipient of numerous awards including the ASCE’s Walter Huber Research Prize in 1982, and Horonjeff Award in 1988, and the James Laurie Prize in 2000. He was elected to the National Academy of Engineering in 2002.

Plenary talk at FEASC 2001: “Serving a Connected Society – The Emerging Paradigm for Engineering and Education”

14

FEA Distinguished Alumnus Award To John Makhoul in recognition of his fundamental contributions to speech analysis and human-computer interaction by voice.

John Makhoul BE 1964 MSc 1965 PhD 1970 Chief Scientist BBN Technologies, A Verizon Company Cambridge, MA, USA

John Makhoul received the B.E. degree from the American University of Beirut in 1964, the M.Sc. degree from the Ohio State University in 1965, and the Ph.D. degree from the Massachusetts Institute of Technology in 1970, all in electrical engineering. At AUB, he received the Penrose Award in 1964.

Since 1970 he has been with BBN Technologies, Cambridge, MA, where he is a Chief Scientist working on various aspects of speech and language processing by computer. In addition to his duties at BBN, he is Adjunct Professor at Northeastern University, Boston University, and the University of Massachusetts.

Dr. Makhoul is a Fellow of the Institute of Electrical and Electronics Engineers (IEEE) and a Fellow of the Acoustical Society of America. His IEEE Fellow citation reads: "For contributions to the theory of linear prediction and its applications to spectral estimation, speech analysis, and data compression."

He is the recipient of a number of professional awards, including the IEEE Third Millennium Medal.

He currently serves on the Board of Directors of the AUB Alumni Association of North America.

Plenary talk at FEASC 2001: “Can Computers Really Understand the Human Voice?”

15

FEA Distinguished Alumnus Award To Hussein Rifaï in recognition of his leadership and

innovative contributions in implementing and advancing telecommunications and information systems, and his support

of education.

Hussein Rifaï BE 1981

Chairman and General Manager

LibanCell Beirut, Lebanon

Hussein Rifaï is currently holding the position of Chairman – General Manager of LibanCell.

He received his B.E. in Electrical Engineering (with Distinction) from the American University of Beirut in 1981.

He started his career in Saudi Arabia where he specialized in low currents and telecommunications systems. He later on moved to Paris as Partner in telecommunications consultancy company I.T. CAL and managed telecom and information systems projects, in France and all over Europe, for large companies, mainly financial institutions. He joined LibanCell, a cellular operator in Lebanon, at its inception late 1994. He succeeded in building the operations, in an extremely challenging environment, from scratch up to a mature and innovative organization serving more than 385,000 subscribers.

Plenary talk at FEASC 2001: “Building Partnerships between the Private and Educational Sectors”

16

The FEA Distinguished Scholar Award

The FEA Distinguished Scholar Award To Thomas Kailath for his many and varied innovative contributions to the advancement of science and engineering, and for the integrity, goodwill, and generous spirit that he brings to everything that he does.

Professor Thomas Kailath Fellow IEEE Hitachi America Professor of Engineering Department of Electrical Engineering Stanford University, Stanford, California

Professor Kailath’s research has spanned a large number of disciplines, emphasizing information theory and communications in the sixties, linear systems, estimation and control in the seventies, and VLSI design and sensor array signal processing in the eighties. Concurrently, he contributed to several fields of mathematics, especially stochastic processes, operator theory and linear algebra. While he maintains all theses interest to varying degrees, his current research emphasizes their applications to problems of semiconductor manufacturing and high-speed digital communications.

Professor Kailath has mentored over a hundred doctoral and postdoctoral students. He is the author of Linear Systems, Prentice Hall, 1980, Lectures on Wiener and Kalman Filtering, Springer -Verlag, 1981; and co-author of Digital Neural Computation (with S. Siu and V. Rochowdhury), Prentice Hall, 1995; Indefinite Quadratic Estimation and Control (with A.H. Sayed and B. Hassibi), SIAM, 1999; Linear Estimation (with A.H. Sayed and B. Hassibi), Prentice Hall, 2000; He has also authored or co-authored over 300 research papers and received outstanding paper prizes from the IEEE Information Theory Society, the IEEE Signal Processing Society, the European Signal Processing Society, and the IEEE Transactions on Semiconductor Manufacturing.

Professor Kailath has held Guggenheim and Churchill fellowships, among others. He served as President of the IEEE Information Theory Society in 1975, and received its Shannon Award in 2000. Among other awards are Technical Achievement and Society Awards of the IEEE Signal Processing Society, the John R. Ragazzini Award of the American Control Council, the Stevin Medal of Deft University of Technology, a Golden Jubilee Medal of the IEEE Circuits and Systems Society, a Golden Jubilee Paper Award of the IEEE Information Theory Society, the IEEE Education Medal, the IEEE Donald G. Fink Price Award, and an IEEE Millennium Medal.

He is a Fellow of the IEEE and of the Institute of Mathematical Statistics, and a member of the National Academy of Engineering, the National Academy of Sciences, the American Academy of Arts and Sciences, the Third World Academy of Sciences, and the Indian National academy of Engineering.

Professor Kailath has co-founded several high-technology companies in 1980, three of which are now public: Integrated systems, Inc., which merged with WindRiver Systems (www.windriver.com)

17

in 1999, Numerical Technologies, Inc. (www.numeritech.com) in 1995, and in 1998 Excess Bandwidth Corporation, which was acquired by Virata Corporation (www.virata.com) in 2000.

Plenary talk at FEASC 2001: “Breaking the 0.1 Barrier in Optical Microlithography via Signal Processing”

18

Breaking the 0.1 Barrier in Optical Microlithography via Signal Processing

Thomas Kailath Hitachi America Professor of Engineering

Department of Electrical Engineering Stanford University, Stanford, California

The current predicted resolution limit for optical lithography is 0.13 µm, using 193 nm laser sources and resists that are still under development. Beyond that, to quote Gordon Moore, there seem to be only three equally unappealing options: X-rays, e-beam, and deep UV (with mirrors instead of lenses). It appears that signal processing ideas can allow us to go beyond the so-called "0.1 barrier".

The first steps in this direction were taken by Marc Levenson and others at IBM who showed that one could get well-separated images for two lines spaced closer than the conventional Rayleigh limits by adding a 180° phase shift to one of the lines. This idea, extended to more complex patterns, is being explored by many semiconductor manufacturing organizations as a way of extending the resolution of existing optical exposure systems. However, for all but very simple patterns, mask design is currently done in a laborious and empirical way, with in fact no guarantee that a suitable phase-assignment even exists.

In this talk, we shall describe a systematic procedure based on the (approximate) solution of a nonlinear inverse problem - going from a desired pattern on a wafer to the specification of a mask that will yield this result by appropriately compensating for the effects of the optical projection system.

Some experimental results will be displayed showing satisfactory printing of lines at a spacing 3 times as close as predicted by the nominal resolution, e.g., 0.11-0.13 µm spacings using a system with 0.35 µm nominal resolution. Comparable reductions have been obtained with systems of higher nominal resolution, providing hope that refractive-optical lithography could be used well into the next decade.

This talk is based on work done with Dr. Y.C. Pati and Dr. Yao-Ting Wang at Stanford University and Numerical Technologies, Inc.

19

Technical Program – Schedule

Thursday, June 6, 2002 8:30 – 9:00 Opening Ceremony (Room: ELH)

§ Welcome by Peter Heath, Provost, AUB § Welcome by Ibrahim N. Hajj, Dean, FEA, AUB § Welcome by the President of FEA Student Representative Committee

9:00 – 10:30 Plenary Session I (Room: ELH) § “Can Computers Really Understand the Human Voice?”

John Makhoul, Chief Scientist BBN Technologies, Cambridge, MA, USA

§ “Building Partnerships between the Private and Educational Sectors” Hussein Rifaï, Chairman and General Manager LibanCell, Beirut, Lebanon

Break 11:00 – 12:20 Technical Sessions I (Room: ELH) & II (Board Room) Lunch Break 14:00 – 14:45 Plenary Session II (Room: ELH) § “Architecture: Theory and Practice”

Nabil Azar, Chief Executive Officer Builders Design Consultants, Beirut, Lebanon

15:00 – 16:20 Technical Session III (Room: ELH) 16:30 – 19:00 Poster & Demonstration Session* 19:00 – 21:00 RECEPTION, East Terrace, Bechtel Bldg. * Displays will be available all day June 6 and 7 for public viewing.

Friday, June 7, 2002 9:00 – 10:30 Plenary Session III (Room: ELH)

§ “Serving a Connected Society – The Emerging Paradigm for Engineering and Education” Adib Kanafani, Chairman Department of Civil and Environmental Engineering University of California at Berkeley, USA

§ “Engineers in Construction and Construction Management” Milhim Ghulmiyyah, General Manager AGAP Arabia Ltd., Saudi Arabia

Break 11:00 – 12:20 Technical Sessions IV (Room: ELH) & V (Board Room) Lunch Break 14:00 – 14:45 Plenary Session IV (Room: ELH) § “Breaking the 0.1 Barrier in Optical Microlithography via Signal Processing”

Thomas Kailath , Hitachi America Professor of Engineering Stanford University, USA

15:00 – 16:00 Technical Session VI (Room: ELH) 16:15 – 16:30 Closing Ceremony (Room: ELH)

20

Technical Session I – Computer Vision, Signal Processing

Thursday, June 6, 11:00 – 12:20 Room: ELH 1 11:00 – 11:20 IRIS Recognition Carine Daouk, Lina El-Esber, and Fouad Kammoun Page 32 2 11:20 – 11:40 Beat the traffic Houda Karaki, Jihad Ibrahim, and Hassane Kamal Page 36 3 11:40 – 12:00 Perspective image formation using omnidirectional vision Adel T. Zouheiry and Mounir M. C. Doumani Page 40 4 12:00 – 12:20 Speech recognition for voice based control Youssef Naous, Ghinwa Choueiter, and Nesrob Ohannnessian Page 44

Technical Session II – Computer and Communication

Thursday, June 6, 11:00 – 12:20 Room: Board Room 1 11:00 – 11:20 Mobile.Music Joe Abou-Rjeily, Jade Bissat, and Mario Harik Page 48 2 11:20 – 11:40 Mobile jammer Paula Atallah, Najwa Hamzeh, and Elias Nahra Page 50 3 11:40 – 12:00 SMS shell Rima Daoud, Jean Ghanem, and Nayla Hamadeh Page 52 4 12:00 – 12:20 MSIS: Mobile student information system Hicham Fadel, Habib Haddad, and Hady Moussa Page 55

21

Technical Session III – Electrical, Computer, Communication

Thursday, June 6, 15:00 – 16:20 Room: ELH 1 15:00 – 15:20 Above and beyond the wires – W.A.M Ayman Chalhoub, Mark Ghibril, and Wissam Haddad Page 57

2 15:20 – 15:40 GIS based computerized traveling guide Chucri Abi Haidar, Mossine Koulakezian, and Fadi Abou Ghantous Page 59 3 15:40 – 16:00 Common sense controller (CSC) using LabVIEW Sima Azar and Sandra Dandach Page 61 4 16:00 – 16:20 Linearization of a D/A Converter Michel M. Azarian, Alain M. Hani, and Roland N. Saad Page 65

Poster & Demonstration Session

Thursday, June 6, 16:30 – 19:00 Bechtel and Raymond Ghosn Engineering Buildings The students of the Department of Electrical and Computer Engineering and the Department of Mechanical Engineering are displaying technical posters describing their Final Year Projects (FYP). The displays will be available all day June 6 and 7 for public viewing.

Refer to pages 25 and 29 for a complete list of the Final Year Projects for the academic year 2001-2002.

Technical Session IV – Mechatronics, Bioengineering

Friday, June 7, 11:00 – 12:20 Room: ELH 1 11:00 – 11:20 The design of a prosthetic glove for a child with cerebral palsy Tarek O. El-Chidiac and Jawad T. Sabbah Page 67 2 11:20 – 11:40 Development of a computer-based measurement station Marianne Sarkis, Michel Sarkis, and Joseph Malkoun Page 71

22

3 11:40 – 12:00 CNC retrofit for milling machines Imad Doumit, Georges El-Mir, and Makram Nehme Page 74 4 12:00 – 12:20 Hoover Saucer Selim Hobeika, Nael Itayem, and Marc Tabchy Page 76

Technical Session V – Mechanical, Electrical

Friday, June 7, 11:00 – 12:20 Room: Board Room 1 11:00 – 11:20 A comparative assessment within a multi-grid environment of the performance of segregated pressure -based algorithms for fluid flow at all speeds Daniel Asmar Page 79 2 11:20 – 11:40 Chocooler – A mini thermoelectric cooler for chocolate bars Fadi Hachache, Ibrahim Geha, and Abdul Karim Younes Page 84 3 11:40 – 12:00 Human electric locomotive project – HELP Raya Abdel Baki, Antoine Abboud, and Chadi Chaker Page 87 4 12:00 – 12:20 Capacitor placement using genetic algorithm and distributed computing Francois Layoun, Maher Salameh, and Raji Sayegh Page 89

Technical Session VI – Civil, Transportation, Education

Friday, June 7, 15:00 – 16:00 Room: ELH 1 15:00 – 15:20 Geophysics and engineering hand in hand Nadim Baalbaki and Jouanna Doummar Page 91 2 15:20 – 15:40 Potential for Rail Freight in Lebanon Rawad Hani Page 93 3 15:40 – 16:00 Civil engineering and novel technology: Where do they meet? Loai Naamani Page 95 4 Cultered Engineers: A Necessity or a Privilege Loai Naamani Page 97

23

Graphic Art Exhibition of Fine Art Prints

Architecture Building Exhibition Hall June 6 –12, 2002 Official opening on June 6 at 18:00 Coordinator: Mohammed Rawas

The exhibition includes a selection of etchings and silkscreen prints (about 50 works) produced by the students of the Graphic Design and Architecture programs at FEA during the last seven years. The Etching workshop was established in 1994 to introduce this printmaking technique as a core course in the Graphic Design Program. The Silkscreen workshop was added later in 1998.

Etching and Silkscreen printing are now offered as elective courses in the GD major. The exhibition works represent a variety of individual artistic statements in their themes, styles and technical printing techniques.

Human Powered Vehicles

Coordinator: Ahmad Smaili

The Human Powered Vehicle (HPV) competition is organized to promote the spirit of competition and teamwork among students and to enhance their entrepreneurial skills while involved in a physically and intellectually stimulating experience of designing and building a vehicle that have many societal benefits such as cleaner and less noisier environment, less crowdedness on the streets, healthier exercising and recreational habits, and other economical benefits.

The HPV’s designed this year by Mechanical Engineering students are displayed on the road between Bechtel and Raymond Ghosn Engineering Buildings.

24

The 50 th – Structures Contest 2002

Coordinator: Mounir Mabsout Formal opening and Evaluation by Jury: June 6, 2002, 16:00 Bechtel/Engineering Bldg., 4 th floor, FEA Library Lobby

The Contest consists of building a model of a Structure and/or Civil Engineering system with a theme that symbolizes the FEA 50th Anniversary. There are no restrictions on the material of which the model is made. Projects should be enhanced with special effects or features (Mechanical, Electrical, Visual, Computer, …). The models will be judged for their soundness and stability, reactivity, complexity, originality, and the level of entertainment they provide. A poster describing the project is also required. Participating teams, in groups of three to six contestants, are made of Civil Engineering undergraduate students, with associates from other fields. Prizes and certificates will be awarded for the winner teams. Eight teams are participating in the Contest:

1 DISCORD Ayman Abou Said, Kristina Araman, Mohamad Faour, Faten Ibrahim, Nadim Rouhana, and Farah Taha 2 ENLIGHTENMENT Pascal Abou Dagher, Nadine Badr, Raed Ghousainy, Mariane Kurban, Ziad Nassar, and Michel Waked Sponsors: CAT-Contracting and Trading Co., DEBBAS Lighting, Engineering Alumni Chapter-AUB, GHALI Copy Center 3 IL CONSTRUCTURA Ali Darwish, Karim-Philippe Eid -Sabbagh, Wissam Khalil, Cinderella Nouweihed, Chadi Sabra, and Ramzi Tafesh 4 INNOVATE Maher El-Sayed, Raed Jarrah, Karim Musallem, and Saleh Sabbidin Sponsor: EL-SAYED Office for Contracting and Consulting 5 MAC CIVIL Linda Al-Atik, Marie -Helene Doumit, Mark El-Khoury, Bechara Harfouche, and Serge Rizk 6 MACSI Sami Bejjani, Mario Chamoun, Anis-Patrick Jabr, Imad Moukarzel, and Camil Tahan Sponsor: MATTA-Entreprise Alfred & Jacques Matta 7 St. KANT Sako Holtian, Raffi Khoshafian, and Charbel Rizk Sponsor: ABS-Antranik Baldjian & Sons 8 THE CIVILIANS Nadim Azar, Chahe Dabanjian, Toufic Fakih, Rania Fakih, and Najib Germanos

25

Final Year Projects (FYP)

As part of their final year, students are required to carry out a project and to submit a technical report. This project is a substantial piece of work that requires creative activity and original thinking. The objectives of a final year project are:

• To allow students to demonstrate a wide range of the skills learned at FEA during their course of study by asking them to deliver a product that has passed through the design, analysis, testing, and evaluation stages;

• To encourage multidisciplinary research through the integration of material learned in a number of course units;

• To allow students to develop problem solving, analysis, synthesis and evaluation skills;

• To encourage teamwork;

• To improve students communication skills by asking them to produce both a professional report and a professional poster and to give an oral presentation on their work.

FYP’s – Department of Electrical & Computer Engineering

Coordinator: Fuad Mrad

1 Iris recognition Fouad Kammoun, Lina E l-Esber, and Carine Daouk Advisor: A. Al-Alaoui 2 Speech recognition using FPGA Youssef Naous, Mesrob Ohannessian, and Ghinwa Choueiter Advisor: A. Al-Alaoui 3 Image coding and compression Shadi Nasser, Nicolas Nasrallah, and Sharfi Nakouzi Advisor: H. Artail 4 A distributed system of network devices Dunia El Hassan, Randa Zakhour, Rola El Tal, and Hania Sabbidin Advisor: H. Artail 5 Controlling a robot arm Elie Maalouf and Samer Bizri Advisor: H. Artail 6 Objects dimensions determination Line Dghaili Abdallah, Rayane Chahine, and Elie Sfeir Advisor: H. Artail

26

7 G.I.S. based computerised travelling system Fadi Abi Ghantous, Mossine Koulakezian, and Churri Abi Haidar Advisor: F. Chaaban 8 Design and performance testing of a renewable electric energy system Wissam Shaar, Tamer Zaidan, and Haitham Awar Advisor: R. Chedid 9 Electric wheel chair Chadi Chaker, Raya Abdel Baki, and Antoine Abboud Advisor: R. Chedid 10 Performance evaluation of image processing algorithms on reconfigurable

computing systems Mohammad Mechli, Ma'en Shaka'a, and Samer Ockaily Advisor: H. Diab 11 Mobile jammer Paula Atallah, Elias Nahra, and Najwa Hamzeh Advisor: K. Kabalan

Sponsored by Modern Electronic Corporation (MEICO) 12 Turbo code simulation Jad Berbary and Bruni El Asmar Advisor: K. Kabalan 13 Capacitor placement using generic algorithms and distributed computing Francois Layoun, Maher Salameh, and Raji Sayegh Advisor: S. Karaki 14 Development of a computer based measurement station Marianne Sarkis, Joseph Malkoun, and Michel Sarkis Advisor: S. Karaki

Sponsored by Modern Electronic Corporation (MEICO) 15 Beat the traffic Kamal Hassane, Hoda Karaki, and Jihad Ibrahim Advisor: K Kayssi

Sponsored by LibanCell 16 Linearisation of a channel Roland Saad, Alain Hani, and Michel Azarian Advisor: K Kayssi

Sponsored by National Instruments USA 17 High accuracy remote data acquisition module Mona El Hassan, Ghada Derbas, and Rula Antoun Advisor: F. Mrad

Sponsored by Y.L. Engineering

27

18 Common sense controller (CSC) Sandra Dandach and Sima Azar Advisor: F. Mrad

Sponsored by National Instruments USA 19 Design of a computer controlled robot for data acquisition using path

tracing Fadi Wakil, Ziad El Bizri, and Jabbour Abdallah Advisor: F. Mrad and A. Smaili 20 Low cost UV alert Nadim Abi Ali, Ahmad Sidani, and Ghayass Tohme Advisor: F. Mrad

Sponsored by SES – Microchip 21 Design and simulation of multiple traffic light controllers for multiple

intersections Johny Jabbour, Bassil Ghazzaoui, and Mohamad Shehab Advisor: J. Saade 22 Bluetooth business card Raya Maxim, Jean Pierre Saad, and Fouad Tarazi Advisor: A. Al-Alaoui 23 Design of an optimum path and general robot for navigation among

moving obstacles Wael Mabsout, Fadi Hawwa, and Nadim Klat Advisor: J. Saade 24 Digital phase meter Rayan Salha, Samer Farhat, and Rami Adada Advisor: N. Sabah 25 Active vision Jihane Assaf and Fadi khoury Advisor: S. M. Abdallah 26 Linear motor, design and application Ayman Ayoub, Nabil Al-Shurafa, and Firas Abani Advisor: F. Chaaban 27 Adaptive fuzzy logic based differential pulse code modulation Rony El Saouda, Louay Lezeik, and Walid Nasr Advisor: J. Saade 28 Digital wireless communication Bechara Moufarrej, Weam Abu Zaki, and Sinno Sihaila Advisor: K. Kabalan

28

29 Advanced control for electro-magnetic suspension Shrayteh Z iad and Tarek Mikaty Advisor: F. Chaaban 30 Computer numerical control (CNC) for a 3 axis lathe machine George El Mir, Imad Doumit, and Makram Nehme Advisor: A. El Hajj

Sponsored by Phoenix Machinery 31 OV-CAR: Omnidirectional vision autonomous robot Adel Zouheiry, Mounir Doumani, Antranik Serge Saatdjian, and

Seboah Pamoukian Advisors: S. M. Abdallah and A. Smaili 32 Mobile.com Hicham Fadel, Hady Moussa, and Habib Haddad Advisor: A. Kayssi

Sponsored by LibanCell 33 Video conferencing over GSM/GPRS Rima Daoud, Jean Ghanem, and Nayla Hamadeh Advisor: A. Kayssi

Sponsored by LibanCell 34 Internet connectivity application Alain Rassi, Gilbert Tawil, Itani Ayman, and Nayla Itani Advisor: K. Kabalan

Sponsored by Terranet 35 Wireless LAN Haddad Wissam, Chalhoub Ayman, and Ghibril Mark Advisor: A. El Hajj

Sponsored by Cisco 36 Internet radio Joe Abou Rjeily, Jade Bissat, and Mario Harik Advisor: A. El Hajj

Sponsored by LibanCell 37 Web portal Joe Awwad, Camille Beyrouthy, and Ghassan Abou Samra Advisor: R. Chedid 38 Computer controlled movable pins for multi-purposes Maher Hatab and Karim Tourbah Advisor: N. Sabah 39 Parallel processing in computational molecular biophysics Nadim Tawileh, Imad Agha, and Leila El Khazen Advisor: H. Diab

29

40 Adaptive solar panel Mohammad Badran, Marc Bitar , Fadi K. Sbaiti, and Jad Bajjani Advisor: A. Smaili and S.M. Abdallah 41 Universal asynchronous bluetooth transceiver Jad Abou Rjeili, Zalek Abdul Hadi, and Maya Haddad Advisor: A. El Hajj 42 Development of a post processing tool for the ap plication "terms

investigation 3.1" Ayman Jomaa, Merib Adnan, and Naji Issa Advisor: H. Diab

Sponsored by LibanCell 43 Distributed shared memory Elias Jureidini Advisor: S. Karaki

FYP’s – Department of Mechanical Engineering

Coordinator: Samer M. Abdallah

1 Hoover saucer Marc Tabchy, Nael Itayem , and Selim Hobeika Advisor: S. M. Abdallah 2 BinBot – The bin loader robot Jad Kahlil, Samer Azzam, and Amer Jaber Advisor: A. Smaili 3 The design of a prosthetic glove for a child with cerebral palsy Tarek El-Chidiac and Jawad Sabbah Advisor: K. Khalaf 4 Load Calc - Creating a software for radiant time series methods using

Visual Basic Nadine Mazzawi, Mohamad Chehade, and Ziad El-Khal Advisor: F. Moukalled 5 OV-CAR: Omnidirectional vision autonomous robot Sebouk Pamoukian, Antranik Serge Saatdjian, Adel Zouheiry, and Mounir

Doumani Advisors: S. M. Abdallah and A. Smaili

30

6 Airbags Nabil Abdel Khalek, Marwan Abou Haidar, Nicolas El-Haddad, and Malek

Kazem Advisor: K. Khalaf 7 Automatic piano tuning Joseph Tilbian and Garo Matosian Advisor: A. Smaili 8 Powder metallurgy Armen Arevian and Hagop Jatalian Advisor: H. Hamade 9 Energy conservation report for the mechanical shops at Wing B building Marwan Khattab, Mohammad Shehab, and Samer Menhem Advisor: N. Ghaddar 10 Adaptive solar panel Mohammad Badran, Marc Bitar, Fadi Sbeity, and Jad Bejjani Advisors: A. Smaili and S. M. Abdallah 11 Active stereo vision Jihane Assaf and Fadi Khoury Advisor: S. M. Abdallah 12 Engine tuning and dynamometer coupling Mohamad Baraa Al Halabieh and Omar El-Chami Advisor: A. Shihadeh 13 Multi-newspaper vending machine Mireille Akilian, Naji Atallah, Salim El-Ferkh, and John Feghali Advisor: A. Smaili 14 Active vision head Mohannad Al-Hakim, Hania Zaatari, and Muayyad Shaaban Advisors: S. M. Abdallah and A. Smaili 15 The PostBot Fadi Abou-Ibrahim, Mahmoud Itani, and Karim Yehia Advisors: A. Smaili and S.M. Abdallah 16 Human-electric hybrid vehicle Joe Youssef Malek, Gilbert Semaan, and Elie Fadel Advisor: A. Shihadeh 17 Focal reactor for a solar parabolic concentrator Armen Abou Hamad and Omar El Jaroudi Advisor: R. Nuwayhid

31

18 Hybrid vehicle energetics: Portable data acquisition unit Nayla Jaber, Jinan Abou Rabia, and Mona Sabra Advisors: A. Shihadeh and S. M. Abdallah 19 Hydrodynamics and boundary layers computer package Fouad Al Sous and Rabih Jawad Advisor: F. Moukalled 20 Design of a computer controlled robot for data acquisition using path

tracing Fady Wakil, Abdallah Jabbour, and Ziad El Bizri Advisor: F. Mrad and A. Smaili 21 Techno-economic study of solar water heating Rami Joueidi, Bassel Kikano, and Charles Tarazi Advisor: N. Ghaddar 22 Implementation of some extensions of the tri-diagonal and penta-diagonal

matrix algorithms Yaqoub El Khamra Advisor: F. Moukalled 23 Thermoelectric fan Wissam El-Hajj Obeid, Rami Maalouf, and Jad Mouawad Advisor: R. Hamade and R. Nuwayhid

32

Iris Recognition

Carine Daouk Dept. Electrical and Computer American University of Beirut

Beirut, Lebanon [email protected]

Lina El-Esber Dept. Electrical and Computer American University of Beirut


Fouad Kammoun Dept. Electrical and Computer American University of Beirut


Abstract In this paper, we briefly describe the design, implementation, and performance of our final year project ‘Iris Recognition’.

Keywords Bilinear Transformation, Biometrics, Canny Operator, Haar Wavelet, Hough Transform, Iris Recognition. INTRODUCTION

The purpose of ‘Iris Recognition’, a biometrical based technology for personal identification and verification, is to recognize a person from his/her iris prints. In fact, iris patterns are characterized by high level of stability and distinctiveness. Each individual has a unique iris (see Figure 1); the difference even exists between identical twins and between the left and right eye of the same person. [7]

We implemented ‘Iris Recognition’ using Matlab for its ease in image manipulation and wavelet applications. The first step of our project consists of images acquisition. Then, the pictures’ size and type are manipulated in order to be able subsequently to process them. Once the preprocessing step is achieved, it is necessary to localize the iris and unwrap it. At this stage, we can extract the texture of the iris using Haar Wavelets. Finally, we compare the coded image with the already coded iris in order to find a match or detect an imposter (Figure 7).

Figure 1: Distinctiveness of human iris

IMPLEMENTATION

Image acquisition Image acquisition is considered the most critical step in our project since all subsequent stages depend high ly on the image quality. In order to accomplish this, we used a CCD camera. We set the resolution to 640x480, the type of the image to jpeg, and the mode to white and black for greater details. Furthermore, we took the eye pictures

while trying to maintain appropriate settings such as lighting and distance to camera.

Image manipulation

In the preprocessing stage, we transformed the images from RGB to gray level and from eight-bit to double precision thus facilitating the manipulation of the images in subsequent steps.

Iris localization Before performing iris pattern matching, we should locate the boundaries of the iris. In other words, we are supposed to detect the part of the image that extends from inside the limbus (the border between the sclera and the iris) to the outside of the pupil [7]. To achieve this, we apply a canny operator to the image and then a Circular Hough Transform. We slightly modified the Circular Hough Transform so that we can fit it to our application. In order to obtain an acceptable running speed, we dow n sampled the image and then detected the outer boundary. The center and radius of the iris in the original image are determined by rescaling the obtained results. Whereas for the second boundary, we determine it from the original image because the pupil center is shifted by up to 15% from the center of the iris, and the pupil radius can be between 0.1 and 0.8 the size of the radius of the iris [2]. This means that we can restrict a lot the region of our search and we can obtain an accurate result in small time. That is why we don’t need to reduce the image size; we might be loosing some accuracy without gaining relatively much speed.

Figure 2: Localized Iris

33

402

Mapping

After determining the limits of the iris in the previous phase, the iris should be isolated and stored in a separate image (Figure 4). The factors that we should watch out for are the possibility of the pupil dilating and appearing of different size in different images. For this purpose, we begin by changing our coordinate system by unwrapping the lower part of the iris (lower 180 degrees) and mapping all the points within the boundary of the iris into their polar equivalent (Figure 4). The size of the mapped image is fixed (100x402 pixels) which makes that we are taking an equal amount of points at every angle. Therefore, if the pupil dilates the same points will be picked up and mapped again which makes our mapping process stretch invariant. When unwrapping the image, we make use of the bilinear transformation to obtain the intensities of the points in the new image. The intensities at each pixel in the new image are the result of the interpolation of the grayscales in the old image. [4]

Figure 3.Original image

Figure 4. Iris isolated image

Feature extraction

“One of the most interesting aspects of the world is that it can be considered to be made up of patterns. A pattern is essentially an arrangement. It is characterized by the order of the elements of which it is made, rather than by the intrinsic nature of these elements”(Nobert Wiener)[4]. This definition summarizes our purpose in this part. In fact, this step is responsible of extracting the patterns of the iris taking into account the correlation between adjacent pixels. After performing lots of research and analysis about this topic, we found that using wavelets transform, and more specifically Haar transform, illustrated in Figure 5, has the best performance1 regarding feature extraction.

1 By performance we mean not only accurate feature extraction, but

also time effectiveness.

Figure 5. The Haar wavelet

Figure 6. Conceptual diagram for organizing a feature vector

Haar Wavelets

Most previous implementations have made use of Gabor wavelets to extract the iris patterns [2], [3], [7]. But, since we are very keen on keeping our total computation time as low as possible, we decided that building a neural network especially for this task would be too time consuming and selecting another wavelet would be more appropriate.

We obtain the 5-level wavelet tree showing all detail and approximation coefficients of one mapped image obtained from the mapping part. When comparing the results using the Haar transform with the wavelet tree obtained using other wavelets we found that the Haar wavelet gave slightly better results. Our mapped image is of size 100x402 pixels and can be decomposed using the Haar wavelet into a maximum of five levels. These levels are cD 1

h to cD 5h, cD 1

v to cD5v

and cD1d to cD 5

d. We must now pick up the coefficients that represent the core of the iris pattern. Therefore those that reveal redundant information should be eliminated. In fact, looking closely at Figure 6 it is obvious that the patterns in cD1

h, cD2h, cD3

h and cD4h are almost the same and only

one can be chosen to reduce redundancy. Since cD4h

repeats the same patterns as the previous horizontal detail levels and it is the smallest in size, then we can take it as a representative of all the information the four levels carry. The fifth level does not contain the same textures and should be selected as a whole. In similar fashion, only the fourth and fifth vertical and diagonal coefficients can be taken to express the characteristic patterns in the iris -mapped image. Thus we can represent each image applied to the Haar wavelet as the combination of six matrices:

100

CD1h

CD1d CD1

v

CD2

CD2

CD2

CD3h

CD3d CD3

V

r

?

34

• CD4h and cD 5

h • CD4

v and cD 5v

• CD4d and cD 5

d

All matrices are combined to build one single vector characterizing the iris patterns. This vector is called the feature vector. Since all mapped images have a fixed size of 100x402 then all images will have a fixed feature vector. In our case, this vector has a size of 702 elements. This means that we have managed to successfully reduce the feature vector of John Daugman who uses a vector of 1024 elements [3]. This difference can be explained by the fact that he always maps the whole iris even if some part is occluded by eyelashes, while we map only the lower part of the iris whatever are the circumstances obtaining half his feature vector’s size. Binary Coding Scheme

It is very important to represent the obtained vector in a binary code because it is easier to find the difference between two binary code-words than between two number vectors. In fact, Boolean vectors are always easier to compare and to manipulate. In order to code the feature vector we first observed some of its characteristics. We found that all vectors that we obtained have a maximum value that is greater than 0 and a minimum value that is less than 0. Moreover, the mean of all vectors varied slightly between -0.08 and -0.007 while the standard variation ranged between 0.35 and 0.5. If “Coef” is the feature vector of an image than the following quantization scheme converts it to its equivalent code-word:

• If Coef( i ) >= 0 then Coef( i ) = 1 • If Coef( i ) < 0 then Coef( i ) = 0

The next step is to compare two code-words to find out if they represent the same person or not. Test of statistical independence

This test enables the comparison of two iris patterns. This test is based on the idea that the greater the Hamming distance between two feature vectors, the greater the difference between them. Two similar irises will fail this test since the distance between them will be small. In fact, any two different irises are statistically “guaranteed” to pass this test as already proven. The Hamming distance (HD) between two Boolean vectors is defined as follows:

∑=

⊕=N

jBA jCjC

NHD

1

)()(1

Correspondence between the name entered and a chosen eye image. The third option is to identify the person through his/her eye.

Where, CA and C B are the coefficients of two iris images and N is the size of the feature vector (in our case N =

702). The ⊕ is the known Boolean operator that gives a binary 1 if the bits at position j in CA and CB are different and 0 if they are similar. John Daugman, the pioneer in iris recognition conducted his tests on a very large number of iris patterns (up to 1013 iris images) and deduced that the maximum Hamming distance that exists between two irises belonging to the same person is 0.32 [2]. Since we were not able to access any large eyes database and were only able to collect 40 images, we adopted this threshold and used it. Thus, when comparing two iris images, their corresponding binary feature vectors are passed to a function responsible for calculating the Hamming distance between the two. The decision of whether these two images are of the same person depends upon the result:

• If HD <= 0.32 decide that it is same person • If HD > 0.32 decide that it is different person

(or left and right eyes of the same person)

RESULTS AND PERFORMANCE In our project we faced several kinds of problems regarding the quality of the pictures and the lighting conditions. Some of these problems are the occlusion by eyelids, shadow of eyelids, noises, inappropriate eye positioning…To solve some of these, we decided to consider only the lower part of the iris, avoiding thus the eyelids problem. The following table summarizes the primary results of our project.

Table 1. Summary of Primary Results

Percentage

of failure Processing

T ime Number

of Pictures

Edge Detection 5% 100 s 40 Mapping --- 4.28 20

Haar Transform and

Code Word Generation

---

0.5

20

Our primary results show that our system is reliable but further testing is needed. GRAPHICAL USER INTERFACE

To easily manipulate the images in our database we built an interface that allows the user to choose between different options. The first one is to select two images to

35

The flow chart in Figure 7 shows in detail how t hese three options make use of the iris recognition software we implemented.

Figure 7.Flow Chart of the Project

Figure 8.Flow Chart of Iris Recognition

ACKNOWLDGEMENTS

We would like to express our gratitude to our supervisor Professor Adnan Al Alaoui for his continuous support and assistance. We would also like to thank our friends Adham Attalah and Mesrob Ohanessian for their help.

REFERENCES

[1] Daugman, J., “Complete Discrete 2-D Gabor Transforms by Neural Networks for Image Analysis and Compression”, IEEE Transactions on Acoustics, Speech, and Signal Processing, Vol. 36, no. 7, July 1988

[2] Daugman, J. “How Iris Recognition Works”, avai lable at

<http://www.ncits.org/tc_home/m1htm/docs/m1020044.pdf>

[3] Daugman, J., “High Confidence Visual Recognition of Persons by a Test of Statistical Independence,”IEEE transactions on pattern analysis and machine intelligence, vol. 15, no.11, November 1993

[4] Gonzalez, R.C., Woods, R.E, Digital Image Proces s ing, 2rd ed., Prentice Hall (2002).

[5] Lim, S., Lee, K., Byeon, O., Kim, T, “Efficient Iris Recognition through Improvement of Feature Vector and Classifier”, ETRI Journal, Volume 23, Number 2, June 2001

[7] Wildes, R.P, “Iris Recogntion: An Emerging Biometric Technology”, Proceedings of the IEEE, VOL. 85, NO. 9, September 1997

36

Beat the Traffic

Houda Karaki Dept. Electrical and Computer American University of Beirut


Jihad Ibrahim Dept. Electrical and Computer American University of Beirut


Hassane Kamal Dept. Electrical and Computer American University of Beirut


Abstract In this paper, we describe briefly the basic structure and operation of our final year project Beat the Traffic. INTRODUCTION

The main idea of this project is to detect the traffic congestion level in different streets in the Beirut area. This information is incorporated into a map that shows Beirut roads with road colors changing according to the current level of traffic congestion. This service can be accessed through the Internet by users to avoid heavily congested areas, thus rendering their daily transit more efficient. Beat The Traffic is a car congestion-detection system that consists of several stages, which are image cap ture sequence, digital image processing, GIS database update, and XML-based web publishing. Its building parts and functionalities have been designed and implemented using MATLAB as the main driving engine, GIS to update road maps, Visual C++ to generate XML documents and XSL to display the XML documents on the web interface.

A sequence of images is taken through a camera in the street. Several Digital Image Processing (DIP) algorithms are applied to the sequence of images to produce the traffic level results corresponding to the street. The results periodically feed a GIS database, enclosing tables and maps that portray the geographic region of interest. The maps in the GIS database are regularly updated depending on the information extracted by the Image Processing Algorithm. They are then uploaded to a specially designed web site were users can check the congestion levels online. The general flow of our project is described in the block diagram (figures 1 and 2).

Image Sequence Capture

The images are next transferred to a remote server via a wireless link: General Packet Radio Service (GPRS). The transfer is accomplished by either peer-to-peer TCP connection or using File Transfer Protocol (FTP), depending on the server used.

A digital camera or a web cam is installed in an appropriate position to get an effective view of the passing cars. The cam periodically captures street image sequences at a proper rate. The image size is considerably reduced using a down sampling algorithm. A C++

program then automat ically saves the images in the suitable format and location. The image capturing program, down sampling and file transfer take place next to the camera on a processing unit (PC or processing board). Digital Image Processing

The server to which the image sequence is transferred contains the image-processing program. The program is written in MATLAB. Through a series of operations (background subtraction, noise elimination, filtering…), the images are converted to monochromatic frames where moving objects are represented by white pixels. The moving objects are identified, labeled and tracked through consecutive frames to determine their effective speed. This information is used to make a decision on the congestion level. Results are stored in a specially formatted data file that will be next read by the GIS database. Geographic Information System (GIS) The traffic data file, mentioned in the previous section, is sent to the GIS database. To show traffic information on a map of Lebanon, a road theme on the map is used; the table behind the road theme should include a traffic field that reflects the traffic condition on the street. To make the display of the traffic condition dynamic, it needs to be updated regularly with data that reflect the latest traffic congestion levels. This update procedure is done through a script written in Avenue that reads the information from a standard text file and then updates the appropriate record in the road table. The script written reads a road ID followed by the congestion level from a file trafficdata.txt. It then opens the road table and updates the record with the matching road ID. The update is saved, and the result shows on the map. After the view is updated, an image file of the view is stored along with other dat a. The Avenue script then calls a function from a DLL file. The function uses the image file and the data, stored previously, to create an XML file. The XML file and an XSL style sheet are used to update the maps and information that appears on the web interface.

Results

We have built a project prototype consisting of a web cam, connected to a laptop and a GPRS phone. The camera takes image sequence from Hamra Street at constant time intervals. The images are uploaded through FTP to the MATLAB server located in the AUB

37

multimedia labs. Results of image processing are fed into the GIS database that displays the results on a map and publishe them on our web site http://beat -the-traffic.freeservers.com The prototype is operational, but still in the stage of fine tuning. Further results will be provided in the final paper.

ACKNOWLEDGMENTS

Our appreciation to Dr. Ayman Kayssi for providing guidance all through this project. Also a thank you for LibanCell for providing us with equipment.

REFERENCES [1] Lorenzo Favalli, Paolo Gamba, Andrea Marazzi and Alessandro Mecocci, “Tracking by Cooccurrence Matrix” Time-Varying Image Processing And Moving Object Recognition,4 – V.Cappellini, editor.

[2] Frank Muller, Michael Hotter and Rudolf Mester, “Moving Object Detection in Image Sequences Using Texture Features” Time-Varying Image Processing And Moving Object Recognition,4 – V.Cappellini, editor. [3] A. Tesei, A. Teschioni, C.S. Regazzoni and G. Vernazza, “Long-Memory Matching of Interacting Complex Objects From Real Image Sequences.” Time-Varying Image Processing And Moving Object Recognition,4 – V.Cappellini, editor [4] Dieter Koller, Joseph Weber and Jitendra Malik, “Robust Multiple Car Tracking with Occlusion Reasoning”. ECCV (1) 1994 189-196 [5] John C. Russ “The Image Processing Handbook” Second Edition, CRC Press [6] Ioannis Pitas “Digital Image Processing Algorithms” Prentice Hall International Series in Acoustics, Speech and Signal Processing. [7] Microsoft Developer’s Network July 2001

A camera on the street takes a sequence of digital images of the street every x milliseconds.

A computer board with a down-sampling algorithm will down-sample the images for transfer over GPRS

The down-sampled images are sent over a GPRS link to the main server

The main server contains the digital image-processing algorithm, which is fed the down-sampled images. The server will run the program, produce the traffic congestion data, and store the data in a file.

Traffic Data File: Information that identifies the street, number of cars, time, traffic level, …

A script written in Avenue does two things:

1) Take the data from the traffic data file and update the appropriate GIS table

2) Call a DLL file that contains a C++ function responsible for creating XML documents.

1

2

Traffic Traffic Data File

Traffic Traffic Data File

Figure 1 Beat The Traffic Flow Chart

39

Web Data

File

A script written in Avenue does two things:

1 2

Web Data File Web Data File

Maps and Info

1.2 1.1

First the record with the corresponding record ID is updated

The road view map is updated and stored, along with other information in a file

.dll

The script calls a function from a DLL file. The function uses the Web data file to create an XML file. The XML document and a XSL style sheet are used to update the web interface.

Web Data File

Figure 2 Beat The Traffic: GIS Module

40

Perspective Image Formation Using Omnidirectional Vision

Adel T. Zouheiry Department of Electrical & Computer

Engineering Faculty of Engineering & Architecture

American University of Beirut Beirut, Lebanon

[email protected]

Mounir M. C. Doumani Department of Electrical & Computer

Engineering Faculty of Engineering & Architecture


[email protected]

Abstract Omnidirectional vision as its name states allows to have a field of view large enough to cover all directions. Many systems have been designed in order to achieve this operation, some using complex refractive elements like fish eye lenses, and some using complex mechanisms such as rotating cameras or multiple cameras. But these systems have their drawbacks, fish eye lenses for example add nonlinear distortion to the image, while rotating cameras need very precise control and high shutter speed in order to obtain blur free pictures without the necessity of applying any processing techniques. An interesting solution that emerged recently is the use of catadioptric sensors. Catadioptric systems involve the use of mirrors and lenses to reach the desired result, but produce a distorted image that needs to be corrected. In this paper, unwrapping, which is the process involved in reconstructing undistorted image from the output of a catadioptric system, will be analyzed thru three different approaches. And finally we will go over an application in which perspective unwrapping could be used.

Keywords Omnidirectional, Catadioptric, Parabolic, Unwrapping, Panoramic, Cylindrical, Perspective, Epipolar.

INTRODUCTION Research, lately, has been directed toward imitating human behavior in order to get more anthropomorphized models. Consequently, cameras were used to mimic the human eye, stereo systems were designed to get the feel of depth we human have, but still, one step further toward breaking the limitations imposed by those kind of scheme needed to be taken. One of the solutions is the use of Omnidirectional Vision which would result in a larger field of view compared to the traditional imaging systems while conserving a single image plane. Omnidirectional vision can be obtained by using many systems built on different concepts as diverse as the use of fish lens image or rotating cameras. But these systems have proven to be expensive or complex to manufacture. Another alternative that appeared in the middle of the 1990’s is the use of catadioptr ic sensors [8], [9]. As their name indicates catadioptric sensors are built out of

a combination of reflective and refractive materials. This paper will start with a brief description of catadioptric sensors that have been studied by Shree Nayar, followed by the formation of panoramic, cylindrical and perspective images out of the wrapped image produced by these kinds of sensors. Finally we will demonstrate the usefulness of perspective image formation for 3-Dimensional reconstruction [3].

CATADIOPTRIC SENSORS Figure 1 depicts a typical catadioptric system formed by a lens and an image. A ray of light emitted from a point in space is reflected on the mirror. The ray is then directed towards the lens that directs it towards the corresponding point on the image.

Figure 2. A Typical catadioptric system.

The main issue when designing the mirror of the sensor is to have a single viewpoint in order to be able to unwrap the resulting image. Many mirrors meet this requirement [2] and have been used in robotics application. As an example, at the Université de Picardie Jules Verne a conical mirror by the name of SYCLOP was used to orient the mobile robot. In our experiments and applications we have decided to use a parabolic mirror due to a considerable advantage that this kind of mirrors have on other conical shapes. The additional property is orthographic projection, meaning that every ray of light directed thru the focal point of the mirror is reflected perpendicularly towards the image plane [1], [7], [9]. Another useful property that is not to be neglected is the fact that the geometry of the parabolic mirror is only a function of a single parameter H. These two properties simplify significantly the

41

Figure 4. Wrapped image

Figure 6. Result of panoramic unwrapping.

computational effort involved in unwrapping and Calibration because the distance between the mirror and the camera is not a calibration parameter and the mirror can be characterized only by H which is the outer radius at the focal point level.

The mirror used in our experiments is not a simple parabolic mirror, but what Nayar defined as a folded optical path, that means a combination of mirrors which can be modeled as a single mirror [10]. In our particular case, we modeled our system by a parabolic mirror.

Figure 3. S ystem used in our experiments.

UNWRAPPING

As stated previously the output of the catadioptric system is a distorted image that needs to be processed in order for it to become a useful source of information. The process of unwrapping is the transformation of the obtained circular image into a desired image formation according to a desired view plane. At this stage the choice of the view plane is critical for the next stage as, depending on the application, perspective views or 360° panoramic views may need to be generated. For robot navigation, panoramic unwrapping may be sufficient and comput ationally efficient as it can be seen in table 1. But for more complex tasks involving object recognition an undistorted perspective image is more adequate.

We will discuss next the following unwrapping methods: • Panoramic unwrapping

• Cylindrical unwrapping

• Perspective unwrapping

.

Panoramic Unwrapping

Panoramic unwrapping is achieved by taking the view plane to be coplanar with the wrapped image. This kind of unwrapping does not take into consideration the particular shape of the mirror; consequently it is subjected to a non-linear transformation about the vertical axis of the obtained panoramic image.

Because the view plane and image plane are coplanar there is only a need to translate the donut shaped wrapped image into a rectangular view plane. A point P(xp,yp) (Figure 4) on the view plane is mapped back to a point P’(xi,y i) on the wrapped image plane by considering yp the arc length of the outer radius of P’ and xp the radius of P’. Due to resolution constraints, mapping from P’ to P end up with increasing pixel loss as P’ has a smaller radius [6]. As a result, interpolation and filtering techniques should be implemented in order to recover the pixel loss. Thus mapping from P to P’ is preferable as it solves by itself to the resolution issue.

Figure 5. View plane and image plane in panoramic

unwrapping.

As yp is the arc length we can say that:

radiusouteryp _×Φ=

Where Φ is the angle formed by 'OP and the Xi axis and outer_radius is a parameter specific of the mirror.

=> radiusouttery p _/=Φ (1)

pxradius = (2)

From (1) and (2):

)cos(Φ×= radiusxi (3)

)sin(Φ×= radiusyi (4)

Now for each point P (xp,,yp) , P’(xi ,,yi) can be calculated and the resulting panoramic image formed. The final result can be seen in figure 5.

42

Cylindrical Unwrapping

A midway between panoramic unwrapping and perspective unwrapping can be obtained using cylindrical unwrapping. The resulting image is also panoramic representation of the surroundings but it preserves proportionality along the Xp-axis. This unwrapping technique consists on taking a cylinder as a viewplane. The properties of this cylinder are:

• Rotation axis corresponds to the axis of the vertical mirror.

• Radius is equal to that of the upper cross section of the mirror, thus the outer radius of the picture as a result of orthographic projection.

• Height is equal to that of the mirror.

Figure 7. Cross section of cylindrical mapping geometry.

The procedure to find xi and yi is the same as in panoramic unwrapping as the final step of the mapping will use equations (3) and (4). The difference lies in the computation of the radius r used in the multiplication. From figure 6, we can find using similar triangles:

xh

zr

= (5)

From the geometry of the mirror we can deduce (6)

hrhz

2

22 −= (6)

By replacing (6) in (5) we end up with a quadratic equation (7) that we solve for r and only one of the two solutions is acceptable as the second is negative.

02 22 =−− hxrr (7) 22 hxxr ++=

knowing r and F we can find to which point on the wrapped image the points of the cylindrical image map.

Figure 8. Result of cylindrical unwrapping

Perspective Unwrapping

To obtain perspective images from the wrapped image, it is necessary to choose a rectangular surface as a viewplane. A choice could be the four planes tangent to the mirror or any other rectangular surface defined by a direction, location and dimension. This method results in a separate image for each plane defined as it can be

seen in figure 8. To obtain the desired perspective images it is necessary to find first a transformation between any world point and the image point which is also a generalized case for the two previous unwrapping methods. Let us define Pw as a point in the world frame, Pm a point in the mirror frame and finally Pi a point in the image frame, and assuming that those three frames coincide we can conclude the following:

• Any point Pw with rectangular coordinates xw yw zw has its spherical coordinates as follows:

)arctan(w

ww x

y=Φ

)arccos(222www

ww

zyx

z

++=Θ

=wρ 222www zyx ++

• Any point Pm will have the same angles F and ? than the world point Pw, because Pm is on the line joining Pw to the mirror’s focal point:

wmwm Θ=ΘΦ=Φ &

• Knowing the equation of a parabola in spherical coordinated:

)cos1(1

Θ+=mρ

• We can now obtain the rectangular coordinates of Pm, xm , ym, and zm using the spherical to rectangular transformation:

)cos()sin( Φ×Θ×= ρmx

)sin()sin( Φ×Θ×= ρmy

• Because of the orthographic property of the mirror, any point Pi has the same x and y coordinates as the corresponding Pm point:

mimi yyxx == &

Combining the transformation between Pw and Pi with a well defined view plane, desired perspective images can be computed. It is important to note that for panoramic and cylindrical unwrapping the transformation between Pw and P i could have been applied, but for better computational efficiency, we described direct geometrical methods.

Figure 9. Result of perspective Unwrapping.

43

Table 1. Figures of unwrapping time on a 900x900

picture and different platforms

Unwrapping Platform Panoramic Cylindrical Perspective Matlab

(Windows Intel PIII 1GHz 256 MB RAM)

4 min. 6 min. 12 min.

C (Linux Intel PII 350 MHz 96 MB RAM)

6 sec. 10 sec. 45 sec.

AN APPLICATION: 3-D RECONSTRUCTION

3D Reconstruction is achieved using stereo vision systems. In omnidirectional vision, stereo vision is obtained thru several methods, to cite a few: dual omnidirectional system [3], mirrors with two focal points [4], [5]. Those techniques in general require the development of new pro cesses for correspondence, epipolar geometry. In these cases, a suitable view plane should be chosen to fulfill the desired requirement. Another method would be to use the generated perspective images with well developed traditional stereo vision techniques. This kind of technique is less time consuming in development and more flexible mainly because of the large number of 3D reconstruction techniques developed for traditional dual imaging systems. This could be achieved by taking two pictures at different but known locations, unwrapping both pictures using a perspective view plane and proces s ing both images in order to extract 3D information using regular epipolar techniques.

CONCLUSION

The use of catadiptric systems revealed to be a successful new trend in 360 degrees imaging systems. But in the transformations developed above, the internal radial distortion was not taken into consideration, and this is perceptible in our results. Also modern omnidirectional sy stems use a folded optical path as in our system so there is a need to take into consideration the transformation resulting from the combination of mirrors. In our case we approximated it to be a single parabolic mirror result ing in acceptable additional radial distortion.

.

ACKNOWLEDGMENTS

Our thanks go to Dr. Samer Abdallah for his precious assistance and unconditional availability.

REFERENCES

[1] Baker, S., & Nayar, S. A Theory of Catadioptric Image Formation, Proceedings of the 6th International Conference on Computer Vision Pages 35-42, Bombay, India, January, 1998

[2] Baker, S., and Nayar, S. A Theory of Single-Viewpoint Catadioptric Image formation , International Journal of Computer Vision, 1999, Kluw er Academic Publishers

[3] Roland Bunschoten and Ben Kröse, Range Estimation from a Pair of Omnidirectional Images , IEEE International Conference on Robotics and Automation, 2001, Seoul, Korea

[4] Conroy T.L. and Moore J.B. Resolution Invariant Surfaces for Panoramic Vision Systems. Submitted to the Intern ational Journal of Computer Vision, April 1999. [5] Conroy T. L. and Moore J.B. View Based Design for Stereo Panoramic Vision Systems. In proceedings of the 4th Asian Conference on Computer Vision (Taiwan ), January 2000.

[6] Gatcher S. Mirror design for an Omnidirectional Camera with Uniform Cylindrical Projection when using the SVAVISCA Sensor, Center of Machine Perception, Department of cybernetics Czech Technical University, Prague, March 2001

[7] Geyer C., & Daniilidis K. Catadioptric Projective Geometry, International Journal of Computer Vision, December 2001

[8] Ishiguro, H. Development of Low-Cost Compact Omnidirectional Vision Sensors and their Application, International Conference on Information Sy stems, Analysis and Synthesis , pp. 433-439, 1998

[9] Nayar, S. Omnidirectional Vision, Proc. of International Symposium on Robotics Research, 1997, Japan

[10] Nayar, S. & Peri V. Folded Catadioptric Cameras, Proc. of IEEE Conference on Computer Vision and Pattern Recognition Fort Collins, June 1999

44

Figure 11. Speech waveform for the word "Nadine".

Speech Recognition for Voice Based Control

Youssef Naous

ECE Department, American University of Beirut, Lebanon

y_naous@hotmail,com

Ghinwa Choueiter ECE Department, American

University of Beirut, Lebanon [email protected]

Mesrob Ohannessian ECE Department, American

University of Beirut, Lebanon [email protected]

Abstract In this paper, we describe a typical approach for implementing a voice based control solution. Isolated word speech recognition is performed using cepstral feature extraction and hidden Markov modeling of speech. The merit of this document lies in the amalgamation of the simplest yet most successful relevant methods into a coherent design guideline, aiming to trivialize the integration of speech technology into daily applications.

Keywords Speech recognition, voice based control, isolated word, cepstral feature extraction, hidden Markov model. INTRODUCTION

Although speech recognition technologies have greatly matured, they are still not widely embraced, apart from computer -based dictation, where a controlled environment and excess computing power even out technical shortcomings, such as noise and speaker dependence. However, in simpler domains such as the voice-based control of various appliances, the adoption is not as prominent. In this case, we believe that what will balance out the situation is the existence of certain design guidelines, incorporating both theoretical and technical aspects of speech recognition, such that they could lead to the quick prototyping of voice-enabled devices. As an attempt to establish such a guideline, we present in this paper the major elements of a project we undertook, on the integration of isolated word speech recognition into arbitrary applications. We address specifically the essential theoretical background required for isolated word recognition, which relies on two elements: the extraction of features from the recorded utterance, followed by the training or the classification of the succession of features into a given word model. Several options have been validly demonstrated regarding these two elements. We describe our own choices: cepstral coefficients and hidden Markov modeling, justified from standpoints of accuracy, performance and simplicity. The adopted pro cess is illustrated in Figure 1. After this software description, we elaborate on the procedures and modifications required for porting such a speech recognizer to a compact hardware system. This consists mainly of dealing with underflow, overflow, floating-point to fixed-point transformation, and various optimizations, all of which are either general or specific

issues of concern, based on whether one is working on a DSP or is designing a custom processor.

Figure 10. Flow chart describing the various stages of an

isolated word speech recognition system.

FEATURE EXTRACTION

In the Figure 2 below, the waveform representing the word “Nadine” is shown. As in most pattern recognizers, front -end processing transforms such a waveform into an ordered set of parameters. This stage, referred to as feature extraction, consists basically of a speech analyzer that will yield some more convenient, or more compact, representation of the signal, as is the case in the area of coding.

45

One such powerful technique is linear predictive coding (LPC hereafter), which provides good quality features, and is also used to encode speech at a low bit rate. LPC decreases the variability of the signal by attempting to determine the envelope of the signal spectra, as illustrated in Figure 3 for a single frame inside the first “N” sound in the word of Figure 2. For more information on LPC, [1] and [2] can be consulted.

Figure 12. Spectral envelope as described by LPC analysis.

Before reaching the LPC stage, our feature extraction section proceeds as follows. First, in order to reduce the high spectral dynamic range of the signal, a first-order pre-emphasis digital filter is applied, represented by equation (1). Figure 4 is the pre-emphasized “Nadine”.

( ) 0.19.01 1 ≤≤−= − aazzH (1)

Figure 13. Pre-emphasized waveform of the word "Nadine".

Next, observing the fact that the vocal tract moves slowly, mechanically, speech can be considered stationary over a relatively short signal frame. Therefore, overlapping frames of 16 ms duration-width are extracted every 5ms of the speech signal. A Hamming window, as expressed in equation (2), is applied per frame.

( ) 1Nn01N

n2cos46.054.0nw −≤≤

−π

−= (2)

The overlap is to increase the correlation of the LPC spectral estimates from frame to frame, and the windowing is to taper the signal to zero at the beginning and end of each frame. In order to facilitate the calculation of the predictor estimates, we use the “Levinson-Durbin” algorithm, which derives LPC coefficients from an autocorrelation computation, where the highest autocorrelation value, p, is the order of the LPC analysis, which we have selected to be 8, at a sampling rate of 8 kHz. The algorithm is an iterative one, and is summarized in Table 1 below. The final iteration α values are the LPC coeff icients. For good coverage of the procedure, refer to [3]. A drawback of LPC estimates is their high sensitivity to quantization errors, and small differences could lead to unreliable results and models.

Table 2. The Levinson-Durbin Algorithm

Autocorrelation ( ) ( ) ( ) pmmnxnxmrmN

n

,...,1,01

0

=+= ∑−−

=

Initialization )0(rE =

Iterations pi ,,2,1 K=

( ) ( ) ( ) Ejirjirki

j

−−= ∑

−

=

1

1

α

( ) ki =α ( ) ( ) ( )jikjj −−= ααα , 1,,1 −= ij K

EkE )1( 2−=

( ) ( )jj αα = , 1,,1 −= ij K

A remedy is to use LPC-based forms, the most common of which are the cepstrum coefficients, which is in essence the Fourier transform representation of the log-magnitude spectrum. These estimates have been proven to be a robust and reliable set of features in speech recognition applications. Table 2 presents an iterative algorithm for the determination of cepstral coefficients from that of the LPC. The cepstral order, q, which we have set to 12, is generally chosen to be greater than the LPC order p.

Table 2. Cepstrum Determination

Iterations pm ,,2,1 K= ( ) ( )∑

−

=

−

+=

1

1

m

kkm ckm

mk

mc αα

Iterations qpm ,,1 K+= ( )∑

−

−=

−

=

1m

pmkkm ckm

mk

c α

However, the low-order and high-order cepstral coefficients exhibit sensitivity to overall spectral slope and noise respectively. Hence, it has becom e a standard method to apply a window to the coefficients in order to minimize the sensitivities. Although a proper window can be mathematically derived, scaling by the following quadratic function proves to be an effective alternative:

qiqq

qiqiiS ,,2,1,

2244

)( 2

2

L=+

++−= (3)

The result is known as the weighted cepstral coefficients, and this set of coefficients is our final desired feature vector. Figure 5 illustrates clearly the advantage of such a representation: the plots represent the first “N” sound, in different utterances of the word “Nadine”. Observe how the coefficients are consistent throughout.

Figure 14. Cepstral coefficients for the first "N" sound

in various utterances of the word "Nadine".

Finally, we optimize the speech recognition system by compressing the feature vectors using vector quantization. The VQ procedure approximates an infinite set of vectors by a limited set known as a codebook, which is generated from speech samples according to some optimality measure and for the

46

Gaussian noise model. We note that we obtained best results using a codebook of 32 elements, generated by a K-means algorithm applied on a long speech sample, representing most of the words to be modeled. The details of the process can be found in [3]. This approach generates the codebook once; all subsequent feature vectors are then classified to one of the codebook-elements, according to the least distance criterion. As such the vector space is divided into quantization (Voronoi) regions. Figure 6 shows how the codebook elements cover evenly the cepstral coefficient space. With this last step, our original speech signal is converted to its final representation: a sequence of codebook indices, which will be our observation sequence, as referred to in the next section.

Figure 15. Vector quantization. (∗)s represent the frames,

while (∇)s represent codebook elements.

MODEL REPRESENTATION

One of the most successful speech models is the first order hidden Markov model, a simplified stochastic process model based upon the Markov chain, and it was the one adopted for our project. A discrete hidden Markov (HMM hereafter) model, λ, consists of a finite number of states, to which we associate certain probability distributions, summarized in T able 3 below.

Table 3. HMM Probability Distributions (∗)

MODEL λ PARAMETERS

Transition Probabilities tisjsPA ttij ∀=== + |1

Observation Probabilities tisOOPOB tti ∀=== |)(

Starting Probabilities isPPi == 1

AUXILIARY DISTRIBUTIONS – TIME S PECIFIC

Forward Variable i,s,...,O,OOP(i)a ttt == 21

Backward Variable isOOOP(i) tTttt == ++ |,,, 21 Kβ

Specific-Time Transition Tttt OjsisPji K11 |,),( === +ξ

A Posteriori Probabilities Ttt OjsP(i) K1|==γ

(*) t = integer time, i.e. frame number. T = number of observations, i.e. frames.

Transitions among states, say from i to j, are the result of a transition probabilities Aij, while there exists another set of probabilities Bi(O) according to which a given outcome or observation O is generated, given the fact of being in a precise state i. Yet a third set, Pi, gives us the probabilities of starting the model in state i. An

important remark to mention here is that by “observation” we mean a feature, which is the codebook index associated with a given frame, as explained in the previous section. Since, the user only sees the outcomes, and not the states generating them, the latter are referred to as “hidden”, whence the name hidden Markov model. Figure 7 illustrates an HMM model, transitional probabilities are shown, while starting and observation probabilities are not. The choice of the number of states is arbitrary, but experience has shown that 6 states are adequate for the task. For one of the best references in hidden Markov modeling of speech, consult Rabiner’s tutorial paper [4].

Figure 16. HMM States and Transitions.

Note that Table 3 also provides four additional convenient distributions. The first two, the forward and backward probabilities, are useful in solving two important HMM problems: training, which will predispose the models to specific words, and recognition, which will choose the model most likely to be the uttered word. It can be shown, [4], by induction, that:

∑=

++ =N

iijttjt (i)Aa)(OB(j)a

111

(4)

∑=

++ ××=N

1j11tt )(OBA(j)(i) tjijββ (5)

The training procedure that we have adopted is known as the Baum-Welch algorithm. It re-estimates the model, i.e. A, B and P, through the two other distributions from T able 3, ξ and γ , computed using the forward and backward variables:

( ) ( ) ( )

( ) ( ) ( )∑∑= =

++

++

×××

×××= N

i

N

jtjij

tjtijt

OBjAi

OBjAiji

1 111tt

11t),(βα

βαξ

(6)

( ) ( ) ( )( ) ( )ii

iiiN

itt

tt

∑=

=

1

t

βα

βαλ (7)

T he re-estimation equations are as follows:

( )( )

( )( )

( )

( )∑

∑

∑

∑

=

==

−

=

−

= ===T

tt

T

OOt

t

iT

t

T

tt

iji

i

i

OBi

jiAiP t

1

1

1

1t

1

11

,

γ

γ

γ

ξγ

(8)

The derivation and other underlying detail of the algorithm can be found in [4], and are not repeated here, for brevity. Each word model should be appropriately trained, i.e. around 5 or more times, with the equations

47

above, using different utterances of the correct corresponding word. Once models are prepared as such, we perform the recognition of an unknown utterance by evaluating the matching likelihood against each model using the following equation:

( ) ( )iiisOPOPN

itt

N

itTT ∑∑

==

===11

11 |,| βαλλ KK

By selecting βT(i) to be all ones, the summation above reduces to a summation of the forward αT(i), and is known as a forward algorithm. The action of recognition will then be the selection of the most likely model, but only if a certain confidence measure is satisfied, i.e. if the difference between the likelihood of the matching model and the rest is greater than a threshold. The only detail that is yet to be elucidated is the initialization of the word models. Incidentally, most initializations meeting the stochastic constraints, i.e. probabilities summing to one, are valid. However, better results are obtained if we use the very first utterance to initialize the corresponding model. For that, we create an artificial traversal of the states, by first assigning random observations to each state, then using K-means distribution, as in vector quantization, to end up with initial state statistics yielding P, transitional statistics yielding Aij, and observation statistics yielding Bi(O).

PORTING TO HARDWARE Most of the algorithms used in the speech recognition scheme described above require floating point arithmetic. This is the main portability issue to settle, since most DSP-based embedded systems are only capable of fixed-point arithmetic, not to mention the difficulty and cost of implementing an FPU in a custom processor. Many design platforms support flexible data types for fractional number representation, such as the Q.15 format adopted by Texas Instruments in their TMS320 series, a specimen of which (the C5402 DSK) was used for the prototyping of our project. However, Q.15 is a normalized number format, i.e. all numbers are in the range [–1,1], and whenever computation results are to exceed that range, some sort of overflow control is necessary. Such a situation is encountered during feature extraction, where both the autocorrelation and LPC computations overflow. For the former, signal scaling solves the problem. For the latter, a new variable is introduced, such as:

Sαβ = (10)

Here, S is a scaling constant larger than one. This substitution is effective, since the only place where overflows do occur is in the determination of α. The resulting modification is the use of β instead, in addition to scaling the summation by S, in the expression where k is determined, and setting β(i) to k/S for every iteration i. Beyond the basic representation issue, it is worthy to mention that HMM algorithms also suffer from very small numbers in temporary variables, which may yield to underflows, even on floating point systems. The

solution, in this case, is in an alternative algorithm that has been proposed in [4]. However, that section of the document contains some errors, corrected subsequently in various citat ions, such as the errata paper [5]. Last but not least, some parameters that are otherwise arbitrary in software are no more so for a hardware implementation. The primary issue being memory limitations, in-place processing is desirable whenever possible. In what concerns DSP applications, the use of optimized libraries, such as the one documented in [6], improves considerably the computational efficiency. CONCLUSION In speaker dependent mode, our current implementation of the speech recognizer is capable of achieving 96% accuracy on a small vocabulary of 10 words and in normal conditions, which however deteriorated to 90% in the presence of noise, i.e. unrelated background speech. Since similar performance is to be expected from any system following as guideline the specifications outlined in this paper, the procedure can be used to build useful voice-enabled devices, or to integrate voice capability into existing ones, and has as such achieved its primary purpose. Further work should deal, amongst other aspects, with the use of the system in adverse conditions. Stress should be put as well on achieving speaker independence, which is currently an active research area.

ACKNOWLEDGMENTS We would like to acknowledge our supervisor, Prof. M. A. Al-Alaoui for his guidance and encouragement. We are grateful as well to Mr. R. Ferzli, for providing us with relevant information and technical help during the various stages of the project. We thank also our colleagues at the ECE department of the American University of Beirut for keeping a high spirit throughout the project.

REFERENCES [1] Barnwell, K., Nayebi, K. and Richardson C.H.,

Speech Coding: A Computer Laboratory Textbook , John Wiley & Sons Inc. (1996).

[2] Makhoul, J. “Linear prediction: A tutorial review'”, Proc. IEEE, vol. 63, pp. 561-580, April (1975).

[3] Rabiner, L. and Juang, B.-H., Fundamentals of Speech Recognition, PTR Prentice Hall, NJ (1993).

[4] Rabiner, L., “Tutorial on Hidden Markov Models and Selected Application in Speech Recognition”, Proc. of the IEEE, vol. 77, no. 22, pp. 257-286, Feb. (1989). Contains also an appreciable bibliography.

[5] Trebbe, H., “Correction of Errata in Rabiner’s Tutorial on HMM”, May 19, (1995), available at <http://santana.uni-muenster.de/Publications/>

[6] “TMS320C54X DSP Library User’s Guide”, (2001), Texas Instruments documentation available at <www-s.ti.com/sc/psheets/spru518a/spru518a.pdf>

48

Mobile.Music

Joe Abou-Rjeily 4th year CCE, AUB


Jade Bissat 4th year CCE, AUB


Mario Harik 4th year CCE, AUB


Abstract In this paper, we provide a brief description of our Final Year Project, one of our graduation requirements for the B.E in Computer and Communication Engineering from the American University of Beirut.

Keywords Streaming, Mp3, Cell phone, Internet Radio.

INTRODUCTION

For our final year project at the American University of Beirut, we have worked on designing and implementing an audio streaming application for mobile devices; hence the project's name "Mobile.Music".

OVERVIEW OF THE CURRENT SCENE

Given the preeminence of cell phones these days, and the "download culture" created by the growth in quantity and quality of digital data in general - and audio mp3's in particular-, it seems particularly interesting to merge these two aspects so as to provide improved services for the user. Thus, we are starting to see burgeoning projects, from Nokia for instance, that relate to these services. However, they require a particular cell phone, specifically designed for that task.

OUR OBJECTIVE

Our objective is to design a prototype of a generic cell phone music download and playback system. It will be a prototype, of course, since we need to use components more powerful (and therefore more expensive) than what would be strictly needed, in order to allow for testing and development. It is supposed to be a generic system, meaning it should be independent of the cell phone used (our only need is to have a GPRS/WAP enabled cell phone).

PROJECT DESCRIPTION

The Different Parts

Our implementation is composed of the following parts: • A GPRS and WAP enabled cell phone.

• A device for the decoding of the music, along with a soundboard for the playback.

• A WAP/CGI server.

• A control server program.

• A control client program .

• Audio streaming servers.

• An audio (mp3) decoding streaming client.

The Design Requirements

When design decisions were made, the most significant parameter for our choice was how close (or far) it would pull us towards (or away from) our three basic tenet s: portability, mobility, and light load. Device design requirements:

The device is a uCdimm Coldfire board, and is a uCLinux powered reduced computer. Therefore we have uCLinux software on it to perform the desired control and decoding tasks. This gives us the ability to work with the C language, which is ideal for our needs. We have selected this device for its small size (which allows mobility) and programmability. Cell phone design requirements:

Since the cell phone (particularly the WAP component) and the device do not have an intrinsic communication capability (and since we want our device to work with any GPRS enabled phone), these services also require the mostly small-scale participation of a control server to transmit control information from the cell phone to the device. This additional step, which might not have been needed if we could communicate directly between the cell phone and the device, is due to our second tenet of portability. The system has the advantage of working independently from the brand/OS of the cell phone. Control layer design requirements:

Finally, instead of having the server capture the stream from the radio, and then sending that stream to the device, that device has the ability of streaming the audio directly from the Internet radio, thus reducing the server’s contribution to a control aspect only. This conforms to our third tenet of low load imposition on the server.

How They Interact

In our design (see Figure 1), we start by assuming the servers are turned on. It becomes then up to user to start the flow. The user starts by turning on the device, which connects through the GPRS protocol to the control server. The user then accesses the WAP site of the system through the cell phone. There, s/he will enter the personal device Id and password, and then enter a search menu. After setting the desired constraints, a choice of accep table audio servers will be listed. The user will then select the desired listening option (note that s/he can also choose to stop any running playback).

49

The WAP/CGI server will then inform the control server of the address and port of the desired choice, and the server will transmit that information to the control client device. That client will then start the mp3-streaming client on the device, and the sound will be outputted from the soundboard. Note that the streaming client also communicates directly with the streaming servers through GPRS.

COMMERCIAL PERSPECTIVE

Financial Aspect

From a business aspect, this project can be viable provided that the cost of the decoding device is slashed. The control server, for instance, does not have a lot of work to do, and its running costs can be covered easily by a subscription scheme. The device, however, is quite costly at this stage. This is because we are building a prototype, and thus we need a programmable device to experiment with our code every time. A dedicated printed board, along with a chip of

limited computational capabilities, could be used to reduce costs.

Potential Obsoleteness

Some might argue that this project will become obsolete after the first reasonably -priced high -level cell phones with such capabilities are on the market. But this will not be the case, since our device could be use for "back-compatibility" of older phones, and will remain viable even after that if combined costs for a simple cell phone and the device are less than the cost of a souped-up "music phone". be centered above the table body.

ACKNOWLEDGMENTS

We would like to express our gratitude towards Dr. Ali Hajj, our esteemed advisor, for his continuous assistance; and Dr. Ayman Kayssi, for his interest in our project. We would also like to extend our gratitude to our sponsor, LibanCell, for their material contribution; and to Salem Itani, for his precious help in the WAP segment.

Figure 1: Operational Project Flowchart

50

Mobile Jammer

Paula Atallah ECE Dept., FEA


[email protected]

Najwa Hamzeh ECE Dept., FEA


[email protected]

Elias Nahra ECE Dept., FEA


[email protected]

Abstract This paper discusses the design and implementation of a mobile signal scrambler, mobile jammer. Such jammers are already in the market, manufactured by few foreign leading companies. However, they have been criticized for not meeting two of their expected specifications, namely, durability and area of coverage. Above all, they are being too expensive. Our goal is to implement a new advanced technique to provide a better product in terms of design, per formance and cost.

INTRODUCTION

A mobile jammer is a device that cuts-off communications between cellular handsets and cellular base-stations. A jammer is not supposed to interfere with any communic ations devices other than cellular phones within its zone of operation. It also must not interfere with cellular communications outside the regulated area. Commercial jamming operations include prevention of cellular phone use and abuse in working environments such as offices and factories, and public buildings like movie theatres, libraries, restaurants, temples, churches, hospitals and pri sons.

RESEARCH

All cellular phone sy stems used in all countries employ simultaneously two different groups of frequencies. This is to enable the user to communicate in full duplex, i.e. to speak and listen simultaneously (unlike a standard talkie-walkie). The first group of frequencies is used for the downlink communication between the cellular towers and the mobile phones themselves. This group is usually assigned the upper range of the GSM bandwidth. The second group is used for the ’uplink’ between the phone and the tower.

DESIGN

We utilized a unique transmission method for our advanced jammer. As mentioned earlier, jammers in the market today are functioning only for a short time; this is due to the fact that their transmission requires high power to cover the whole GSM uplink or downlink bandwidth which is 25 MHz in Lebanon, thus the electronic components they have been using are burning too easily out of the high heat dissipation. In our design, we made sure to avoid this problem: we will be transmitting a signal of low bandwidth (500 KHz) and then modulate a GSM level carrier by it, then we will be

shifting this carrier up and down the downlink bandwidth at a rate high enough to cancel any GSM signal for a calculated area.

We thoroughly considered several design techniques and accordingl y have arrived to the most suitable and feasible design. Figure 1 clearly shows our method of generating and transmitting noise at radio frequency.

Figure 17 – Block diagram design

Knowing that the downlink GSM channels in Lebanon range from 935 to 960 MHz, we set VCO1 to 934 MHz and VCO2 will be controlled by a variable capacitor, varactor, and will range from 1 to 26 MHz. Now multiplying the outputs of both VCOs, we will have one signal to be transmitted which ranges from 935 to 960 MHz. Thus, we achieved our purpose.

PRELIMINARY RESULTS

To start applying the theoretical design that we have agreed upon, we had first of all to look for the necessary electronic components that function at GSM frequencies. We could hardly find any of them in Lebanon. We then searched for them on the web and asked our sponsor, MEICO, to get them as soon as possible. Meanwhile, we had been working on the same design but on lower frequencies (FM level). We were able to jam the transmission of various FM radio stations over a certain area.

The sponsor also had problems getting the components necessary for implementing the design. The only option left for us was to call the manufacturers of the components and order the components ourselves. Some

Cry stal

VCO 1

Mixer

Low Noise

GSM Filter

Varac tor

Cry sta l

VCO 2

PLL

51

companies also were not ready to ship them overseas to Lebanon due to the 11 September crisis; a friend of us living in the US then offered to order them and reship them to Lebanon. Until now most of the components arrived and few are left.

In the meantime we tried to simulate our design using ORCAD 9.2; this was not possible because the ICs we were using were not found in the ORCAD database. Nevertheless, we did a rough simulation using MATLAB. Figure 2 shows how the carrier frequency we are transmitting moves along the GSM downlink bandwidth used in Lebanon (935 MHz to 960 MHz).

Figure 2 – MATLAB simulation of the design

CONCLUSION

Having spent few months on our final year project, we gained great experience through our progressive work. We learned how to organize a big project on a group

basis and to manage our time schedule in an efficient manner. We acquired tremendous knowledge on handling large-scale projects. We became familiar with advanced engineering topics related to telecommunications such as GSM transmission and Spread Spectrum techniques; we also strengthened our knowledge in certain topics or fields we were taught during the four years spent at the university.

We expected to accomplish our jammer some time before the deadline. Unfortunat ely, we were not able yet perform the testing and troubleshooting; that is due to the fact that the components arrived on May 24, 2002. We hope that, in the coming few days left, we will be able to introduce a new product -cellular jammer with high quality and competitive price to the worldwide.

ACKNOWLEDGMENTS

We are very grateful to Prof. Karim Kabalan who provided us with the academic encouragement and motivation. His remarkable support was essential for us to keep up our progress in the project. We would like to thank Dr. Walid Ali Ahmad for his technical advices through which we were able to enhance our design. Special appreciation goes to MEICO, our sponsor, who provided us with a work bench at their company and assisted in building the PCBs necessary for testing. We are most grateful to Mr. Mohamed Darwish who made our final design possible by getting us the ICs needed from the US.

REFERENCES

[1] Mc Swiggan, F., “To Design and Build a Portable, Miniaturized, Multichannel FM Transmitter”, University Of Limerick, avai lable at < www.csn.ul.ie/~francis/fyp_report/report.html >

[2] Glas, J., “The principles of Spread Spectrum communication”, Last modified: August 29, available at < http://cas.et.tudelft.nl/~glas/ssc/techn/ tech niques.html >

52

“SMS Shell”

Rima Daoud ECE Dept, AUB Beirut, Lebanon

[email protected]

Jean Ghanem ECE Dept, AUB Beirut, Lebanon

[email protected] m

Nayla Hamadeh ECE Dept, AUB Beirut, Lebanon

[email protected]

Abstract Short Message service (SMS) is no longer a simple chat tool between friends; it can be used for important as well as operational tasks. “SMS Shell” takes advantage of the simplicity of this service to provide the user with a well-built functionality that allows him/her to access a computer, perform operations and get the results back in form of SMS messages. Such operations include checking/s ending mail, sending messages to multiple users by the use of one command (SMSBox) as well as executing system functions on both Windows and Linux operating systems.

INTRODUCTION

The proposed project is to implement an “SMS Shell” ; an integrated firmware that communicates with your computer’s operating system using Short Message Service. The end product will enable you to invest in the advantages of SMS without the need for Internet connectivity using a PC, laptop or any hand -held device other than your mob ile phone. It allows multiple simultaneous users due to its multithreading feature. Our program works on both Windows and Linux operating systems.

PROJECT DESCRIPTION Components Used • SIM card – any card that can receive and send SMS

will do the job.

• Cellular phone – Any cell phone with possibility of sending SMS messages will do the job.

• GSM modem – Ericsson GM 12 – it is a mobile phone for the GSM 900 MHz network. It is a class 4 mobile station and has the features of a regular cellular phone: it handles voice conversations and message service. The GM12 contains a SIM card through which all the communication takes place. In our project, we will only use one feature of the GSM module: the Short Message Service (SMS) since we are only interested in sending and receiving SMS messages.

Software Used

The software languages employed in the implementation of the programming algorithms are the following:

• AT commands for communication with the module.

• C for the Linux Mandrake 8.0 Operating System.

• Visual Basic 6 from the Visual Studio package for the Windows Operating System.

Starting the program

Upon running the program, whether on Windows or Linux, the first thing to be done is to complete the configuration, i.e. managing the users and commands file. Our program supports multiple users with different operation levels (administrator and regular users). On the other hand, the administrator can specify the commands that each level of users can execute as well as the maximum number of messages that each command can send as a response to the user’s enquiry and that by managing the command file. All of this is done through a friendly user-interface in both Linux and Windows.

The only difference between the two operating systems is that in Windows, the user and command files are stored in a database format using MDB files whereas in Linux, both are stored in regular text files with the passwords encrypted.

Establishing a Session

The session starts when the user sends an SMS message containing the username and password. The Computer gets the message through the GM 12 modem and checks if the user is valid by comparing the received username and passwords to the ones existing in the users database. If there is a match an authentication message is sent back to the user indicating that further messages containing commands can follow. If no match was found, an error message is sent back. This operation is illustrated in Figure 1.

Figure 1. Authentication Process

Once the user is authenticated, there will be a series of messages exchanged between the user and the program.

Each thread has an “inbox” related to it; each new message from an authenticated user will be placed in it, and the thread can later retrieve that message and perform

53

the required action. The response is put by the thread into the main thread’s outbox to be sent later on by the program. Figure 2 shows the above-described architecture.

Figure 2. Program Architecture

Ending a Session

It is important to point out at the beginning that our program has a multi-user feature; however, ending the session for one user doesn’t mean ending it for all users. In fact, each user has his own working place where he can execute commands and perform his own tasks.

Ending the session can be using one of the following two methods:

• Explicit command sent by the user through an SMS message.

• When the timer set for the user expires: each user has his own timer that gets set after sending the response to a particular command. If no messages are received form the users within a given time, then the space corresponding to that user is erased and the only way for that user to continue performing tasks is by re-establishing a session.

Performing Tasks within a Session

The user has the possibility to perf orm a lot of tasks via SMS messages. The exchange of SMS messages is possible as long as the user asks for programs to be executed and commands to be run.

Once the authentication is complete, the user can take advantage of services and tools offered by our product “SMS Shell”.

In what follows we describe some of the features of our end product.Manage Users

The “SMS Shell” user, if he/she has administration authority, can manage users via SMS messages.

This is similar to the configuration process used when running the program for the first time.

Using SMS messages, the administrator can add a user by sending the corresponding username, password and level (i.e. a regular user or administrator state). Deleting a user

is also possible by a specific command indicating the user to be removed. Modification in the users file is also possible: the administrator has the authority to change both the passwords and the level of capabilities. On the other hand, a simple user can only use the “change password” option of this feature (his own password of course).

Email Utilities

The user, regardless of his level, can use this feature to check his emails, read them and send new ones. The only requirement is to have a Microsoft Outlook installed on Windows with several accounts (corresponding to the different users present in the users database) already setup. The program, when receiving a SMS message containing the check mail command, will communicate with Microsoft Outlook, and return the new messages in the format shown in Figure 3.

Figure 3. Check Mail reply

At this point the user can send a read email command specifying the number of the email he wishes to read, and he will be sent up to a definable number of SMS messages. The user can also send an SM S indicating that he would like to send an email and send several SMS messages with the content of mail to be sent.

SMS Message Box A user can send a message to multiple users by the means of only 1 SMS message. This is done through the SMS Message Box feature. The lists are already defined in a database and each user has a set of predefined lists (and can be managed by the administrator directly on the PC), so that the user, when wanting to send a collective SMS, will only indicate the name of the group and the text to be sent. This feature also enables the user to manage his own lists (define a new list, add a number to the list, delete a number form the list, delete an entire list).

Shell Utilities

Maybe the most important application of our program is the shell utility. The user can establish a session similar to telnet with the system where he can open a shell and execute commands as if he was in front of the computer working on the command prompt of Windows or on the shell of Unix with the only difference being that the output is formatted and squeezed so that it can be displayed on the tiny cellular screen: a special filter has been constructed to accomplish that task. An example of such application is when the administrator can control a device connect ed with the PC through an already existing application where he can invoke the wanted device from

54

his cellular phone through the shell utility. Another application is to administer the network users or to check for the network status. The advantage of such a utility is that it can provide a backbone facility to let developers create their own applications, ones they wish to launch while they are on the move.

The feature described above with all its capabilities is only restricted to the administrators who are authenticated with respect to their telephone number due to the highly dangerous applications, some of which even harmful to the system, that can be executed through this shell. However, a similar application is available for the regular users where they can execute a predefined set of commands available in the command table or the command list on both operating systems.

ACKNOWLEDGMENTS

We would like to extend our deepest appreciation to our supervisor Dr. Ayman Kayssi, for his continuous help and support. We would also like to thank LibanCell for believing in us and being our sponsor by providing us with the necessary equipment.

REFERENCES

[1] GM12 GSM Module, Integrator Manual

55

MSIS: Mobile Student Information System

Hicham Fadel ECE department, American

University of Beirut. Hamra, Lebanon

[email protected]

Habib Haddad ECE department, American


[email protected]

Hady Moussa ECE department, American


[email protected]

Abstract In this paper, we describe the functionality of our project and some possible ways the American University of Beirut could use it for future upgrade of its services and better interaction among students.

INTRODUCTION

Msis stands for Mobile Stud ent Information System, a subset of AUBsis. The purpose and aim of this project is to allow AUB students register their courses through cellular phones or regular telephone lines, and getting alerts and notices of courses updates, grades, short messages… AUB labs are becoming very crowded during the registration period; AUBsis is facing many breakdowns due to excessive web uses… Msis presents a practical solution for this problem and a very efficient value-added service for the American University of Beirut . The project will be based on CTI, Computer and Telephony Integration, for the registration and “Drop and Add” parts, which will allow the communication between AUB users and AUBsis database, and on the GSM modem (GM 12) for the alerts and notices parts. Msis is sponsored by LIBANCELL and supervised by Dr. Ayman Kayssi.

REGISTRATION PROCEDURE

Registering courses will be easy and convenient using the CRN provided by AUBsis for each course.

§ You can register by telephone only with a touchtone telephone.

§ A Personal Identification Number (PIN) is a unique number that allows you to electronically register for classes and access your academic records through Msis.

When you first start, your PIN is set to your birth month and day. You may use your birth month and day or you may change your PIN to any four-digit number you prefer. By creating a unique PIN, you will provide increased security for your student records. It is important that you do not forget your PIN since you will be

required to enter it any time you use Telephone Registration.

§ STEP 1: Call AUB number, and the extension

for Msis

§ STEP 2: The computer will ask you to enter the year and term for which you want to register. You can only register for one term per call.

§ STEP 3: The computer will ask you to enter your Personal Identification Number (PIN) followed by the pound (#) key - Your PERSONAL IDENTIFICATION NUMBER (PIN) is your birth month and birthday, unless you have changed it.

§ STEP 4: Listen for the spelling of your name:

Enter 1 if correct enter 2 if incorrect § STEP 5: The computer will ask you to enter

your Current Enrollment Status.

§ STEP 6: You are now ready to register. The computer will ask you to enter one of the follow ing:

1 to register for courses or 2 to drop courses

Registering/Adding

If you entered 1 to register for courses, enter the CRN unit numbers for the courses you want. Continue to add courses by entering the CRN unit number for each course. Press 9 when you have completed entering your courses.

The computer will repeat all the courses for which you are registered and tell you if any of your selections were not accepted due to schedule conflicts, closed classes, etc. Listen for instructions to enter alternate selections if necessary.

Dropping If you entered 2 to drop classes, listen to the computer’s instructions. It will call out each of the courses in which you are registered and ask you which one(s) you want to drop. Press 9 when you have completed dropping courses.

56

§ STEP 7: Confirmation of courses. The computer will ask you if you want your courses read back.

§ STEP 8: When the registration process is complete, the computer will give you final instructions and say “good-bye”. Do not hang up until the computer has completed step 8 and said “goodbye.”

SMSBOX

An additional feature of our project is the SmsBox. It consists of specialized boxes for each course where users are divided into two categories:

• Students: they have the lowest priority and are only allowed to ask questions, chat in forums, and some additional simple tasks.

• Professors: they have a higher priority (administrator) and are allowed to send announcements, quiz dates, homeworks, grades, etc…

This feature is in some way similar to the WebCT service that is implemented on the Web at AUB, and our objective will be to make it available via mobile.

As its name implies, SmsBox relies heavily on the Short Message Service (SMS) capability to send short text messages relevant to the various functions of students and professors mentioned above.

To achieve this type of service, we basically need a GSM Modem (GM 12), and a database associated with each course:

Ø The GSM modem will handle the SMS

messages and will be responsible for sending and receiving such messages from/to the various databases.

Ø The database of each course will be used to store

the necessary information about each user (in this case, every student registered in this course), and also some information related to the messages.

o The user information will be made of:

§ User ID

§ User Name

§ Status (ON-OFF) § Last message retrieved

§ GSM number

The Status field is used to provide some flexibility to the users: a user (student) is able to decide whether to block the messages from peer entities (other st udents) or to receive them. (Blocking is equivalent to setting his status to OFF).

However, a normal user (student) is not allowed to block messages whose source is a higher priority user (in this case, a professor).

o The message information has the

following associated fields: § Message ID

§ Data

§ User ID This information is used to match the messages with the users in order to deliver these messages correctly to their intended destination.

We may say that the Msis service offers many exciting features that will help both professors and students in their quest toward efficient learning.

57

Above and Beyond the Wires W.A.M

Ayman Chalhoub Dept. of Electrical and Computer

Engineering American University of Beirut

[email protected]

Mark Ghibril Dept. of Electrical and Computer


[email protected]

Wissam Haddad Dept. of Electrical and Computer


[email protected]

Abstract Because wireless LAN technology is still new and dynamic, in addition to our enthusiasm to investigate and explore the fascinating world of wireless, we decided our project to be a full exposure of the wireless technology: a complete design study of a wireless network in the AUB campus with enhanced security, in addition to a full scale business plan to reveal its commercial viability.

In addition, using inexpensive off-the-shelf components and free software, we will detect when an unauthorized 802.11 card or access point was powered up and began broadcasting within range of a local WLAN. Furthermore, the event could be tracked, activity monitored, and the offending card or access point physically located.

We will also be implementing a wireless LAN in the Engineering Building as a demonstrating prototype. This prototype will be used to demonstrate the effectiveness of the wireless network and the power of our security, detection and tracking software. INTRODUCTION

Today, technology plays a central role in sparking the imagination, facilitating learning and creating new possibilities in education and research environments. In particular, networking technology can deliver to colleges and universities a wide range of vital broadband capabilities, such as E-learning, IP telephony, and affordable, high-bandwidth Internet access throughout the campus.

Colleges and universities face enormous challenges developing and maintaining infrastructures that keep pace with the demands of today’s high-tech society. Besides supporting administrative and faculty requirements, educational institutions must have the appropriate technology available to attract the best students and to prepare them to work, live, play, and learn effectively in the Internet economy. [3]

In the last decade, the Wired High-Speed LAN was the principal means of delivering needed broadband applications for univers ities and colleges. Special features were, and are still, deployed in order to achieve

maximum performance on the LAN, such as Robust Quality of Service (QoS), Continual Network Availability, and Security.

Today’s emerging technology, Wireless LAN, WLAN, affordably extends your network’s flexibility when it is not practical to install cable or when flexible, mobile access solutions are required.

Wired and wireless technologies each have a place in the campus LAN and collectively deliver LAN-to-LAN capabilities.

STUDY OF UNIVERSITY CAMPUS

The first part of our project will consist of examining microwave signals in the entire campus in order to determine the architecture and scheme of the network that will guarantee coverage i.e. Optimum placement of wireless access points to provide full coverage of the designated location without causing interference among the indi vidual access points.

After having determined the architecture, the plan and the design specifications of the wireless network that is most suitable for the university campus, we will proceed with the testing and implementation of the system.

IMPLEM ENTATION OF THE NETWORK

The implementation will be on a part of the campus (Engineering building) that will be a prototype to test the efficacy of the wireless network. This implementation will consist of testing the equipment, adjusting the locations of antennas, and adjusting other configuration parameters.

AUTHENTICATION SOFTWARE In parallel, we will be working on an authentication software, which will supply the network a boost in its wireless security. In other words, this software will give authenticated users accessibility to the university’s network. This open source software will be a very huge leap in the economic face of this project, since it will cost the university a number close to zero do llars. (Figure 1) [2]

58

Figure 18 DETECTION AND TRACKING S OFTWARE

The first goal is detection. This can be done with off-the-shelf components and free software. The monitoring station will work in a monitor mode, where it permits promiscuous monitoring of raw 802.11 frames to detect if thee are any telltale frames broadcast by a rogue card or access point.

The second goal is tracking, where a number of data points will be collected from the detected rogue device and the signal strength to distance curve is drawn. Then, a curve fitting equation is deduced, having the coordinates of the device as variables. Finally, the solution of this equation will realize the position of this rogue device. (Figure 2) [1]

Figure 19

BUSINESS PLAN

Finally, a complete business plan will be conducted to provide the university, with a detailed study on the importance, commercial viability, and technicality of such project. This plan will have full coverage of all the technical, economical and financial aspects of the project.

CONCLUSION

Today, the WLAN has redefined what it means to be connected. It has stretched the boundaries of the local-area network. It makes an infrastructure as dynamic as it needs to be. With this project, we provided the university with the complete study that reveals the power of Wireless with open source authentication software that enhances the security aspect of WLAN while decreasing the total cost of ownership. Moreover, a detection and tracking system that the business study reflects the practicality of such a project.

ACKNOWLEDGMENTS

We would like to express appreciation to several individuals and groups who made this project possible. Particular thanks, of course, go to our supervisor Professor Ali El Hajj for his assistance and support through out the project.

We would like to explicitly acknowledge the wonderful support and enthusiastic cooperation which we received from our co-sponsoring organizations: Cisco Systems, and Data Consult.

In particular, we would like to thank Mr. George Gha rios, Mr. Elie Howayek, Mrs. Lara M alek, Mr. Mohammed Al-Hashimi, and Mr. Osama Shwaihat for their technical assistance in the project.

REFERENCES

[2] A Practical Approach to Identifying and Tracking Unauthorized 802.11 Cards and Access Points <www.interlinknetworks.com>

[3] NoCat Authentication <http://www.nocat.org>

[4] Writing the Classroom Rules: An End-to-End Network for Delivering Broadband in Higher Education Environments

<http://www.cisco.com/warp/public/cc/pd/si/casi/ca3550/prodlit/hedun_dg.htm>

59

GIS Based Computerized Traveling Guide

Chucri Abi Haidar American University of Beirut

03-224805; Box 1777 [email protected]

Mossine Koulakezian American University of Beirut


Fadi Abou Ghantous American University of Beirut


Abstract GIS (Geographical Information Systems) is a very useful tool that is widely used. It has the ability to link spatial information to alphanumeric information, thus developing geographically referenced database-in both spatial and tabular formats .

We will use this ability to create a Computerized Routing Guide. This program is built using Arc View 3.1, the extension Network Analyst 3.0, and Avenue. In this report we will try to describe the functionality of this guide through describing its parts and options.

INTRODUCTION

In our FYP, we were to create, design and program, a “Computerized Travel and Touristic Guide”. Our main objective is to create a friendly interface that guides the user in various traveling, touristic, and routing applications.

These applications are:

• Shortest route in Beirut and Tripoli (Detailed) • Fastest route in Beirut and Tripoli

• Closest facility in Beirut and Tripoli (Detailed)

• Shortest route in Lebanon

• Buffering applications • Inter Kada’a connections

• Visualization of touris tic/historic locations

• Directions of all the above routing applications

• Printing the results (maps and directions) of the above applications.

As shown above, our software contains several applications that are very essential to help the tourists, as well as the local residents, in their different activities in Lebanon.

EXPLANATION OF APPLICATIONS As shown above, our software contains several applications that are very essential to help the tourists, as well as the local residents, in their different activities in Lebanon.

Before starting, we had to find the data that we can use in our project. We gathered them from Khatib& Alami, ACT, and Mr. Hani Naghi. Due to their inconsistency, we had to convert all the shapefiles/data to one projection system (Lambert), and finally digitize some parts ourselves in order to update their attributes.

Shortest Route Shortest Route analysis gives the shortest route to be taken form a point to other point(s) chosen by the user, based on the length of the streets. These points could be chosen anywhere; they could be hotels, restaurants, nightclubs, hospitals, universities, etc. This application has the ability to give the user the shortest route of the chosen locations by the order entered or by the best order to cover all the given points. Shortest Route is now functional for Beirut and Tripoli, in a very detailed format, and for Lebanon with less detail, due to lack of data. Fastest Route

Fastest Route analysis is similar to the shortest route but the base data for our route is the traffic data of the streets and not their length. Consequently routing time is our concern in this part.

Closest Facility

Closest Facility analysis gives the user the ability to locate the closest facility (Hotels, restaurants, etc.) from his current location. For example, if the user is at Pt#1 and wants to locate one or more banks, Closest Facility will show him the closest in ascending order with respect to their distance from Pt#1 (see figure 1). The user can choose any number of banks to see the corresponding maps.

Figure 1: A sample output sorted from the shortest to the longest route for each location point

60

Buffer Analysis

Buffer Analysis gives the user the ability to choose a buffer radius from his location to find the available landmark s that lie within his buffer.

Inter Kadda’a Connection

Inter Kada’a Connection gives the user the ability to go from one Kada’a to another using the inter city highways and major roads.

Block Roads

In all of the above applications, the user is always able to prevent his route from passing certain roads. This is by using the option Block Roads that eliminates the possibility of passing through selected areas or streets and therefore doesn’t include them in the routing applications. Further more; the user can always print this map, and get the directions for it .

Figure 2 is an example of the shortest route by order and a part of the corresponding directions. The green points are the points that the user chose and the yellow roads are the blocked roads.

Figure 2: A sample Route

Directions: Starting from Graphic pick 1 Travel on for 33.01 m Turn right onto Jeanne Darc Travel on Jeanne Darc for 19.97 m Turn left onto Emile edde Travel on Emile edde for 756.17 m Continue straight onto Rue Dunant

Travel on Rue Dunant for 17.79 m Continue straight onto Emile edde Travel on Emile edde for 169.01 m Continue straight onto Spears Travel on Spears for 64.94 m Turn right onto Madhat Pacha Travel on Madhat Pacha for 49.06 m Turn left onto Salim Boustany Travel on Salim Boustany for 80.94 m Turn right onto Bouhtouri

Travel on Bouhtouri for 243.42 m …

Figure 3: Corresponding Directions for the Sample Route

Visualization of touristic/historic locations

Visualization of touristic/historic locations gives the user the ability to see pictures of the location he chooses. It helps the user see different locations, before deciding his trip destination(s).

INTERFACING

We are trying as much as we can to make the dialogs and the pop -up windows user friendly in a way that any person who is non-familiar with Arc View and Avenue to use the software with ease. We are also following universal conventions in naming our dialogs, scripts, and variables used.

ACKNOWLEDGMENTS

At the end we would like to acknowledge Dr. Farid Chaaban for his wisdom and experience in his supervision for our project. His help and guidance were crucial for the success and development of our final year project.

Also we would like to credit Mr. Hani Naghi for his help in providing data, hospitality and constant advice. In addition, we would like to thank Khatib & Alami and ACT for supplying us with the data needed to complete our project.

REFERENCES [1] <http://www.esri.com>

[2] ESRI, Avenue, 1st ed., Environmental Systems Research Institute, Inc., United States of America (1996).

61

Common Sense Controller (CSC) using LabVIEW

Sima Azar

Dept. of Electrical and Computer Engineering American University of Beirut


Sandra Dandach Dept. of Electrical and Computer Engineering


[email protected]

Abstract This paper discusses the design of a universal controller based on fuzzy logic, and implemented in National Instrument’s LabVIEW. Design of conventional controllers often requires heavy computational modeling of the process, which can be very difficult, and sometimes unnecessary, for complex processes. The goal of Common Sense Controller (CSC) is to drive an arbitrary plant to a desired reference output without knowledge of its transfer function, thus relieving the operator of the need for expert control knowledge.

Keywords Fuzzy, Process Control, LabVIEW, Controller Design.

INTRODUCTION

The typical application in control is to drive a process to a desired reference output. Most often, systems we encounter in real life are too complex to be mathematically modeled. Designing a controller for such systems can therefore be a difficult task. It also requires that the user have expert knowledge in co ntrol and on the process to be controlled, which is not always the case. Fuzzy control methods can be very useful when such uncertainties exist. Also, rather than developing a specific controller for each process, certain techniques can be used to create a universal controller based on intuitive knowledge. Such a controller will be oblivious to the process it is controlling, but will learn from error measurements about its progress, which is the task of Common Sense Controller, or CSC. Many configurations of fuzzy controllers exist depending on the application. Some work to assist other controllers, such as in a Fuzzy-Assist PID controller. Our controller will work to directly drive the process.

COMMON SENSE CONTROLLER DESIGN

In designing our controller, we treat each plant as a black box, meaning we do not have knowledge of its transfer function, nor do we seek to derive its transfer function. We rely only on learning from the online error measurements to determine the progress of our controller as it attempts to drive the process to the desired reference. Since, we assume that the operator does not have expert control knowledge, the only information we require is:

• Desired reference value.

• Plant input minimum/maximum ranges.

The plant input minimum/maximum ranges are needed for calculation of an initial condition, as well as for safety precautions, so that the controller does not try to drive the process with an input it cannot handle. All of the controller’s decisions are taken based upon two inputs: the error and the change in error.

Figure 20 – Block diagram of CSC driving a process

The output of the controller is a corrective measure to the previously delivered control signal. It is important to mention that we are building the fuzzy controller offline, and so CSC’s parameters are predefined, and remain fixed. The parameters are not designed to change online as possibly in an adaptive fuzzy controller. We used LabVIEW’s built in fuzzy logic toolkit that provides an interface for designing fuzzy controllers. Algorithm

Fuzzy control processing occurs in three stages: Fuzzification, Inference, and Defuzzification. In building CSC, extra pre-processing and post-processing is needed. As mentioned earlier, the controller performs corrective actions to the control signal according to the error, and the rate of change in the error. We explain the steps of the algorithm below: 1. Calculate initial condition: IC

For our algorithm to begin, we supply a small initial condition to the process. This IC is taken roughly as a small percentage of the process input range.

2. Calculate normalized error: e(k)

Fuzzy controllers designed for a specific system do not necessarily require a normalized set of error values. The normalization step is very crucial to making our controller universal. Since we cannot fix our controller param eters to “real” error values that may pertain to only

CSC Plant

Reference

+ -

Output

z-1

+ -

62

one or a few systems, we chose the error range to be (-1, +1).

The error is then calculated as follows:

e(k) = [r – o(k)] / r

This error tells us how far away we are, above or below, the reference value.

There is a special case when the reference is equal to zero (regulation). In this case, the error is simply:

e(k) = – o(k)

The error range remains (-1, +1), because values beyond -1 negatively are defaulted to -1, and those beyond +1 positively are defaulted to +1. In each case, the controller must make a very strict change to the control signal.

3. Calculate change in error: de(k)

Knowing the rate of change in the error provides vital information to our controller, by telling it how fast the error is growing or shrinking. Here, we also chose to restrict the range of de(k) to reside between –1 and +1. Calculation of the change in error is done as follows:

de(k) = e(k) – e(k-1)

Combined with the error, the controller can decide what changes to make to the control signal.

4. Fuzzification of controller inputs, e(k) and de(k)

Designing membership functions may take several trials. It is best to begin with basic membership distributions such as symmetrical triangular ones, and to later tune them. We discuss the tuning of our membership functions at a later stage, but we will now work with the preliminary distribution to continue the discussion of our algorithm. To start with, we chose 7 membership functions for each e(k) and de(k): Negative Large (NL), Negative Medium (NM), Negative Small (NS), Zero (ZE), Positive Small (PS), Positive Medium (PM), and Positive Large (PL).

Figure 2 - Membership distribution of error e(k) and

change in error de(k) are similar

We initially allowed full overlap of the membership functions for e(k) and de(k). Some overlap is necessary to make sure that all states are well defined. The terms derived from the fuzzification process will be used in the next phase, inference.

5. Inference of e(k) and de(k)

There are many different configurations for a rule-base consisting of the error and the change in error. Some rule-bases emulate PD or PI controllers [1]. We built the rule-

base of our controller on intuitive knowledge, taking into account that the output of our controller is an increment or a decrement to the previous control signal we supplied. This is unlike usual controllers, whose outputs are either a direct control signal, or an additive/subtractive factor to the previous control signal.

Table 1 - Initial CSC rule-base with seven membership functions

NL NM NS ZE PS PM PL

NL NL NL NL NL NM NS ZE

NM NL NL NL NM NS ZE PS

NS NL NL NM NS ZE PS PM

ZE NL NM NS ZE PS PM PL

PS NM NS ZE PS PM PL PL

PM NS ZE PS PM PL PL PL

PL ZE PS PM PL PL PL PL

This set of rules steadily drove the plant to the desired set point, but caused some oscillations around the reference. We later tuned our controller to be less sensitive around the zero error region, for an improved response.

Figure 3 – Control surface of initial CSC

The control surface provides a valuable analysis tool for our controller. A reasonable controller will have a smooth control surface, especially around the zero error region.

6. Defuzzification of e(k) and de(k) lead to x(k)

As we previously mentioned, the output of our controller represents an increment/decrement change to the previous control signal. The output membership functions of CSC are also initially distributed symmetrical. They are similar to the input membership functions, but the universe of discourse here was chosen to be (-0.5, +0.5). This means that we allow a ± 50% correction to the previous process input.

Figure 4 - Membership distribution of output x(k)

de e

63

Using the full range of (-1, +1) will not work with our algorithm, because in case the correction factor turns out to be –1, then all future process inputs will be zeroed. We found that the range (-0.5, +0.5) was more than sufficient to drive the process to its reference.

7. Calculate new control signal: i(k)

Based on the above discussion, the control signal is calculated to be:

i(k) = [1 + x(k)] i(k-1)

Where, x(k) is the current correction to the previous input i(k-1), and i(k) is the current corrected control signal. This is a very significant difference from conventionally generated control signals. A typical control signal would look like:

i(k) = i(k-1) + u(k)

Here, u(k) directly increases or decreases the control signal. When some knowledge of the system is available, it is possible to use such a control signal. The designer can then know where to define u(k), and what universe to assign it. If we are to make our controller universal, then it is impossible to define u(k)’s universe in such a manner. CSC is completely oblivious to the system it is controlling at any given time, and so defining a fixed range for this controller output would severely limit its applications, not to mention its ability to accommodate different systems.

Therefore, the solution is to use a correction factor x(k), to build on previous control signals, since that is the only information available to us. In this sense, our controller can learn online from the response of the plant to each control signal it feeds it.

Notice the need for the initial condition mentioned earlier. Our algorithm cannot begin with i(0) = 0, as this will cause all future inputs to be zero, and will not allow our controller to function as it should.

8. Generate new process output: o(k)

Once our controller has computed the present control signal, this is input into the process and the process is allowed some time to respond. We require from the system for it to be bounded -input, bounded-output (BIBO) stable. This is necessary because we need to witness the steady -state effect of the control signal on the process output, before taking any further action.

9. Repeat steps 2-9 to reach reference

Steps 2 through 9 of the algorithm are repeated until the error is satisfactorily small, meaning the desired reference value has been attained. At this point, the controller output is zero, and the control signal computed reaches a constant value. Naturally, due to discrepancies and system dynamics, the correction factor may not necessarily reach zero, but rather a very small fluctuating number. This produces some oscillat ions in the control signal, and henceforth in the output as well. The

simulation results on the process tested show that these oscillations are only occasionally encountered, and when they are present, they are very minimal and do not distort the output reference much.

CSC Tuning

We performed a great deal of tuning to the initial CSC design. We found that 7 membership functions were not sufficient to provide us with satisfactory results, and so we extended the membership functions of the error and the output as follows: NL, NM, NS, ZN, ZE, ZP, PS, PM, PL. The new additions, Zero Negative (ZN), and Zero Positive (ZP), demonstrated excellent results around the sensitive area of zero error. Adding them to the output membership functions allowed for finer tuning of the control signal. The membership functions for the change in error remained the same. The new membership functions for the error and the controller output are shown in the figures below:

Figure 5 - Tuned membership distribution for error

e(k)

Figure 6 - Tuned membership distribution for output

x(k) Naturally, the rule-base has also been changed to suit the new membership functions added, as well as the new distribution. The tuned rule-base is shown in the following table.

Table 2 - Tuned rule-base of CSC

NL NM NS ZE PS PM PL

NL NL NL NL NL NM NS ZE

NM NL NL NL NM NS ZE PS

NS NL NL NM NS ZE PS PM

ZE NL NM NS ZE PS PM PL

PS NM NS ZE PS PM PL PL

PM NS ZE PS PM PL PL PL

PL ZE PS PM PL PL PL PL

de e

64

Figure 7 - Control surface of tuned CSC

The resulting control surface after tuning is shown to be different from the initial one. The differences are to properly compensate for large errors, and to be much more sensitive around smaller errors.

SIMULATION

To simulate the performance of CSC, we tested on several known systems. We illustrate the results of only one typical system here. Of course, the transfer functions of the systems are used only for simulation, and do not in any way affect the controller decisions.

This example is of a real DC Motor whose transfer function is: G(s) = 672 / s (s+6.07)

This is a 2nd order type 1 system. For a reference of 300 rpm, the response is shown in the next figure.

Figure 8 - DC Motor response (reference 300 rpm)

Notice that there is no overshoot, and very little oscillations at the steady -state region.

HARWARE EXPERIMENTATION

Not only did CSC provide significant results during software simulation, but also in a practical implementation on a real servo motor. We used a Data Acquisition Card (DAQ) connected to the CE2000 Servo Trainer, whose input ranges are ±10V. With a set point of 500 rpm, CSC was able to drive the motor steadily and suitably to the reference. Due to some discrepancies in the speed sensor, as well as system dynamics, we

witnessed some oscillations around the reference. This can be further improved with more tuning later on.

Figure 9 - Servo Trainer response (reference 500 rpm)

Of course, in testing on the Servo Trainer, its transfer function is not known to us. For further testing, we changed the load on the motor while it was running, which changes the systems transfer function. CSC was able to compensate for this added load by increasing the control signal. The reference was therefore maintained, despite some oscillations.

CONCLUSIONS

Designing a universal controller is a novel idea, and may at first seem infeasible. However, with the original techniques we used in our design, we were able to reach our aim with more than satisfactory results. This controller was originally developed for SISO LTI systems, but it can further be expanded to accommodate MIMO, Linear Time Variant, and Non -linear systems.

ACKNOWLEDGMENTS

We would like to sincerely thank our supervisor Dr. Fouad Mrad for his valuable support and direction during the development of this project. Thank you to Mr. Ghassan Deeb for also providing us with his experience and knowledge.

REFERENCES

[1] F. Mrad, G. Deeb, “Experimental Comparative Analysis of Adaptive Fuzzy Controllers”, IEEE Transactions on Control Systems Technology, 2002.

[2] K. Ogata, “Modern Control Engineering”, Prentice Hall Inc., New Jersey, 1997.

[3] National Instruments Corp., “Fuzzy Logic for G Toolkit Reference Manual”, March 1997 Edition.

65

Linearization of a D/A Converter

Azarian, Michel M. Dept. of Electrical and Computer

Engineering Beirut, Lebanon

[email protected]

Hani, Alain M. Dept. of Electrical and Computer

Engineering Cornet Chehwan, Metn, Lebanon

[email protected]

Saad, Roland N. Dept. of Electrical and Computer

Engineering Awkar, Metn, Lebanon [email protected]

OBJECTIVE

This research paper addresses non-linearity errors, which occur in a classical digital to analog (D/A) converter. This class of converters uses weighted current sources that are controlled via the bits of the digital word to generate its equivalent analog signal. Due to non-ideal behavior of these current sources, the outputs deviate from their ideal values. The objective of this paper is to discuss software methods that will reduce the non-linearity errors and thus improve the precision of the D/A converter. However, these met hods are mainly effective in the relatively low frequency range.

BRIEF DESCRIPTION

Although there are hardware methods capable of correct ing the outputs of the current sources, these methods are usually complicated and expensive. For this reason, this paper focuses on software implementations that can be easily incorporated into existing software packages.

In order to verify the theory, these software techniques were applied on the D/A converter internal to National Instrument’s data acquisition board NI5411, using LabVIEW as the programming environment. The main idea is to generate a pre-distorted signal that when fed into the D/A converter would show a significant improvement in the output by decreasing its noise content around the working frequency range. Two main stages are required in generating the pre-distorted signals. The first phase is to simulate the D/A

converter. To achieve this, more than one procedure can be followed. The objective is to acquire er ror values introduced by the converter for later use in the simulated model.

The second phase is the addition of an error correction loop that receives the output of the D/A converter model and passes it through a noise shaping structure. The result is the pre-distorted signal sent to the board.

PRELIMINARY RESULTS

All the above explanations were transformed into action. The errors of the D/A converter were found using three different schemes:

• Direct method where the error values were found while varying the converter’s output in steady steps

• Random method where the error values were found while varying the converter’s output in a random order

• Sinusoidal method where the error values were found while outputting a sinusoidal wave

So far, the best improvements were obtained while using the error values found using the random method. As an example, a sine wave at a frequency of 920 Hz sampled at a rate of 100K was output by the board. The result was read by another data acquisition board: NI5911, which is basically an A/D converter. Figure 1 shows the output of the D/A converter with no correction, while Figure 2 shows the same signal after being passed through the error correction loop.

Figure 1. The output of the D/A converter with no correction

66

Figure 2. The output of the D/A converter with correction

Comparing both outputs appearing in the figures above shows the improvement added to the signal.

ACKNOWLEDGMENTS

We would like to express our sincere gratitude to Mr. Niels Knudsen of NI, Denmark for his daily assistance and to Dr. Kayssi for his constant supervision.

REFERENCES

[1] Knudsen, N. O., Moriat, A., “A New Method for Linearization of a Classical-Type D/A Converter” Presented at the 98th AES Convention 1995

[2] Norsoworthy, S. R., Schreier, R., Temes, G. C., “Delta-Sigma Data Converters Theory, Design, and Simulation,” IEEE Press 1997

67

The Design of a Prosthetic Glove for a Child with Cerebral Palsy

Tarek Omar El-Chidiac Mechanical Engineering, Class of 2002

Beirut, Lebanon Email: [email protected]

Jawad Talaat Sabbah Mechanical Engineering, Class of 2002


Abstract In this paper, we describe our Final Year Project, which discusses the design and manufacturing processes of a prosthetic glove, indented to be used by a child suffering from cerebral palsy and partial hand amputation.

INTRODUCTION

The history of prosthetics and amputation surgery began at the dawn of human medical advancement. Prosthetic history began with human kind's spiritual and functional need for wholeness. The range of prostheses and their uses is vast. A major boost for prosthetic advancement occurred in the 20th century in the form of wars. Nowadays, especially with the emergence computers, prosthetic development has achieved a new level. In the United States, statistics show that nearly 3,000 individuals become amputees each week. Worldwide, there are 10 million amputees. Amputations usually result from a disease, trauma or car accident, birth defects, illnesses and warfare.

CEREBRAL PALSY

C.P., or Cerebral Palsy, is a disorder affecting the motor skills, muscle movement, and muscle tone. “Cerebral” refers to the brain while “Palsy” is a synonym for paraly sis. Around 10,000 new cases of cerebral palsy are identified each year in the United States, and five out of 2,000 babies born in the U.S. have cerebral palsy. Approximately 50% of children with cerebral palsy are born early. Cerebral palsy is not progressive; that is, it does not get worse. It is also not contagious.

ABDULLAH KURMALLY

The name of the amputee we are building the prosthesis for is Abdullah Kurmally. He is a 7 year old Saudi who has been suffering from spastic cerebral palsy, a disability occurring at birth and resulting from damage to the brain which has resulted in a limb becoming affected significantly (Refer to section 1.3 for more information on CP). He is a regular patient at the “Ghassan Kanafani Cultural Foundation” in Beirut, Lebanon. His right hand is partially amputated and his left hand is the hand that he cannot use to carry out normal tasks. Abdullah cannot sit on a chair by himself, he has trouble moving his legs and arms without help, and has some trouble with his speech and emotions (all of those symptoms are those of cerebral palsy); to make things worse, the fingers of his right hand have been amputated, due to gangrene, leaving only the met acarpal bones free to move.

Figure 2. Abdullah Kurmally. CABLE CONTROLLED P ROSTHESES

In spite of the significant amount of research and development in externally powered prostheses, it has been estimated that up to 90 percent of the upper-limb amp utees who wear prostheses still wear the "conventional" body powered prostheses. Those are actuated either by body power or externally applied power. The VO and VC system hands belong to the cable controlled prostheses or active gripping arms and are controlled via cables. The motion is passed on to the prosthesis through a cable control system that usually runs from the prosthetic arm across the back to a loop around the healthy shoulder. This makes it possible to specifically open the prosthetic hand or the hook by moving the healthy shoulder or both shoulders simultaneously.

Possible Design

The systems mentioned above are mainly designed for patients who suffer a complete arm or hand amputation. Since Abdullah has muscular control of the amputated right hand with no functionality of the left hand (due to Cerebral Palsy), we chose to adop t a special kind of design where fingers would flex and extend by the action of a harness cable. Unlike the VO and VC systems, where the shoulder controls the finger movement, artificial fingers would be adjoined to the amputated hand. Those will be controlled by the movement of the partial fingers. Our design has been based upon work supported by the Rehabilitation R&D Center at the Department of Veterans Affairs Healthcare System in Palo Alto, the Temesvary Fund of

Figure 1. Amputation of Abdullah’s Right Hand.

68

the Rehabilitation Engineering Center at Packard Children's Hospital at Stanford University, and the Stanford University Medical School Medical Scholars Program. Taking the Stanford design as a starting point, we modified it to suit Abdullah’s case. The endoskeleton would be composed of four separate active fi ngers, each capable of bending at the (MCP), (PIP), and (DIP) joints.

Figure 3. Endoskeletal plastic fingers offer bending MCP, PIP, and DIP joints, providing realistic finger movement.

The endoskeleton would also contain a passive thumb that could be positioned by the user to provide different grips, power and precision. The prosthesis is voluntary closing. The fingers remain extended until the harness cable, made of Spectra, is displaced. The harness cable actuation causes flexion of all four fingers against the stationary thumb. Upon release of the cable, the fingers extend, opening the grip. This voluntary -closing mechanism offers the benefit of graded apprehension, allowing the user to vary grip the force depending on harness cable force. This feature assists in grasping fragile objects.

Infeasibility of Such a D esign to a CP Patient

Although this design is an outstanding biomechanical prosthesis, it proved to be impractical or feasible from several aspects.

• Abdullah’s medical condition

• Manufacturing process

• Time constraint

• Budget

As a result of all those constraints, we had to resort to a more practical yet effective prosthesis that would be an assistive tool for Abdullah.

COSMETIC ARM PROSTHESES

Arm Prostheses are designed to restore the essential functions of the missing hand, such as opening and clos ing the hand. Arm prostheses mimic the outward appearance of the hand due to their natural form, therefore meeting the desires of p atients for whom the restoration of their outward appearance is of paramount importance. Generally, cosmetic hands consist of an inner hand and a cosmetic glove. The shape, color, and surface structure of the cosmetic glove reproduce the natural hand in detail. The cast foam inner hand ensures

high stability with low weight and thus increases wearing comfort. Thanks to various fastening possibilities, it is universally useable. Many manufacturing companies have different models and designs to satisfy a wide range of patients.

Significance of Abdullah’s Pros thesis

Since prosthesis can never fully replace a human hand, decisions have to be made about what functions should or should not be possible with a prosthetic device, as well as its appearance, durability, and other issues. The design of this prosthesis will act as an extension of the remaining part of the fingers. It will consist of an artificial hand that could be worn as a glove using a zipper. A zipper is needed in order to take the prosthesis out at night and clean it with special agents since the skin area surrounding the prosthesis will experience a lot of perspiration. We chose the material of the glove to be PVC, or Poly-Vinyl Chloride due to its economical price. The elasticity of the glove will be manipulated by the use of plasticiz ers. We chose the color of the PVC as white as possible to match Abdullah’s light skin. This design will allow Abdullah the elementary movements of the fingers by acting as an extension. The design of Abdullah’s prosthesis is based on three criteria:

• Comfort • Cosmetic Appeal • Control

The glove will help Abdullah in grasping a wide range of items with different dimensions, such as an apple, a toy, a spoon, even a pencil or a pen, etc. This design is just the right one for Abdullah due to the cer ebral palsy that he suffers from. It is impossible to train Abdullah to use a complicated prosthesis due to the difficulty in communication between us and him. He is not a normal patient. He has deficiency in learning and he will not always follow orders or listen to instructions because of the brain damage that cerebral palsy causes.

MANUFACTURING OF THE PROSTHESIS

As in any design and manufacturing process, there are trials and errors before reaching the appropriate prot otype. We have undergone several steps before presenting Abdullah with a functional and reliable prosthesis.

Casting

The first step in any prosthesis fabrication is the casting. An accurate cast of Abdullah’s right hand will provide us with invaluable information regarding the dentition, measurements and particularities of each of the partial fingers before proceeding to manufacturing. Below are the steps that were taken to cast Abdullah’s hand.

1. We first apply a lubricant (preferably Vaseline) on his hand up to the elbow in order to minimize friction and irritation.

2. In general, a plaster cast is designed to protect a broken bone and prevent movement of the

69

aligned bone ends until healing has progressed sufficiently. Casts are made from Plaster of Paris. When mixed with water, Plaster of Paris sets hard, forming interlocking crystals of gypsum.

3. We stabilized Abdullah’s wrist with a wood stick, in order to prevent him as much as possible from closing his fingers. We then started casting the hand by wrapping the wet Plaster of Paris around the wrist, palm, and between and over the fingers.

4. The first cast was not as successful as we hoped, because Abdullah had stiffened his muscles a lot and was not in a relaxed state. Therefore, we took another cast, this time for the fingers and palm only.

Figure 4. We have the Vaseline, water bucket, wood stick, scissors and the two casts we took.

5. These casts will complement each other, allowing the calculation of the exact measurements of the fingers and hand for developing the proper prosthesis.

6. As discussed earlier, the prosthesis will be made of PVC (Poly-Vinyl Chloride), which comes in different colors and shades. We chose from several PVC shadings and decided to use the one that best matches Abdullah’s skin color.

7. The amputated hand was drawn on an A4 sized paper using a pencil to have all the needed data for the construction of the prosthesis. Using a scaled rope, we took measurements of the wrist diameter, fingers and palm.

First Fitting Session After obtaining a prototype hand prosthesis based on the measurements and properties we di scussed in the previous section, we went to the “Ghassan Kanafani Center for the Handicapped” for a first fitting session. This session will determine whether our prosthesis meets the measurements we have and whether the length of Abdullah’s partial fi ngers can go through it, allowing for fine-tuning to be made later. Moreover, it will show the ease with which Abdullah puts the prosthesis. Finally, it will let us see how much he is comfortable with the extensions, and to what extent he seems to accept it psychologically.

Figure 5. First Prototype of Prosthesis

1. By directly comparing the prosthesis to Abdullah’s complete left hand, we focus on the dimension of the prosthesis. Its size should be close to his normal hand. We also examine the closeness of the PVC’s color to the child’s skin. As can be seen, the prosthesis’ shading is somewhat different. That is due to the PVC which, unlike silicon-fabricated prostheses, does not match the exact skin color. This comparison is important when a large difference between the prosthesis and the normal hand would be noticeable and awkward.

2. While fitting the prosthesis to Abdullah’s hand, his fingers were contracted, and that prevented the prosthesis from sliding and prohibited the partial fingers to assume their positions. This was due to two reasons: first, the spasticity related to Abdullah’s CP prevented him from relaxing his fingers; second, the child was uncomfortable with the prosthesis being applied to his hand for the first time.

3. After rubbing Abdullah’s hand and the prosthesis with Vaseline, in order to reduce friction, we managed to fit the glove into his hand. We had to make sure that the prosthesis slipped with ease; otherwise, immense pressure would be exerted on the hand’s veins and soft tissue, preventing blood from circulating prop erly. An examination of the depth of the partial fingers into the prosthesis had been conducted for two reasons. To start with, we should know where we ought to apply the foam as will be explained in the next section. Second, the amputated fingers will be the driving forces behind the flexion and expansion of the prosthesis. Thus, to have optimal performance, the fingers have to reach the maximum possible depth into the prosthesis without any discomfort to the child.

Second Fitting Session

Before scheduling a second fitting session, we managed to solve the problems we encountered in our first fitting session, mainly the difficulty of inserting the prosthesis into Abdullah’s hand. By introducing small plastic cones halfway through the prosthetic fingers, we were able to dilate them wide enough for Abdullah to slide in his fingers with relative ease. In addition to the

70

modification we made to the fingers, this enhancement is a reflection of the fact that Abdullah is starting to get used to and accept the prosthesis as a future-helping device in his daily activities.

We sewed a zipper along the prosthesis going from the wrist to the arm, offering a convenient way to apply and remove the prosthesis. The zipper provides firm stability once it is closed, with minimal irritation to the skin.

Figure 6, 7. The prosthesis with the Zipper Closed and Open.

A problem we realized during this fitting session is that Abdullah, still closes his fingers involuntarily even when he wares the prosthesis. This is due to the spasticity and involuntary movements caused by his CP condition, which makes the fingers slip out of their repective places inside the glove. To correct this obstru ction, we will attach a wrist fixation inside the glove, which will be applied over Abdullah’s palm. This plate will make difficult for him to close his fingers, making them stay inside the glove. Last Fitting Session

On our last fitting session we had the prosthesis ready to be used by Abdullah after the addition of silicone along with the wrist fixation. He seemed to accept it reasonably well and was able to grasp the objects that he could not hold on the first day we met him.

FUNCTIONALITY

A prosthesis should not be labeled as “functional” if it only enables the amputee to take hold of objects. That definition ignores the majority of tasks for which the prosthesis is actually utilized. Tasks, such as holding an object up or down, pushing or pulling an object closer, are by the prosthesis. If the role of our prosthesis in supporting, stabilizing, pushing, pulling, holding and facilitating balance in every day life situations is fulfilled, the prosthesis should be qualified as functional. The follow ing classes rate the functionality of Abdullah’s prosthetic glove based on crit eria adopted by the field of rehabilitation medicine normally performed.

• Active Function

• Emotional Function

• Social Function

• Economic Function

ACKNOWLEDGMENTS

We, Jawad Talaat Sabbah and Tarek Omar El Chidiac, would like to extend our gratitude and appreciation to the following individuals and institutions, whose names are listed below, for their co ntinuous help and support throughout the preparation of this report.

• Dr. Kinda Khalaf (FYP advisor)

• Dr. Samer Abdallah (FYP coordinator)

• Orthopedic Instruments Manufacturing for donating the needed materials and providing free consultation to achieve our design.

• The Kurmally Family

• Ms Nahla Ghandour (Director of the Kanafani

Center for the Handicapped) • Miss Lama Abi El -Mona (Occupational

Therapist at the Kanafani Center for the Handicapped)

REFERENCES

[1] North Western University, Prosthetic History; Available Online: http://www.nupoc.northwestern.edu/index.shtml

[2] Schools Online, Plaster of Paris; Available Online:

http://sol.ultralab.anglia.ac.uk/pages/schools_online/Contents.html

[3] Livingskin, Aesthetic Concerns Prosthetic Inc.; Available Online:http://www.livingskin.com/default.asp

[4] Rajiv Doshi, BS; Clement Yeh, BS; Maurice LeBlanc, CP, MSME; The design and development of a gloveless endoskeletal prosthetic hand; Journal of Rehabilitation Research and Development Vol. 35 No. 4, October 1998 Pages 388-395; Available. [Online]: http://www.vard.org/jour/98/35/4/doshi354.htm

[5] CEREBRAL PALSY PROGRAM; THE ALFRED I. DUPONT INSTITUTE http://gait.aidi.udel.edu/res695/homepage/pd_ortho/clinics/c_palsy/cpweb.htm

71

Development of a Computer- Based Measurement Station

Marianne Sarkis 4th year Electrical Engineering

student, AUB Beirut, LEBANON [email protected]

Michel Sarkis 4th year Electrical Engineering


Joseph Malkoun 4th year Electrical Engineering


Abstract In this paper, we give a description of our Final Year Project, which consists of developing a computer-based measurement station. The Modern Electronic Industry Corporation (MEICO) is sponsoring our project. The objective is to measure accurately the power factor and the total harmonic distortion of the electronic ballast designed by the company. The program that we are designing is needed because it can measure the requir ements that will enable the company to obtain the ISO certification for its products. In this paper, we are presenting the purpose of the project, the software (LabView) and equipment used, the theoretical analysis performed, the work achieved so far and the preliminary results that were attained.

INTRODUCTION

Our Final Year Project, as fourth year Electrical Engineering students at AUB, consists of developing a computer based measurement station. The Modern Electronics Industry Corporation (MEICO)-Lebanon sponsors our project. The main objective is to be able to measure accurately the power factor and the total harmonic distortion of the electronic ballast that the engineers at MEICO have designed. We are to plan this application such that it will accomplish the following tasks to their product:

ü Measur e the input and output voltage and current of the device.

ü Measure the power factor of the device. ü Measure the harmonics created by the

device: evaluate the total harmonic distortion by measuring the power in the harmonics and comparing it to the fundamental one.

OVERVIEW OF THE SOFTWARE AND EQUIPMENT USED In order to carry out the work on our project, MEICO provided us with the manuals and kits that are useful to our job. The following is a brief description of the equipment and software that we are employing:

1) The NI 5102 instruments are analog input devices that combine the benefits of digitizers and oscilloscopes. The manual that comes with this package explains the basic information we need to understand about making measurements with digitizers. It contains also

the theories that should be taken into account when measuring our data, i.e. Nyquist theorem, Analog Bandwidth, ADC resolution. In addition, it provides us with an overview of the hardware components we are using to get familiarized with the DAQ and the Scope cards.

2) The main software that coordinates our project

is LabView. It is a programming system different from the conventional text based languages. Its main feature is the ability to create programs in a graphical layout. The commands used in this language are of block logic format that enables the user to visualize the task he is performing. The sequence of the program is acquired by wiring the blocks used as in an ordinary circuit. LabView includes many libraries for data acquisition, data analysis, data presentation, and data storage.

THEORETICAL ANALYSIS

We started by performing a theoretical analysis to investigate how to measure the power factor and the total harmonic distortion of a device.

a) Measuring the power factor:

There are several definitions for the power factor and hence several methods to calculate it. However, we have considered the power factor as the ratio of the real power to the apparent power.

werApparentPoalPower

rPowerFactoRe

=

Measuring the total harmonic distortion:

Total Harmonic Distortion (THD) is the ratio of the RMS value of unwanted components to the fundamental. %THD is the percent total harmonic distortion present in the input auto power spectrum of the signal. The %THD computation is actually made according to the following equation:

72

Where A(f1) is the amplitude of the fundamental component A(fn) is the amplitude of the nth harmonic N is the number of harmonics

PRELIMINARY RESULTS

After performing the theoretical analysis, we started with the implementation part. We have worked on the design of a LabView program (see figure1) that combines all the measur ements that we are required to take.

Figure 1: LABVIEW Diagram

Initially, we have managed to build a diagram, based on the LabView software application, which allows us to obtain and read correctly the signal generated by the Function Generator, with the corresponding fundamental frequency and amplitude. The built-in functions in the LabView NI scope library generate the signal in an array format. We have added t he Fetch Data function and we have made the required connections in order to regenerate the signal in a waveform signal format. Hence, we are now able to input this signal to any functions and to any diagrams that require the input to be in a waveform format. In our illustrating example, we are reading a square wave from the function generator, with a frequency of 50 Hz, and a peak voltage of 0.8 V (see figure 2). We have chosen a square wave because it contains many odd harmonics.

After accomplishing this important first step, we have inputted the signal to the Auto Power Spectrum function. The Total Harmonic Distortion function of the LabView requires that one of its inputs be connected to the Auto Power Spectrum item (see figure 1). We have used the THD function in order to generate a digital display of the frequencies and of their corresponding amplitudes (see figure 2). We have implemented a simple X-Y graph function in order to plot the amplitudes versus the frequencies and thus, to obtain the amplitude spectrum of the signal (see figure 2).

Figure 2: LABVIEW Graphs and Displays

For measuring the power factor, we have used the Basic DC/RMS function (see figure 1) of the LabView to generate the DC and RMS values of the power signal, and we have displayed these values on a digital display (see figure 2). We divide these two values to get the power factor of the device as we have explained before. The power factor value is also shown on a digital display. We note that to obtain the power signal, we simply multiply the voltage and the current. We have designed our program such that the user is able to specify the number of harmonics and the fundamental frequency. In this example, the fundamental frequency is equal to 50 Hz, the number of harmonics is 21 (including the fundamental), and the sampling rate is 10 KHz. The signal is sampled at a rate of 10000Hz (which is greater than twice the maximum frequency in the signal to satisfy Nyquist The orem).

)(

)(100%

1

2

2

fA

fATHD

N

nn∑

=×=

73

The user can also choose between channel 0 and channel 1 to view and analyze the voltage or the current signal. We are going to use attenuating probes (with ratios of 1 to 100 and 1 to 1 respectively) to read the voltage and the current.

CONCLUSION

Before beginning testing on the electronic ballast, we need to provide an isolation solution to avoid phase to ground shortage and thus, in order to avoid damaging the scope card. After conducting a small research over the internet, we have found that the ISO 122 Texas Instruments Isolation Amplifier is the solution to our problem. We expect to be receiving the isolation amplifier in the very near future.

Finally, let us note that measuring the power factor and the THD of a device is very important in evaluating its performance. The more the power factor is less than 1, the more the circuit carries excess currents or voltages that do not perform useful work. For instance, if the power factor drawn by a given load is 0.5, the supply system must provide twice as much current as is really necessary. Losses in the supply system are actually proportional to the square of the current, so supply losses increase by a factor of four if power factor drops

from 1 to 0.5. This has substantial effects on the efficiency of the system. Hence, it is important to increase the power factor to acquire a better efficiency for the system.

On the other hand, the THD of good electronic ballast is internationally known to vary around the value of 10%. ACKNOWLEDGMENTS

We would like to thank our supervisor Dr. Sami Karaki for his continuous help, advice and his invaluable assistance during the work on the project. We would like to thank MEICO and all the engineers working there for sponsoring our project and providing us with all the equipment and facilities needed. We would like to thank also the EC E (Electrical and Computer Engineering) department at AUB and Dr. Fuad Mrad for providing us with a station where we have installed the cards and are conducting our work on a regular basis. REFERENCES [1] LABVIEW USER MANUAL

[2]Lander, C., Power Electronics, McGRAW -HILL, ENGLAND, (1981).

74

CNC Retrofit for Milling Machines

Doumit Imad ECE Dept. AUB, Beirut, Lebanon

[email protected] .lb

El-Mir Georges ECE Dept. AUB, Beirut, Lebanon

[email protected]

Nehme Makram ECE Dept. AUB, Beirut, Lebanon

Maknehme@hot mail.com

Abstract As a requirement for the curriculum of bachelor of Eng ineering, major Electrical Engineering, we are required to work on a final year project which constitutes a practical application to the theory taken in the previous years at AUB. For the sake of usefulness of the project and being concerned to deliver a project that will be in use after implementation, we chose to fulfill the need of a company in the industry. After several meetings with companies seeking projects that could enhance their work in the industry we landed on Phoenix Industries a part of INDEVCO Group who proposed that we implement a Computer Numerical Control (CNC) on a 3-Axis Milling machine.

INTRODUCTION Our goal is to implement a CNC on a Milling machine. At the end of the semester we should deliver a package that will process an AutoCAD drawing and obtain the object represented on the drawing using the Milling machine. The project contains both software and hardware implementation. For our final project and because of time limitation our project will consist only of the software part. The hardware to be purchased will require a delivery time of at least four weeks in addition to a cost of $10,000. We therefore agreed with both our advisor and the company to deliver by the 6th of June 2002 a software part that will be perfectly working in addition to a detailed research on the hardware to be used. We will also include a LabView simulation of the operation carried by the software. The hardware consists of a data acquisit ion unit that will control three servomotors; each servomotor will drive one axis of the Milling machine. The first step for milling will be to identify the object in the drawing in order to know with what kind of metal we are working because the hardness of the metal affects the speed that will be used in drilling. The next step will be to locate the zero reference point of the piece of metal. After locating the zero reference point a cleaning of the surface of the metal is to be done. Then the process of working the metal to obtain the shape pointed out by the drawing starts. Next we will see in which manner we are planning to process each step, which programming languages will be used, what hardware will be needed etc.

HOW TO APPLY CNC ON THE MACHINE

Figure 1: CNC Machine

The CNC Milling machine is controlled by 3-axis in the orthonormal space. As we can see from the figure, the 3-dimensional movements occur by turning the 3 screws in the 2 directions for forward or backward movements. Each of our three servomotors is going to be applied to an axis. On each motor, we will apply a sensor to control the number of rotations; this will be an indication of the linear distance covered in each direction. A fourth sensor will be set on the drill; it will be an indication of our zero reference point. Software part: As we can see from the diagram, our project will be divided into different parts. We will give a general overview of each part. • AutoCAD drawing: An object is drawn on

Aut oCAD; our program will transform the drawing to a readable file by the Milling machine.

75

• Visual Basic: In order to transform the AutoCAD drawing to a text file, we need to extract only the important information; this will be done using visual basic. We will save our drawing as a (.dxf) file and then use visual basic. This process consists of converting the drawing to a specific format that we will be able to use in the following steps.

• Transforming to a text file: Now that our drawing is transformed to a text file, we need to save it as a temporary file so we can use it to extract information to create our code by using visual basic.

• Visual C++: This part of visual C++ is used to convert from the previously obtained text file to our interface code.

• Interface Code: For explaining t he interface code we will give a short example: Xa will make the object move by a cm in the X direction. Yb will make the object move by b cm in the Y direction. Zc will make the object move by c cm in the Z direction. If we need to move in the X direct ion alone, we write in the code Xa. If we need to move in the X and Y directions simultaneously, we write in the code XaYb. If we need to move in the 3 axes simultaneously, we write in the code XaYbZc. This code defines the trajectory from our reference point to the end point, but not the path we need to follow. This path could be a straight line, an arc or etc. The relation between the motor speeds and accelerations on the different axis will define this path. This can be represented in the following manner:

• LabView: Now that our trajectories are defined, we need to make the interface with the data acquisition

unit to the motors. We will program this card using LabView according to the interface code.

Hardware part: Concerning the hardware part, after several researches lead on the subject, we were faced to two alternatives: Siemens SINUMERIK Hardware option: The Siemens provided the SINUMERIK 802C for CNC application. The package consisted of a PLC in addition to drivers to the servomotors. The problem with Siemens was the delivery time, which is very long in addition to the high cost of the package and absence of technical feedback in case of malfunctions. National Instruments motion controller NI7344 used with YESKAWA servomotors and drivers. This option although harder to implement and to program was preferred on the previous one due to the better delivery time in addition of a better cost as compared to the Siemens package. We therefore advise the use of the NI motion controller with the YESKAWA drivers and motors.

ACKNOWLEGEMENTS

We would like to thank our advisor professor Ali El-Hajj for his valuable contribution to the success of the project, and Dr. Fouad Mrad for his continuous assistance and support throughout the project. We will not forget Mr. Khaled Joujou for his precious interventions and advice through a precedent experience on similar projects. Last but not least, our greatest recognition goes to Phoenix Machinery in the person of Mr. Chady Zablit who showed infinite initiative and support for us and offered all the help and support available for the success of the project.

.

76

Hoover Saucer “Defying Balance”

Selim Hobeika Dept. Mechanical Engineering,

Faculty of Eng. and Architecture, American University of Beirut


Nael Itayem Dept. Mechanical Engineering,



Marc Tabchy Dept. Mechanical Engineering,



Abstract Our Final Year Project consists of designing, realizing and building a flying saucer “The Hoover Saucer” that is remote controlled. The characteristics of this flying vehicle are vertical lifting due to the downward propulsion of air cr eated by one two-blade propeller, hovering at low altitudes and moving by directing the flow of air by the rotation of the four flaps located below the propeller (in the flow of air). Our main challenge in realizing the Hoover Saucer is to counter balance the torque created by the propeller on the body by rotating the flaps. The improvements in our design over exis ting vertical take off spacecrafts like helicopters is that they use a rotor placed on a long tail to counter balance the torque caused by the main blades while us, by using this technique we were able to elim inate the tail and thus got a more compact and safer flying vehicle.

INTRODUCTION

The unusual and very disputed idea of designing and building a flying saucer came to us in the fall semester of 2001, during class, when we were preoccupied by submitting our FYP proposal. Building something that flies was to us a lot more interesting than staying on the ground and designing some machine or robot. The idea was a bit different from what everyone thought. In fact, we wanted to design a flying saucer that would actually hover the ground at low altitudes. Like every idea, it begins very simple in your mind but as it developed, things became a lot more complicat ed. We were however aware of the implications and the budget involved, which somehow limited our capacities and resources.

The application we had in mind for our Hoover Saucer was in the beginning very limited and not very convincing. It was more a challenge than an invention. But as things evolved, more and more applications came to our minds, and our idea was finally accepted as an FYP.

THE DESIGN The Shape Symmetric Circular Shape The Hoover Saucer has a circular shape, much like a disk with the center having a thickness greater than the extremities, which gives it a shape proper to flying saucers. It is most importantly symmetric with respect to the horizontal and vertical planes passing through its center. This design was chosen in order to have the most

aerodynamic shape to minimize friction with air during flight, vert ically and horizontally. It also enables us to easily obtain an extremely balanced vehicle.

Compactness We insisted on having that symmetrical shape so we could avoid using the tail of helicopters with a rotor preventing the rotation of the vehicle, therefore obtaining a more compact vehicle. This is a feature we will talk about in more detail below, in the “safety” section. The Orientation Although symmetrical in shape, it will be orient ed in a specific direction. Moving forward, backward and to the sides will be possible through four rotating flaps that will change the direction of the air. Full rotation will also be possible by the movement of the same flaps in both directions. These flaps will be located under the propeller, on the bottom surface of the body. Their initial position is vertical and each one of them rotates by a small electrical steering system (servo). This navigation system is very simple, and relies mainly on the person holding the controls, with no additional gyroscopes or autopilots, which make the handling of the vehicle very delicate and difficult. The forward direction is indicated by a red strip that serves as a reference to orient the vehicle. In the following discussion, we will take the front side of the flying saucer to be the side pointing in the forward direction.

Turning Left and Right The right and left motion is obtained by simultaneous rotation, in the same direction, of the flaps in pairs. So to turn left, both front and rear flaps rotate towards the right while to turn right, they rotate toward the left.

Moving Forward and Backward The forward and backward motion is obtained in a similar manner but with the other two flaps. The forward motion is obtained by the rotation of the left and right flaps toward the rear. And to go backwards, the same flaps rotate toward the front. Rotation The rotation of the saucer on itself can also be achieved in both clockwise and counter-clockwise directions by a simultaneous rotation of the four flaps by pairs in opposite directions. This is actually the same command that allows us to counter-balance the torque of the propeller as will be seen later. All these simultaneous movements are possible through the mixing of the

77

commands from the digital radio, and this will also be explained later on.

The Stability

The main challenge in our design is to maintain the veh icle in a stable position, preventing it from rotating on itself and in a practically horizontal position. The Counter-Rotation

Our main concern was to cancel out the counter-balance caused by the torque developed by the rotating propeller. Our initial design consisted of placing two counter-rotating propellers on top of each other. This would have solved our problem. But after some brainstorming and gathering information about small-scale fuel engines and their specifications, we realized that with an engine rotating at a speed of around 15000 rpm, we would never be able to use a gearing system that would give us a counter-rotating system consisting of two propellers, because of the needed accuracy and the very small tolerance for error. Furthermore, such a system would also require considerable additional weight, a weight that would be very difficult to lift for the available engines. We managed to solve this problem by using the four flaps rotating simultaneously, and in opposite direction, in both ways as shown in the figure above. This was made possible through the four independent electronic steering systems (servos) and the mixing of the commands from the digital radio. Furthermore, we mixed these four motions with the fuel throttle control servo of the engine at five different points of operation (a curve is automatically interpolated through these points), so that the rotation of the saucer is prevented at any operating point. Of course, many factors can affect the accuracy of the system (like the wind for example), and the person controlling the flying saucer should always account for some rectification. The H orizontal Stability

To begin with, the body we used is symmetric, which means it is mass balanced about the center. Regarding the horizontal stability, this was easily achieved by placing the components in a way to achieve a perfect mass balance. To find the proper position of each element, we hung the saucer by a resistant nylon wire from the middle of the engine mount and placed these elements in a way to obtain the most horizontal position. The fuel tank is the only variable mass used since its mass decreases as the fuel is consumed. We tried to balance the saucer by placing two fuel tanks symmetric about one of the middle planes, but it decreased the engine power (due to the usage of longer fuel pipes and T -junction) and increased the weight of our vehicle, negatively affecting the performance. We then decided to use only one fuel tank and placed it on the engine mount nearly in the middle, on the center of gravity, which solved our problem. This way, even when the fuel level decreases, the center of mass remains in the center.

The Propulsion

The vertically propelled air is obtained by means of a central rotating propeller with two blades, driven by a small-scale fuel engine. This propeller is enclosed inside the body. The air propelled downward causes a reaction from the air on the vehicle upward, with a magnitude greater than the weight of the vehicle, resulting in an upward force sufficient enough to elevate the vehicle. The use of flaps very close to the origin of the propelled air causes some losses in the lift force when turning, moving backward and forward, and even when counter-balancing the rotation. The analysis for the thrust and power obtained from the propeller and engine is discussed later on. The Safety

For safety measures, the blades are out of the reach of hands, covered from the top by a metallic fence and from the bottom by the flaps and the engine. The lightweight of the vehicle and especially the body that is made of Styrofoam will prevent it from being a dangerous mass if it happens to crash on someone. The absence of a tail on which a rotor is mounted like in helicopters is also another safety feature, and the reason is that these rotors are a main cause of accidents in helicopters. APPLICATIONS

The applications of the Hoover Saucer are numerous. It could be used either to detect mines by equipping it by a thermal camera, or it could be for fun purposes or even for large scale transportation. STUDIES AND ANALYSIS

The following studies were performed:

• Engine Mount Stress Analysis (Pro/Engineer, Pro/Mechanica)

• Propeller and air flow analysis (JavaProp, DesignFoil)

• Control Volume Balance Analysis

ACKNOWLEDGMENTS

The help and knowledge of Prof. Samer Abdallah, from the American University of Beirut, as our advisor for our project during the Fall and Spring 2001-2002 is very much appreciated. The expertise of Mr. Hadi Esta in flying R/C models and the advice he gave us for our project was of a great help, especially that he lend us very expensive and necessary components (Radio and Receiver). References

• Components catalogue

• Softwares:

o Pro/Engineer, Pro/Mechanica

o Working Model 2D, 3D

o DesignFoil R5.27

o JavaProp

78

Figure 1. Final shape back view of the Hoover Saucer

Figure 2. Isometric drawing of the Hoover Saucer uncovered (Pro/Engineer)

79

A Comparative Assessment within a Multi-grid Environment Of the Performance of Segregated Pressure-Based Algorithms

for Fluid flow at All Speeds

Daniel Asmar Mechanical Eng. Department, American University of Beirut


Abstract This paper presents a computational assessment of six pressure-velocity coupling algorithms (SIMPLE, SIMPLEC, SIMPLEX, SIMPLEST, PISO, PRIME) implemented within a finite volume multigrid framew ork. Continuity, conservation of momentum and conservation of energy equations are discretized to produce general algebraic equations that are solved using di fferent grid configuration using, single grid, prolongation grids and full multigrid algorithms. Three test problems were used as the basis for the comparison. For most of the single and prolongation tests SIMPLE required the least CPU time but in cases involving anisentropies became unstable and diverged. SIMPLEC and SIMPLEX were the most stable and reliable in all test cases; they were the fastest in multigrid test and lagged slightly behind SIMPLE in the single and prolongation tests. PISO required the least number of iterations to converge in all tests but was slower than the above three. SIMPLEST and PRIME required the largest number of iterations and longest CPU time; the difference between them and the above four algorithms was large. Moreover, they were very unstable in most of the test cases and required considerable bleeding (50%) for the advection term in order to converge. The highest acceleration from single grid to multigrid was recorded for SIMPLEC during subsonic flow through a nozzle at 18.4. The lowest was for hypersonic flow through a nozzle where no acceleration was observed.

Keywords Compressible 2D flow, all speeds, multigrid environment.

INTRODUCTION Over the last two decades important advances in Computational Fluid Dynamics (CFD), pertaining to the development and maturity of solution algorithms, have been achieved. In this work, a solution algorithm, such as the SIMPLE [2,3] algorithm, denotes the procedure

1 Patankar, S.V.,Numerical Heat Transfer and Fluid Flow,

Hemisphere, N.Y., 1981. 2 Patankar, S.V. and Spalding, D.B.,”A Calculation Procedure

for Heat, Mass and Momentum Transfer in Three-Dimensional Par abolic Flows,” International Journal of Heat and Mass Transfer, vol. 15, pp. 1787-1806, 1972.

used to resolve the coupling that arises in the solution of Navier -Stokes equations between velocity, density, and pressure. Many difficulties connected with these algorithms have been resolved and better insight gained. In specific, work has been directed towards settling a number of pending issues related to the choice of prim itive variables (density-based versus pressure based), the grid arrangement (staggered versus collocated arrangement), and the solution approach (semi-direct versus segregated approach). In this paper a collocated grid is used and the solutions are obtained using a segregated pressure based approach. In the segregated approach, the discretized forms of the various differential equations are solved separately, but over the whole domain. This has the advantage of requiring considerably less computer storage than the semi-direct method in addition to providing the flexibility of easily solving additional partial differential equations (such as turbulence kinetic energy, turbulence dissipation rate, concentration of chemical species, etc…) when needed. In this technique, while the velocity components are obtained from the corresponding momentum equations, there is no apparent equation governing pressure. To derive a pressure or an equivalent pressure-correction equation, the discretized forms of the cont inuity and momentum equations are combined together. Moreover, for compressible flows, density is replaced by pressure through the equation of state. Once the pressure (or pressure correction) field is calculated the velocity and density fields are updated to satisfy the continuity equation. In the cell-centered or collocated grid the momentum equations link the velocity to the respective pressure gradients, the continuity equation, apparently having no direct link to pressure, acts as a constraint on the velocity field. Consequently, the convergence and stability of pressure-based algorithms depend largely on how the pressure gradients and velocities are evaluated in the continuity and momentum equations. In order to avoid the checkerboard splitting of the pressure field, a special interpolation procedure for evaluating the control volume face velocities is adopted. In this method, the momentum equations at the cell faces are reconstructed by interpolating the coefficients at the cell centers. Within this context, a large number of solution algorithms have been proposed. The first of these

80

algorithms was the SIMPLE (Semi-Implicit Method for the Pressure Linked Equation) algorithm of Patankar and Spalding [2,3]. A number of modifications to the SIMPLE algorithm have been suggested with the aim of improving the robustness and/or convergence rate. The SIMPLEC (SIMPLE Consistent) algorithm of Van Doormaal and Raithby [4] proposed in 1984, the SIMPLEST algorithm (SIMPLE ShorTened) of Spalding [5], the SIMPLEX algorithm of Van Doormal and Raithby [6], The PISO (Pressure Implicit with Split Operator) algorithm of Issa [7], the PRIME (PRessure Implicit Momentum Explicit) algorithm of Maliska and Raithby [8] are the most important attempts to improve on the original solution algorithm. The problem with the iterative methods is twofold. Firstly, the convergence rate of the iterative method slows down considerably as the “coarse wavelength error” is eliminated and the elimination of the “smooth wavelength error” begins. This leads to time consuming and costly solutions to flow problems. Secondly, as the grid becomes finer and the number of algebraic equations increases, the convergence time increases exp onentially with the number of grid points. The above dilemma promoted the need for an alternative method that can alleviate these convergence problems. Multigrid methods have been successfully used in the past two decades and have been found to greatly improve the convergence rates of single grids. The main objective of this work is to provide a comparative assessment of the performance of six pressure-velocity algorithms in a multigrid framework, namely SIMPLE, SIMPLEC, SIMPLEX, SIMPLEST, PISO and PRIME by using them consecutively in the solution of three flows problems at all speeds and comparing the convergence rates for CPU times and number of iterations. In what follows the governing equations for compress ible flow are presented and their discretization procedure is out lined.

THE GOVERNING EQUATIONS

3 Van Doormaal, J. P. and Raithby, G. D.”An Evaluation of

the Segregated Approach for Predicting Incompressible Fluid Flows,” ASME Paper 85-HT -9, Presented at the National Heat Transfer Conference, Denver, Colorado, August 4-7, 1985.

4 Spalding, D. B. “Mathematical Modelling of Fluid Mechanics, Heat Transfer and Mass Transfer Processes,” Mech. Eng. Dept., Rept. HTS/80/1, Imperial College of Science, Technology and Medecine, London, 1980.

5 Van Doormaal, J. P. and Raithby, G. D.”An Evaluation of the Segregated Approach for Predicting Incompressible Fluid Flows,” ASME Paper 85-HT -9, Presented at the National Heat Transfer Conference, Denver, Colorado, August 4-7, 1985.

6 Issa, R.I.,”Solution of the Implicit Discretized Fluid Flow Equations by Operator Splitting,” Mechanical Engineering Report, FS/82/15, Imperial College, London, 1982.

7 Maliska, C.R. and Raithby, G.D.,”Calculating 3-D fluid Flows Using non-orthogonal Grid,” Proc. Third Int. Conf. on Numerical Methods in Laminar and Turbulent Flows, Seattle, pp. 656-666, 1983

The governing equations are those derived from the continuity law, Newton’s second law and the first law of thermodynamics. In addition, since the fluid is compressible, ideal gas law applies. These equations are written as: Continuity:

0)( =+∂∂

udivt

rρρ

Conservation of momentum:

xBuGraddivxP

uudivtu

++∂∂

−=+∂

∂))(()( µρ

ρ r

yBvGraddivyP

vudivtv

++∂∂

−=+∂

∂))(()( µρ

ρ r

Conservation of energy:

( ) ( ) ( )

+Φ+

⋅∇−⋅∇+∂∂β+∇⋅∇=ρ∇+

∂ρ∂ qPP

tPTTk

c1)T(

t)T(

p

&vvv.

Where:

( )

∇−

∂∂

+∂∂

+

∂∂

+

∂∂

µ=Φ 2222

32

xv

yu

yv

xu

2 .v

An adequate manipulation of these equations allows their representation in a unified equation of a general scalar as follows:

( ) ( ) φφ φφρρφQ

t+∇Γ⋅∇=⋅∇+

∂∂

v)( (1)

THE DISCRETIZATION PROCEDURE

The general equation presented above is integrated over a differential control volume to yield:

( ) ( ) ( ) ∫∫∫∫ΩΩΩΩ

Ω+Ω∇Γ⋅∇=Ω⋅∇+Ω∂

∂dQddd

tφφ φφρ

ρφv

After completing the following steps: • Replace the volume integrals by surface integrals

by applying the divergence theorem. • Replace the surface integrals by a summation of

fluxes. • Discretizing these fluxes by using a suitable

interpol ation profile (a high resolution scheme applied within the context of NVSF methodology [9])

The following is obtained: φφφ φφ P

PNBNBNBPP baa += ∑

)(

(2)

Where Pφ is the value of the scalar at every grid point, and NB are the neighbors of P, i.e. north, south, east and west.

8Darwish, M. S. and Mouk alled, F. , NormalizedVariable and

Space Formulation Methodology for High Resolution Schemes, Numerical Heat Transfer, Part B, 26, 1994, 79-96

81

THE PRESSURE CORRECTION EQUATION All of the algorithms considered in this paper commence by an initial guess to the flow field variables, P*, u*, v *. These guesses are related to their corresponding exact values P, u and v by pressure and velocity correction:

P=P*+P’ u=u*+u’ v=v*+v’ The difference between the momentum equations applied to the exact values and the guessed values results in the follow ing expressions for the velocity corrections:

ji

jijijinbnbji a

APPuau

,

,',

',1

'',

)( −+= −∑

(3)

ji

jijijinbnbji a

APPvav

,

,',

'1,

'',

)( −+= −∑

(4)

At this stage the assumptions that are made to simplify (3) and (4) are the basis by which SIMPLE and its affiliates are defined. The discretization of the continuity equation yields:

0)()()()( ,1,,,1 =−+− ++ jijijiji vAvAuAuA ρρρρ (5)

Adding u’ to u* and v’ to v* and subbing them into (5) allows the derivation of the pressure correction equation as follows:

'

)(

''P

PNBNBNBPP bPaPa += ∑

THE SOLUTION PROCEDURE The procedures involved in the implementation of the six algorithms in this paper are very similar. Therefore, a samplealgorithm, namely SIMPLE will be presented as an example and the explicit implementation of the remaining algorithms can be found in the references. The SIMPLE algorithm consists of: 1) Guess the initial flow field 2) Solve the discretized momentum equations 3) Solve the pressure correction equation 4) Correct the pressure and the velocities 5) Solve all other discretized transport equations 6) Repeat loop until convergence is obtained.

RESULTS AND DISCUSSIONS

The physical correctness of the results obtained for the three test problems was verified by comparing the pressure and Mach number distributions of each of the three test problems to previous publications [10,11] and were found to closely agree. A comparison of the

9 B. Favini, R. Broglia, A. Di Mascio, ‘Multigrid acceleration

of second-order ENO schemes from low subsonic to high supersonic flows. International Journal for Numerical Met hods in Fluids, Vol. 23, 1996

10 Demirdzic, Lilek and M. Peric, “A Collocated Finite Volume Method for Predicting Flows at all Speeds” . International Journal for Numerical Methods in Fluids, Vol. 16, 1993

performance of each of the algorithms in the three test problems is presented in the following figures.

Flow over a Bump In the subsonic flow tests PISO is the algorithm that requires the least number of iterations for single grid, prolongation and multigrid 3 and 4. It is however not the fastest except for the case of single grid. Figure 1 shows the convergence rates for a multigrid configuration. While SIMPLE is the fastest for the prolongation configuration, SIMPLEC and SIMPLEX are the fastest for the multigrid configurations. This is due to the much more realistic simplifications that are assumed in the velocity correction equations. PRIME required the largest number of iter ations to converge and SIMPLEST took the longest time. Figure 2 provides information on convergence CPU time of the algorithms for all grid configurations.

S IM P LE S T

P

P

P

P

P

P

P

P

PP

P

P

P IS OP

P RI M ES IM P LE XS IM P LE C

1 00 20 0 3 0 0 4 00 50 0# of I te ra tions

1 0- 5

1 0- 4

1 0- 3

1 0- 2

UR

esid

ual

S IM P LE

F U LL M U LTIG R ID 4 L E V E LS

Figure 1.Convergence rates for flow over a bump at

subsonic speed

0

200

400

600

800

1000

1200

1400

1600

Single Prolong Mgrid 3 Mgrid 4

Grid Configuration

CP

U t

ime

PISOSIMPLEPRIMESIMPLECSIMPLESTSIMPLEX

Figure 2.CPU convergence time for subsonic flow

over a bump. In figures 3 and 4 convergence rates deal with supersonic flow and are much faster. PISO requires the least number of iterations for all configurations. SIMPLE was the fastest for the single grid and prolongation and SIMPLEC and SIMPLEX were the fastest for the multigrid. In the single grid through to the multigrid, PRIME took the largest number of iterations to converge but SIMPLEST took the longest time.

82

SIMPLEC

100 200 300 400 500# of Iterations

10- 5

10- 4

10- 3

10- 2

UR

esi

dua

lSIMPLE

FULL MULTIGRID 4 LEVELS

SIMPLEST

P

PPP

P

P

P

P

P

P

P

PISOP

PR IMESIMPLEX


supersonic speed

0

100

200

300

400

500

600


Grid Configuration

CP

U t

ime


Figure 4.CPU convergence time for supersonic flow

over a bump. In figures 5 and 6 the flow is transonic. The least number of iterations is again PISO for all algorithms. SIMPLE is the fastest for single grid and prolongation, closely followed by SIMPLEC and SIMPLEX. SIMPLEC and SIMPLEX are the fastest for multigrid. Simplest is the slowest one and the largest number of required iterations is PRIME.

S IM P LEXS IM P LEC

100 200 300 400 500 600# of I teration s

10 -5

10 -4

10 -3

10 -2

UR

esid

ual

S IM P LE

FULL M ULTIGR ID 4 LEV ELS

S IM P LES T

PPPPP

P

PPPPP

PPPPP

PPPPPP

PPPPPPPP

PPPPPP

PPPPP

PPPPPPPPPP

PPPPP

P

P IS OP

P R IM E


transonic speed

0

500

1000

1500

2000

2500


Grid Configuration

CP

U t

ime


Figure 6.CPU convergence time for transonic flow

over a bump

The above results advocate a repeating pattern for all flows over the bump at all speeds: 1) PISO requires the least number of iterations

followed by SIMPLEC, SIMPLE, SIMPLEX and finally SIMPLEST and prime.

2) SIMPLE is the fastest algorithm to converge for the single and prolongation grid configurations and SIMPLEC and SIMPLEX are the fastest for the mult igrid configurations.

3) PRIME requires the largest number of iterations and SIMPLEST takes the longe st time to convergence.

4) PRIME and SIMPLEST are very unstable and required additional excessive bleeding to the advection term in order to obtain convergence.

Flow through a Converging-Diverging Nozzle The results that are presented in the following figures for subsonic flow inside a nozzle concur with the previous findings. PISO requires the least work for single grid and multigrid 4 but in prolongation and multigrid 4 SIMPLEC performs better. As far as CPU time is concerned SIMPLE is the fastest for single grid and prolongation configurations and SIMPLEC and SIMPLEX are the fastest for the mult igrid problems.

SIMPLEST and PRIME were gain the worst performers.

PRIMESIMPLEXSIMPLEC

250 5 00 750 1000 12 50 1500# of I terations

10-5

10-4

10-3

10-2

10-1

UR

esid

ual

SIMPLE


SIMPLESTP

P

P

P

P

P

PP

PP

PPPP

PP

P

PISOP

Figure 7.Convergence rates for flow inside a nozzle

at subsonic speed.

83

0

1000

2000

3000

4000

5000

6000

7000

8000


Grid Configuration

CP

U t

ime


Figure 8.CPU convergence time for subsonic flow

inside a nozzle

Flow around a NACA 0012 Airfoil In the following graphs the convergence times and number of iterations for flow around an airfoil at transonic speed are presented. Again in this section SIMPLEC and SIMPLEX were the only two algorithms that didn’t require additional bleeding for the advection term. Therefore these two algorithms will be considered the best performers and the remaining algorithms will be compared to each other. PISO required the least number of iterations followed by SIMPLE. The fastest of the four was the SIMPLE algorithm for single grid but PISO was faster for the prolongation and the multigrids. SIMPLEST and PRIME were the slowest and requires the largest number of iterations.

SIMPLESTP

P

P

P

P

P

P

P

PP

P

P

PISOP

PR IMESIMPLEXSIMPLEC

100 200 300 400 500# of Iterations

10- 5

10- 4

10- 3

10- 2

UR

esid

ua

l

SIMPLE


Figure 9.C onvergence rates for flow around an airfoil at transonic speed.

0

2000

4000

6000

8000

10000

12000

14000

16000


Grid Configuration

CP

U ti

me

PISO

SIMPLEPRIMESIMPLECSIMPLESTSIMPLEX

Figure 10.CPU convergence time for transonic flow

around a NACA 0012 airfoil.

CONCLUSIONS

The results of this paper lead to the following conclusions: • SIMPLEC and SIMPLEX are the most stable and

consistent between the algorithms. Their usage is highly advisable.

• Although PISO required the least number of iterations it was prone to instability in problems that involved non isentropic regions (shocks).

• SIMPLE was fast in most of the problems but was also sensitive to shocks and became unstable in such problems.

• PRIME and SIMPLEST are very poor performers in respect to convergence CPU time and number of iterations. It is recommended that they not be used in future problems.

ACKNOWLEDGMENTS

The help and support provide by Professor Fadl Moukalled and Professor Marwan Darwish is greatly appreciated.

84

CHOCOOLER©

A Mini Thermoelectric Cooler for Chocolate Bars

Fadi Hachache Dept. of Mechanical Eng’g, AUB


Ibrahim Geha Dept. of Mechanical Eng’g, AUB


Abdul Karim Younis Dept. of Mechanical Eng’g, AUB


Keywords Reynold number : Re Prandtl number : Pr Nusselt number : Nu

INTRODUCTION Thermoelectricity is the direct conversion of heat energy into electrical energy (the Seebeck effect) or vice versa (The Peltier effect). The solid -state phenomenon is electronic in nature and occurs due to charge carriers diffus ing from one junction to another. The Thermoelectric (Peltier) effect is a very convenient way of moving heat from one surface to another particularly when an electric supply is existent and when no moving parts can be tolerated. Thermoelectric cooling has little competition when applied to small size system since usual vapor compression systems are not forthcoming at small sizes. Since it is currently of relatively low COP (0.1-0.5), it is feasible when small to very-small scale application are required. Of course, the niche market will always be there. The working heart of a thermoelectric cooler is the thermoelectric module. The module consists of a number of p -n semiconductor (Bi2T e3) couples connected electrically in series and sandwiched between two thermally conductive but electrically insulating ceramic (alumina) plates. The cooler itself is made up of one or more thermoelectric modules attached to the container to be cooled via the required setup. The current task involved the design, building and testing of a Mini-thermoelectric-cooler (MTC) to cool some two typical chocolate bars to below ambient conditions. The design was to: be effective in cooling, light in weight, be capable of using a 12 V supply, have an easy-t o-open door, be such that replacing the thermoelectric module was easy, have good looks, and be inexpensive.

CHOCOOLER © CHOCOOLER© is approximately 17*15.5 cm long and 13cm high. Two heat sinks (aluminum) are provided, one from outside on the flat surface and one from inside. Inner Compartment The inner surface is made of roughened aluminum (0.5mm), which along with the curved floor aids in the circul ation of air within the cooler.

Becau se air becomes less dense as heated, it will flow to the top of the cooler where the inner heat sink is deliberately (and conveniently) installed. The sink is in complete contact with the flat cei ling and the module for

better conduction. The sink is attached to a 0.9W, 12V fan in order to circulate the air inside the cooler cavity. CHOCOOLER © is provided by a removable, holed shelf inside for the object being cooled (chocolate bars) to rest on.

Insulation The walls contain 4cm of fiberglass thermal insulation between the inner and outer walls. Fiberglass has a low thermal conductivity (k=0.031 W/m.K), and thus is a good choice for insulation. The choice of 4cm was made according to heat loss calculations (see below).

Outer Compartment

Steel was used for outer walls of CHOCOOLER© with a thickness of 0.5mm. The top contains a 4.2cmx4.2cm opening at the center where the module is installed. As for the heat sink, it covers the whole top such that a larger surface is available for heat removal. Its size was chosen upon calculations made (see below). A 2.0W, 12V fan (diameter = 8cm) is installed 1cm above the heat sink to move the air over the fins. The fan has a speed of about 2.66ms- 1 (cfm=30.7 ft3/min). An animometer was used to measure the speed of the fan.

Opening

The door of CHOCOOLER© opens to the right side, and a rubber lining ensures air- tightness.

Gages and Electrical Supply The fans are connected, along with the module, in parallel so that the module is provided with 12VDC. CHOCOOLER© has an ON/OFF switch to turn off the input current when not needed. A LED indicating good operation is included together with a fuse.

TESTING RESULTS Preliminary tests to date indicated a temperature of about 10oC achieved within CHOCOOLER© 6-7 minutes from onset of operation starting at an ambient temperature of 22oC.

CHOCOOLER’s net weight, measured without heat sinks is about 450 grams.

MODULE SPECIFICATIONS

Geometry Side length of each leg : 1.4mm Height of each leg (L) : 1.25mm Thickness of alumina ceramic: 0.65mm

85

Number of couples : 127

Bismuth Telluride (Bi2Te3) properties Seebeck coefficient : a = 190x10- 6 V/K Resistivity : ? = 1.6x10-5 O.m Thermal conductivity : k = 1.5 W/m.K

CALCULATIONS

Basic Calculations Figure of Merit : Z = a 2 / ( ? k ) = 1.5x10-3

Seebeck Coefficient for module : Sm = 2Na = 0.048 V/K

Resistance of module : Rm = 2N (? L/A) = 2.59 O Thermal Conductance of module :

Km = 2N ( kA/L) = 0.597 W/K

? Tmax = ½k T L2 = 62.6K (assume TL =15oC =288K)

IQ,max = aT L/Rm = 5.337 A

Vmax = a? Tmax + IQ,maxRm = 16.8 V

Taking ?T opt = 30K , at IQ,max = 5.337 A ð Qc,Imax = aT LI - k?T – ½ I2Rm = 19 W

Supplying the module with 5A from a 12V source ð Qc = 18.84W and P generated = IV =60 W ð Heat dissipated to outside = 78.84 W 80 W ð Heat sink should be designed to expel 80 W.

Heat Sink Calculations

Fin Calculations Fin dimensions:

L = 15cm t(thickness) = 1.8mm

e (extrusion of fin outside) = 2.6cm

Assuming at T 0 (temp. at bottom surface of sink) = 70 oC

and at T 8 (at top of fins) = 20 oC :

?0 = ( T 0 - T8 ) = 50K n (number of fins) = 15

Lc = e + t / 2 = 2.69 cm (corrected fin lengths)

Am = t . Lc = 4.842x10-5 m2 (profile area)

A fin = [(2L+2t) . e +L. t ] . n

Heat Convection (h) Calculations Diameter of fan ( d ) = 8cm

kair = 0.03 W/m.K ? air (density) = 0.998 kg/m3

vair = 2.66 ms -1

Q (flow rate) = v.A fan = 48.134226 m3/hr

cp (specific heat of air) = 1.009 kJ/kg.K

µ (viscosity of air) =2.075x10-5 kg/m.s

Re = (vair ? air d) / µ =10234.91

Pr = ( cp µ ) / k = 0.697892

Nu = 0.453 Re½

Pr?

= 40.650844

KAl = 206 W/mK

• Forced convection (due to fan)

h = [(Nu.k ) / e] x 80% =37.524 W/m2.K

(20% safety factor)

Figure 1. Efficiency for rectangular fin (Ref [1])

Lc 3/2 (h/kAm)1/2 = 0.271 nf = 0.9 (see Figure 1)

Forced flow on mid fins (7), along Lavg = 11.5cm

q = nf A finh?0 = 74.19 W

• Free convection (still air)

Along ( 8 ) fins h=6.25 W/m2.K

Lc 3/2 (h/kAm)1/2 = 0.1 n f = 0.97 (see Figure 1)

q = 19.797 W

ð qtot = 94 W ( >80W )

Heat Loss Calculations (U-value analysis) Path : Still Air =>Steel => Fiberglass=>Aluminum=> Inner Air Space

Air Rair_inside = 0.16 m2.K/W Rstill_air = 0.18 m2.K/W

Aluminum RAl = 0.00485 m.K/W x 0.5 mm

Steel Rsteel = 0.0221 m.K/W x 0.5mm

Fiberglass Kfiber = 0.031 W/m.K tfiber = 4cm R fiber = tfiber / kfiber = 1.29 m2.K /W

R total = 1.63 m2.K/W U = 1/ Rtotal = 0.613 W/m2.K ?T (bet. in and out) = 25 – 7 = 18K area of door form inside = 222cm2 Ainside = 41cmx 17cm + 222 cm2 x 2 Q =Ux Ainsidex ?T = 1.26 W (can be accommodated for) Reference [1] Heat Transfer, 7th Ed. – J. P. Holman [2] MCG 051 notes – TE cooling, Prof. R. Nuway hid

86

Isometric View

Front View

87

Human Electric Locomotive Project “HELP”

Raya Abdel Baki Faculty of Engineering and

Architecture Electrical engineering


Antoine Abboud Faculty of Engineering and


Beirut, Lebanon tony [email protected]

Chadi Chaker Faculty of Engineering and


Beirut, Lebanon chadi [email protected]

Our project, entitled HELP is a first attempt at combining our theoretical knowledge, with our professional skills; it is an integration of science and engineering. But above all, HELP is a humane endeavor, aimed at providing a simple but effective design for an electric wheel chair. This design will later be adopted by arcenciel, an association providing public utilities for the disabled people, in hope that it will be an asset of great value to its end users, the physically challenged.

HELP is essentially an assembly of electric drives, micro-controllers and electronic devices. It consists of two permanent magnet (PM) DC motors coupled to the rear wheels of a metallic frame constituting the base of the chair. The movement of the chair is manipulated by using a voltage-controlled circuit achieved using the four-quadrant (H-Bridge) chopper circuit depicted in figure 1.

Our basic contribution to HELP is designing the joystick circuit. The prominent feature of this design is the use of a micro-controller (pic 16F84A) to manage the direction and speed of motion of our vehicle. The four different positions of the joystick constitute part of the input to the micro-controller, based on which the direction of movement is decided: Front, backward or sideways. The pic processes the input data, using a downloaded program that we devised for our own purpose. The output of this signal is used to activate two of the four mosfets constituting the chopper circuit for each motor. Thus the direction of motion is produced.

A selection switch also contributes input to the micro-controller. Three switching positions correspond to three speeds at which HELP is operated. The output, a pulse width modulated signal is later stepped up and fed to the chopper circuit. The output of the H-Bridge is the desired

voltage level corresponding to the required speed.

An interesting feature of HELP is the use of a permanent magnet to activate the brakes: By injecting a rated field current the motor will unblock, so a button will be assigned to brake in case of emergency. We also propose the addition of a timer, in such a way that after a given time in the neutral position, the brakes will be automatically engaged. This procedure is essential to save power by eliminating the need for field excitation.

Another unique feature we wish to add is an anti-collision technology implemented so that the user of our chair can avoid colliding with objects when moving in the backward direction. This we hope to achieve using optic sensors installed within a special circuit attached to the rear of the wheel chair, so that at a certain distance from an obstacle, the sensors will trigger the latch on the brake, stopping the vehicle.

HELP implements two rechargeable batteries, each of rating 12V, 17Ah. The recharging of these batteries is achieved using a circuit utilizing a voltage regulator, a micro-controller (pic14C000), as well as other electronic devices. The preliminary proposed circuit is shown in figure 2. To allow for a high initial current, the best plan is to use a constant voltage current limited source, however, we chose to use a constant current allowing the voltage to rise until the battery is full. This current is controlled by a comparator and the pic14C000. The voltage level is compared to another fixed voltage established across a resistor. The output of the comparator is then fed to a FET, which controls the on-off state of the voltage. The charge current is switched off when the voltage level reaches a maximum value, preventing overcharge. It is important to note that the voltage level is monitored by the pic, which interrupts the charging operation once

88

per second to take a reading. We should also add that this feature of HELP is still at the design stage, and we have yet to study and perfect the best circuit to use for a battery charger.

Given enough time, we wish to improve on our design by adding different secondary options that will provide more comfort and convenience to our users. These include installing a fan for ventilation, and a 12V DC outlet which can be used for different purposes (charging a cellular phone, inflating the wheels..)

Our ultimate aim remains the safety and comfort of the disabled person, established at a minimal cost and an optimal engineering design. These qualities render HELP a marketable product, one which can be improved and manufactured for the industry. We hope to contribute to the efforts of arcenciel in that respect.

ACKNOWLEDGMENTS

We wish to acknowledge the efforts of each of: Professor Riad Chedid, our project advisor, for his time and invaluable technical assistance.

Professor Farid Chaaban, for his guidance and help. Professor Sami Karaki, for facilitating our endeavors with his professional knowledge. Mr. Fady Moujaes of arcenciel, for believing in us, and giving us the chance to help in a noble cause. REFERENCES [1] Blaurock, C., Make your own simple Rx/Tx

battery charger..

available at <http://www.uoguelph.ca/~antoon/gadgets/peakchrg.htm [2] Chapman, S. J., ElectricMachinery

Fundementals, 3rd ed., McGraw-Hill, Boston (1999).

Figure 1. H-bridge chopper circuit

Figure 2. Preliminary Battery charger circuit

89

Capacitor Placement Using Genetic Algorithm and Distributed Computing

Francois Layoun

ECE Dept., FEA, American University of Beirut,


Maher Salameh ECE Dept., FEA,

American University of Beirut, Beirut, Lebanon

[email protected]

Raji Sayegh ECE Dept., FEA,

American University of Beirut, Beirut, Lebanon

[email protected]

Abstract The purpose of our project is to find the fastest-optimal method of placing capacitors on a printed circuit board to reduce simultaneous switching noise. We illustrate the steps of the Genetic Algorithm used to find the best solution(s). Then we discuss the protocol that we developed to run on multiple computers and speed up the application.

INTRODUCTION This paper proposes the implementation of a capacitor placement scheme using genetic algorithm (GA) and distributed computing. GA is a new simulation-based solution generator. It adopts an innovative high-level model that enables efficient simulation and guarantees efficient results. In addition to that, distributed computing speeds up the execution of the program by dividing it into multiple fragments that can execute simultaneously, each on its own processor. GENETIC ALGORITM Our approach exploits an evolutionary algorithm to drive the search of effective patterns within the gigantic space of all possible sequences. The genetic algorithm, represents a solution to the problem as a genotype (or chromosome) given by a combination of 1’s and 0’s forming individuals. The genetic algorithm then creates a population of solutions and applies genetic operators such as mutation, crossover and fitness to evolve the solutions in order to find the best one(s). The three most important aspects of using genetic algorithms are: (1)definition of the objective function (capacitor placement), (2) definition and implementation of the genetic representation (combinations of 0 and 1 to represent the individuals), and (3) definition and implementation of genetic algorithms (the functions of the GA program: initialize(), keep_the_best(), elitist(), select(), crossover(), swap(), mutate(), report() ). One of the main goals of computer applications is the processing time. Small programs can be run on a

single processor in short instances; however longer programs will require more time and therefore the shift to parallel processing methods is necessary. A program being executed across N processors might execute N times faster than it would when using a single processor. This is why the system is cas caded in a master-slave scheme where the genetic algorithm procedures will take place on the main processor, and the circuit solvers on secondary processors. Main and secondary processors communicate continuously transferring necessary files. Sharing data is the fundamental motivation for networking computers. There are two machines involved in file transfer transaction, a client machine, and a server machine. The client is always the machine that initiates the transfer.

IMPLEMENTATION The system is implem ented on three computers representing a local area network with Microsoft Windows operating system. However, this system is scalable to many more computers. In order to carry out the transfer of information, we based our work on windows networking, more specifically WinSock applications. There are many popular protocols that we could have used in our program as the File Transfer Protocol (FTP) application described by the RFC 959. But in our program we wanted to minimize the time of transferring data (bytes), so we had to create and program our own protocol that can send and receive from the client to the server and vice versa with the “minimal amount of data”, i.e., the minimal amount of transferring time.

Winsock is a programming interface and the supporting program that handles input/output requests for Internet applications in a Windows operating system. Sockets are a particular convention for connecting and exchanging data between two program processes within the same computer or across a network. So the functionality of our program can be summarized in these steps:

• The server creates the socket and sends the circuit data to it.

• The clients connect to the server, take the circuit data and save it.

90

• The server generates the first population and sends to the socket.

• The clients take generations/(number of clients) from the socket, evaluate the “fitnesses” of each individual, and send them to the server

• Knowing the fitnesses, the server arranges the population, and the GA produces the new generation, and then sends it to the socket.

• The clients again take the generations and so on so forth…(check figure)

ACKNOWLEDGMENTS We would like to thank Dr. Sami Karaki for his supportive supervision throughout the stages of our work. His guidance has been invaluable in putting us in the right direction and motivating us to stay on track. We also would like to thank Mr. Weam Abou-Zaki for his help in the winsock applications. FYP WEBSITE Recommended to visit: www.geocities.com/fyp_ssl

REFERENCES

1- ParaScope, IEEE Comput er Society’s, 2002, Available Online: http://cac.psu.edu/beatnic/Edu_Train/HPF_tutorial/HTMLNotesnode173.html

2- Parallel Processing, Hank Dietz, 1999, Available Online: http://www.linuxdoc.org/HOWTO/Parallel-Processing-HOWTO-1.html#ss1.1

3- Parallel Processing Course, Prof. David E. Culler, 1999, Available Online: http://www.cs.berkeley.edu/~culler/cs258-s99/

4- lawrence T.pillage , Ronald A.Rohrer, and chnadramouli visweswariah, "Electronic circuit and system simulation methods". Mcgraw Hill 1994

5- Winsock Applications, 2002, Available Online: http://msdn.microsoft.com/library/

6- Chambers,L(Ed.) Practical handbook of genetic algorithms.Applications Volume I.

Client-Server Communication Client: Circuit Solver Client: Circuit Solver Send Send Fitness Fitness

Send population/

Number of computers

Server: Genetic Algorithm

First time: Send circuit

91

Geophysics and Engineering Hand in Hand

Nadim Baalbaki

Department of Engineering American University of Beirut


Joanna Doummar Department of Geology


Email: [email protected]

Abstract

In our report, we study the inter-relation between engineering and geophysics. On the one hand, increasingly challenging requirements for speeds, accuracy, versatility and economy of geotechnical surveying are bringing about the development of sensing equipments, techniques and data processing methods. On the other hand, better data acquisition and processing is providing a broader and more accurate understanding of engineering problems, thus reducing the costs of some projects and providing faster and safer solutions. We focus on the ground probing radar technique, and the resistivity method, which are used in shallow and deep subsurface investigations. We also discuss, through case studies, the big impact of these techniques on engineering dams.

INTRODUCTION Geophysics consists of locating shallow and deep subsurface structures and bodies and measure their dimensions and relevant physical properties. In this report two met hods of geophysics will be exposed the Ground probing RADAR method, and the resistivity method. Then this report will emphasize on the relationship between measurements and nature of the subsurface layers, or objects, then we will show the importance of these two methods in solving engineering problems

GROUND PROBING RADAR Brief History RADAR stands for Radio Detection And Ranging. It was first used in 1904 by Hulsmeyer and has, since then, undergone a lot of development, enabling it to probe at considerable depths into various media. This field was received by the coming of the lunar investigations because of its advantages over other methods. It is currently applied to numerous fields ranging from structural non- destructive testing and detection to archaeology. Although this method is relatively expensive, it is very useful accurate, and versatile.

Working Principles RADAR sensing is based on the principles of electromagnetic resonance. Different materials have different electromagnetic properties such as dielectric permittivity, conductivity, and magnetic probability, and thus will reflect electromagnetic waves differently.

Moreover, different spatial configurations and geometrical situations will yield different resonance patterns, which will also be affected by the equipment setup. Both frequency modulation and amplitude modulation are used.

Design Considerations As electromagnetic resonance is affected by material properties and spatial configuration, the design of both Sensor and emitter are affected. The working frequency range, the geometrical setup and the signal proceeding are specifically adapted to different situations, based on the desired resolution, penetration depth, and target nature. Furthermore, the design and its complexity depend on the economic considerations.

Equipments A RADAR (see Fig. 1) consists of an emitter that sends electromagnetic waves into the ground, and of a receiver that senses their reflection. This is however insufficient, as the sensed reflections have to be properly processed for noise elimination and then interpreted. GPRs usually use frequencies ranging between 10 MHz and 30 MHz for FM RADARS, and 20- 200 MHz for AM RADARS. GPRs can work down to a depth of 30 meters. Different antenna configurations can be used based on the situation of the target and the media in which it is.

Working Principles Resistivity in subsurface materials is measured based on Ohm’s law, which relates resistance to current and potential drop. Subsequently this law permits the calculation of true resistivity (ρ=VA/LI), where V is the potential dropped across the measured distance, A is the cross sectional area, L is the length and I is the passing current. However it is the apparent resistivity (Average of all the true resistivity) that is calculated because of the anisotropic nature of most soils.

RESISTIVITY METHOD Brief History The development of resistivity methods began in 1900s and was boosted in the 1970s by the progress in the computer technology, which provided better data processing and analysis means. This method is of great utility in ground water exploration and detection of seepage through dams.

92

Working Principles Resistivity in subsurface materials is measured based on Ohm’s law, which relates resistance to current and potential drop. Subsequently this law permits the calculation of true resistivity (ρ=VA/LI), where V is the potential dropped across the measured distance, A is the cross sectional area, L is the length and I is the passing current. However it is the apparent resistivity (Average of all the true resistivity) that is calculated because of the anisotropic nature of most soils.

Design Considerations This technique is very versatile because it exhibits many configurations and techniques of measurements suitable for profiling, vertical sounding, resistivity grid mapping 3D resistivity distribution. The strength of this method lies in the fact that each configuration is adapted for sp ecific requirements. Moreover the electrode spacing is the factor controlling the depth of resolution of the measurements not the length of the electrode.

Equipments This method uses four polarized electrodes Electric current is applied to the ground surface through two current electrodes (C1 and C2). The variation in the potential of the electrical field that is set in the earth is measured through two or three additional potential electrodes (P1 and P2). Four arrays are usually used: The Schlumberger, the dipole-dipole, the Wenner and square arrays (See Fig. 2). After having taken all the measurements, resistivity is plotted versus the electrode half separation. From this plot a resistivity contour map is drawn relatively to depth.

Subsurface Layer Characterization In subsurface layers, resistivity is controlled by electrolytic conduction through pores, fractures, faults, and shear zones. This conduction occurs in interconnected pore spaces, along grain boundaries. That is why there is a close relationship between porosity and resistivity given by the Archies law (ρ=aϕ -ms-nρe), where ρ/ρ e is the formation factor, ρ and ρ w are respectively, the resistivity of the solution in the pores, and the resistivity of the solution, ϕ is the porosity, S is the degree of Saturation, and a, m, and n are constants. The resistivity measurement allows to

draw a relationship between the data collected, and the type of soil,rock, and degree of porosity. On the one hand resistivity depends upon the nature of the soil, As shown in Fig. 3, a sandstone layer will exhibit low resistivity because of its high porosity and lack of compaction, unlike basalts, which have low effective porosity or granite, which has low porosity. Moreover the deeper the layer is the higher the compaction. On the other hand, geological processes have an important impact on resistivity (See Fig. 4), generally they tend to decrease it, faulting, for instance, increases the cracks in the rock, therefore increases its conductivity. Also resistivity is a function of salinity and water content, hence the solution with large amount of ions will exhibit low resistivity, and furthermore resistivity decreases as the water content increases. It might seems that clayey materials have high resistivity because of their low effective porosity, however electrolytic conduction is increased because of the formation of a double layer exchange cation formed at the boundary of the non- neutral clay particles.

COMMON APPLICATIONS TO ENGINEERING Through case studies, these geophysical methods will help in identifying first buried debris and objects such as tanks, pipes etc…, leaking through dams, which are represented on resistivity maps by zones of low resistivities because of the increase in water content in this zone. Moreover resist ivity method will also help in identifying possible buried faults on dam sites. These faults decrease the safety of the dams, on the one hand because if the fault is reactivated it might displace or shake the dam foundations, and on the other hand, it will increase the degree of seepage tremendously through the dam.

REFERENCES [1] Azzam, I., «Resistivity investigations of a dam site,

Chabrouh area, Lebanon». Masters Thesis, American University of Beirut (1974)

[2] Daniels D J., Surface penetrating Radar. Institution of electrical engineers, London (1996).

[3] Kayal R., « Geophysical investigation of bisri dam site », Masters Thesis, American University of Beirut (1985)

[4] Reynolds J., An introduction to Applied and environmental geophysics. John Wiley and sons England (1997).

[5] Ward S. H., Resistivity and induced polarization methods, Earth science laboratory university of Utah. Geotechnical and Environmental Geophysics, Nabighian M.N (1992).

93

Potential for Rail Freight in Lebanon

Rawad Hani Department of Civil and Environmental Engineering

American University of Beirut P.O. Box: 110236 / 2859

Beirut – Lebanon Telephone: +961- 3 -793171

email: [email protected]

Abstract This paper aims at an assessment of the market potential for rail freight in Lebanon, taking into consideration the country’s strategic position as a link between the Eastern and Western countries.

INTRODUCTION

Land Transport in developing countries is very important to their economies, since transport cost manifests itself as a significant component within the price of consumer commodities. Besides, business and industry in Lebanon are particularly dependent on communication links with other markets, and if prosperity is to take place, excellent links to major markets are essential. Hence it is necessary to promote a high quality transport infrastructure to address the regional and local distribution issues.

PRESENT CONDITIONS O F THE FREIGHT INDUSTRY The paper starts by examining the existing conditions of the freight industry in Lebanon. The vehicle fleet, which is mainly composed of pick-ups, is described. Then the types of shipping movements (imports/exports, transit flow, and domestic flow) are analyzed according to recent national statistics of freight movements at Lebanon’s border crossings. The present conditions of warehouses are then briefly discussed.

COMPETITIVE ADVANTAGES OF RAIL VERSUS FREIGHT The competitive advantages of rail versus truck are presented based on cost, economic trip length, and environmental concerns. The railroad industry is a high fixed cost industry, where high capital costs are involved in building the infrastructure. But once that infrastructure is in place there are relatively modest variable costs for operating the system. However, the economic trip length of the railroad is still to be considered, since rail freight is economical only for long distances. The environmental advantages of rail over trucks are also considered, where rail appears to be more environmentally friendly. The discussion touches on new technologies such as just in time systems (JIT).

INTERMODAL MOVEMENTS BY RAIL Potential benefits of intermodal rail transportation and the impact of intermodality on ports are then discussed. The technologies are defined along with the benefits which include lower accident levels, slower

deterioration of highways, and greater fuel efficiency. The importance of intermodality for Lebanese ports is then illustrated.

FORCASTED REVENUE FROM POTENTIAL RAIL FREIGHT Forecasted revenue from potential rail freight is then presented based on NCHRP (National Cooperative Highway Research Program) procedure, which identifies some measures that can be used in the demand forecasting of new facilities. The potential freight market is first identified which includes mainly the hinterland countries. Next, changes that are likely to occur in the market are forecasted based on projected economic growth rates. The final step consists of estimating the demand and consequently calculating the forecasted freight revenues.

FACTORS INFLUENCING FREIGHT TRANSPORT & POLICY MEASURES TO ENCOURAGE RAIL FREIGHT The factors influencing freight transportation and policy measures to encourage freight by rail are then discussed. The factors are adopted from NCHRP reports where they are classified as direct and in-direct. Policy measures that need to be adopted by the government are then proposed based on similar measures adopted in developed countries that might apply to the Lebanese context.

JOINT USE OF RAIL RIGHT-OF-WAY FOR PASSENGERS AND FREIGHT Joint use of rail for passengers and freight is then examined. This procedure which is mainly adopted in Germany and Japan helps in reducing the capital costs involved in acquiring the right of way by utilizing the same track for both passengers and freight where freight cars are usually operated overnights at times where passenger cars are not in business. Although this procedure seems to be very helpful, regulatory and operational considerations need to be accounted for.

REGIONAL RAIL NETWOR K CONSIDERATIONS Regional rail network considerations which include border crossings, transit, integrated development, and other operational constraints are presented. The rail system can not operate efficiently if the procedures adopted at the borders remain as they are nowadays for trucks, where freight shipments are delayed for a long time so that they can be inspected. Moreover, integrated development is needed so as to have a reliable system

94

that enables transport of people and freight. This last issue has been the emphasis of a major project started by the Economic and Social Commission for Western Asia Countries (ESCWA) and called “Integrated Transport System in the Arab Mashreq” (ITSAM).

COST OF ESTABLISHING THE RAIL FREI GHT INFRASTRUCTURE AND ROLLING STOCK The cost of establishing the rail freight infrastructure is then presented based on previous studies. The studies referred to were completed by the SOFRETU and SOFERAIL in 1994 and Frederic Harris Team in 2000. The costs of the above studies were different due to the fact that the later study only examined the Jounieh – Jiyeh corridor.

CONCLUSION AND RECOMMENDATIONS Finally an analysis is performed based on the aforementioned points and a set of recommendations and conclusions are presented. In short the results show that the rail option might be viable if we look at the wider regional context which includes Syria, Jordan, Iraq, and the Gulf Countries. The revenues as well as the costs of the rail freight industry are estimated, and based on them conclusions were built.

ACKNOWLEDGMENTS I would like to express my utmost gratitude for Professor Isam Kaysi for his continuous help and support on this project. Without his knowledge and dedication for help at all times this pap er might not have come into being.

REFERENCES

[1] A Guidebook for Forecasting Freight Transportation Demand, NCHRP Report 388, Transportation Research Board, National Research Council, 1997

[2] A Guide to Buying Rail Freight Services, Traffic Management v31, Cahners Publishing Company, 1992

[3] Bawarchi Fouad, International Freight Movements, United Nations – ESCWA

[4] El Boustani Bassam, Social Impacts of the Transport System, United Nations – ESCWA

[5] El-Fadel Mutassem, Environmental Impacts of the Transport System, United Nations – ESCWA

[6] Fredric R. Harris, Inc./Daniel, Mann, Johnson, Mendenhall & IBI, Alternative Analysis and Preferred Alternatives Report, December 2000

[7] IBI Group, Freight Demand and Modal Choice, Beirut, September 2000.

[8] Integrated Transport System in the Arab Mashreq (ITSAM), Transport Section, Sectoral Issues and Policies Division, ESCWA, United Nations

[9] Joint Operation of Light Rail Transit or Diesel Multiple Unit Vehicles with Railroads, Transit Cooperative Research Program (TCRP 52), Transportation Research Board, National Research Council, 1999

[10] Kaysi Isam, El-Husseini Ahmad & Ramadi Rabih, Computer based Modeling of National Goods Movement, 1999

[11] Kaysi Isam, Extent and Emissions Impact of National Freight Transport: Case Study of Lebanon, May 2001

[12] Kaysi Isam, Advanced Technologies for Fleet Management and Intermodal Freight Operations in ESCWA Region, July 1997

[13] Kumar Shashi, Intermodal Transportation: Need, Strategies, and Competitive Ramifications. Loeb-Sullivan School, Maine Maritime Academy

[14] Laitila Thomas & Westin Kerstin, The Eco Factor in Freight Transportation Demand, University of Umea

[15] Muller Gerhardt, Intermodal Freight Transportation, Intermodal Association of North America & Eno Transportation Foundation

[16] National Trade Facilitation Bodies, Economic and Social Council, United Nations 1999

[17] Nsouli Abdel Wadoud, Obstacles and Cost Analysis, United Nations – ESCWA

[18] Rail Transportation Policy, US Public Law 104 – 88, December 29, 1995

[19] Railtrack, UK, www.railtrack.com accessed December 2001

[20] Reviews of Freight Demand Forecasting Studies, Cambridge Systematics Inc.

[21] Sussman Joseph, Introduction to Transportation Systems, Boston: Artech House, c2000

[22] Seventeenth National Conference on Light Rail Transit v.1, Transportation Research Board, National Research Council, 1995

[23] SOFERTU & SOFRERAIL, Rehabilitaion de la ligne Ferroviaire Tyr – Beyrouth – Tripoli – Frontiere Syrienne, March 1994

95

Civil Engineering and Novel Technology: Where Do They Meet?

Loai Naamani Civil and Environmental Engineering Dept., AUB


Abstract This paper explores a diversity of technologies and IT tools that can serve civil engineering needs and help this sector restore its past glory by retaining a leading edge among peer disciplines. The targeted audience is the engineering body as a whole (students, faculty, and practitioners), along with high school students, namely eng ineering candidates.

While some assume that computer engineering and information technology (IT) are the future and that civil engineering is obsolete, there are many who can touch beyond this and see that such technological advancements, if properly directed, can add a lot to civil engineering, which in turn would proportionally reflect on our quality of life. A guest speaker (Nassif, 2002), in a seminar presentation given at the Civil and Environmental Engineering Department, put it bluntly by raising this caution flag (addressing faculty members and students): “It’s up to you guys to revolutionize this sector and instill novel technology into it; it’s by this and this only that you can put it back on the pedestal off which other engineering disciplines have recently had it displaced.” [1] He also commended the transportation sector people for being the civil engineering pioneers to realize and start implementing this. It is up to the new breed of civil engineers to attend to this and assess the means to and consequent benefits from such a full-fledged utilization of computerization and information technology in conventional areas/metho ds of civil engineering. While their fellow colleagues develop those high -tech tools, civil engineers’ efforts should lie in knowing how to successfully exploit such tools in every implicative manner. It is only then that a civil engineer achieves both: the satisfaction from subduing IT and computer technology skills, and the obligation towards his profession through applying them to serve numerous civil needs. The following is an e xemplary overview of where such utilization is underway, awaiting the energy and enthusiasm of civil engineers to come: Intelligent Infrastructures and Geographic Information Systems ‘Intelligent infrastructure’ development involves the integration of infrastructure building/modeling and information management using modern computer techniques and graphics technology with advanced database management sy stems (for maintenance and/or customer billing, for instance). Working with spatially networked facilities and land records sy stems would

highly benefit from a tool like a Geographic Information System (GIS). GIS have become a popular item on the wish list of many m unicipalities and water agencies. It would help the planning group perform estimates of future water demands, evaluate the transmission system utilizing these estimates, and specify subsequent system improvements. Then the engineering group can use the GIS in mapping such expansions, since it provides the spatial analysis tools necessary to efficiently assess the important factors (demographic, geographic, and economic) influencing the siting decisions for a wastewater treatment plant, for example. At a later stage, the O&M group can use it to manage work groups at geographically distributed facilities by using the geodatabase to provide work order management, work scheduling, and work history logging on a daily basis. Its use in this domain can even stretch to setting up hydraulic network models whose ‘input data’ is directly derived from the ge ographic and demographic aspects of the area under study; this is known as ‘coupled modeling’. (Of course, there is a diversity of GIS applications in other civil sectors too; the most pronounced would be those in transportation & traffic engineering.)

Innovative Monitoring and Inspection Methods A variety of advanced monitoring and inspection methods are being employed nowadays for maintaining countrywide infrastructures. In pipe rehabilitation, for example, mobile robotic systems (CCTV, ultrasonic sensors, stationary & zoom cameras…) are being used for remote inspection, and many ‘trenchless’ renovation techniques are being employed in refurbishing defective pipes, so as to minimize excavation requirements and disruption to surrounding structures and utilities. Real-time monitoring and predictive modeling (to provide reasonable projections of the remaining useful life of a structure before actual failure occurs) of existing infrastructures would help municipalities undergo preventive maintenance and repair/replacement, therefore, minimiz ing the need for emergency repairs. In the area of high-performance structures, monitoring gadgets such as stress-sensors installed into bridge girders for assessing vehicular stresses and frequencies, and infrared reflectors/receivers for monitoring scour at bridge foundations, are also being installed; not only for maintenance purposes, but also to serve as a source of input for undergoing research on rel evant topics. Similarly, actuated signalization and ITS (Intelligent Transportation Systems) in traffic engineering require

96

real-time monitoring systems (cameras, traffic sensors, etc…) to be able to quantify vehicular demand and characterize traffic patterns. Finally, we need not mention that in the age of the internet, what is recorded in ‘real-time’ becomes also accessible in ‘real-time’ through the luxury of having online databases accessible anytime anywhere, thus adding a whole new dimension to the p otential of IT in civil engineering. The fact that MIT has an ‘Information Technology’ area of concentration only in its Civil and Environmental Engineering Department and not in any other is also worth mentioning. One only needs to skim through its IT program (which is gradually becoming application-oriented and not crude IT), course offerings, and student projects to touch on the magnitude of contribution that IT can have to civil engineering. The proceedings and evolution of the CVEV118/698 course, which Kaysi and Mabsout [2] have formally documented for another conference, also easily attests to this ever-escalating concern and success of senior students in applying freshly acquired IT skills and computer methods to conventional civil-related problems. Finally, a distinctive line should be drawn between the two specialty areas to help steer the civil engineer’s IT pursuits in a way that serves the civil profession without him/her mistakenly wandering out of it. After all, such novel technologies are not but an interface between the two same old bodies of information: (1) the civil engineer and his wide-ranging knowledge, and (2) the

diversity of civil ‘duties’ manifested in facilities to erect, problems to tackle, etc… As those two bodies stretch in scope, the need for the interface (engineers’ tools in this context) to evolve arises. However, the newer the interface or tool the more tempting it gets, and by getting over occupied with the ‘tool’ we, as engineers, risk losing perspective on the central goal this tool was originally employed to serve. Heim said: “The deepest peril of the interface is that we may lose touch with our inner states.” – where the ‘inner state’, in this case, translates into our commitment as civil engineers to enhance the quality of life; we do not want to lose touch with that. ACKNOWLEDGMENTS

I would like to acknowledge Dr. Mounir Mabsout for his unowed encouragement and support.

REFERENCES

[1] Nassif, H., "Instrumentation, Condition Monitoring, and Evaluation of Bridges", Weekly CEE Seminar Series, FEA, AUB, Beirut, Lebanon, January 10, 2002.

[2] Kaysi, I. and Mabsout, M., "Computing in Civil Engineering Education: Dynamics of Change", Proceedings of The 9th International Conference on Computing in Civil and Building Engineering (ICCCBE-IX), Taipei, Taiwan, April 3-5, 2002

97

Cultured Engineers: A Necessity or a Privilege

Loai Naamani Civil and Environmental Engineering Dept., AUB


Abstract This paper encourages well-roundedness and pursuit of culture in engineering students’ careers. It also questions the soundness of compressing a 5-year program into 4-calender -years, as it is the case in the FEA. The targeted audience is the engineering body as a whole (students, faculty, and practitioners), along with high school students, namely engineering candidates.

There was a time when I considered my week futile if I could not squeeze reading a book, writing something worth writing, watching a movie, and enjoying a new album by the end of that week. I strain myself to squeeze any of this in a month’s time nowadays, and I am yet to succeed. Ever since my internship last summer in Florida, my academics-related commitments have been piling up, leaving me with hardly any time to do what I once prioritized over all else; it is also unfortunate that I am readily growing accustomed to this. Nonetheless, I do not consider myself disinves t ing my time nor misplacing my resources by pursuing my engineering career this wholeheart edly. All I imply is that I miss the ‘multidimensionality’ I once exercised as an engineering student, and that wanting to stand out as an engineering student comes at a rather high price, the price of having to cast off much of one’s selfhood. This is my take on the engineering educational sy stem at AUB, perhaps the only system on the planet that believes squeezing a 5-year curriculum into 4 is more reinforcing to a student’s career. If this were the case, then I assure you we would not be alone exercising this idiosyncrasy.

Although we do finish equally versed and a year earlier than those enrolled in 5-calendar-year programs at other universities, of what significance is this year when compared to a lifetime of balanced affairs and cultural involvement? The intensity at which engineering students are instructed and expected to harvest back at FEA (along with the ‘competition’ and other socioeconomic factors), leave committed students with virtually no time or energy to pursue other intrinsic aspects of a well-rounded life, which would most probably reflect on and reinforce (directly or indirectly) their professional careers in the long run. I believe a more ‘relaxed’ system will not only make engineers more ‘civil’, but also reflect positively on their attitude towards acquiring the same technical knowledge when

spread over a 5-year period. Students would then approach their su bjects/courses with far more receptivity for knowledge, less eagerness to finish, and ample time to flourish.

Our Engineering Lecture Hall boasts of a number of doctrines spread across its two stairways, one of which summarizes my whole argument: “What we should aim at producing is men who possess both culture and expert knowledge in some special direction. The expert knowledge will give them the ground to start from, and their culture will lead them to as deep as philosophy and as high as art.” – Alfred North Whitehead. Yet, again, I see too much expertise and barely any culture; unless the 2 or 3 ‘cultural electives’ imposed on engineering st udents are to be considered their share of culture. I believe ‘culture’ is not to be injected from a syringe, but rather to be quaffed from a well. And, as provisional as the effect of an injection is the knowledge we retain from imposed culture; hence my request for a ‘relaxed curriculum’ and not additional cultural courses. Obviously , this essay assumes that the majority of st udents would understand a ‘relaxed curriculum’ as an opportunity to drink more from that ‘well’ and would readily see to that, had the ‘relaxation’ been granted. (However, this essay does not tackle the ‘cultural gap’ problem our society faces as a whole; its relative indifference to fine arts, cultural studies, literature, philosophy, world history, etc… are all intrinsic problems beyond the scope of this abstract.)

I forgot who said, “An expert is he who knows more and more about less and less, until he knows everything about nothing.” In a sense, I cannot but agree. Technical expertise is nothing when one cannot answer his son to a question like, “Was it Plato who taught Socrates or vice versa?” Plunging deep into one’s field of expertise with indifference to all else, is like staring at one’s favorite painting while standing 1 inch away: one can discern the brush stroke and canvas fabric, yet he is oblivious to its subject matter, color theme, surrounding frame, and relative position on a wall among other paintings. One simply starts to lose perspective, and losing perspective leads to detachment from one’s social circle. This brings us to the most important non-professional merit of ‘multidimensionality’: the convenience of being able to relate with anyone on anything; be that from engaging in technical discourse with a colleague, to improvising a set of verses before an anxious lover. As to its direct professional revenue, let us refer back to A. N. Whitehead’s statement: “… and their culture will lead them as deep as philosophy and as high as art.” Take the

98

case of a civil engineer: stretching the artistic dimension of his profession to its limits brings us to architecture. That is, with some appreciation for artistry along with a sense of perspective, the civil engineer becomes a semi-architect; aesthetics readily becomes an as integral of a component in his designs as safety, serviceability, and economy presumably are. Regarding the “… as deep as philosophy…” edge of it, consider designing an interface for a software application: with a slight insight into the psychology of human-computer interaction, we can highly optimize the overall ergonomics of our software tool; hence making it more user-friendly and popular. It also cannot be stressed more how much one’s programming skills would benefit from the slightest literacy in the philosophy of logic and syllogistic reasoning. Furthermore, this versatility would also reflect on how we present/market ourselves as di stinguished engineers: be that from enriching our communication skills and expanding our library of analogies, to helping us design appealing layouts for PowerPoint presentations.

From conversations with a few older colleagues and faculty members, I understood that such ‘cultural enlightenment’, if I may call it, was cherished at a later stage: du ring graduate and post-graduate school abroad. It was then that interests traversed other non-technical fields, and course offerings in such fields were surprisingly in demand. They spoke of different interests ranging from photography to Arabic literature; yet they all agreed on one fact: regretting such ‘enlightenment’ had not come (nor been feasible) at an earlier age. Speaking of ‘abroad’, I also find it worth mentioning that conversations held during lunch meetings with my superiors and international coworkers in Florida never revolved around our proceedings at work, but rather around each one’s cultures and his/her respective view of the others’. It proved to be the common language of the global co mmunity. After all, an Italian is unlikely to be as overwhelmed by your knowledge of the Pisa Tower’s annual inclination rate, as he would if you knew that Pisa is Galileo’s birthplace!

All in all, I ache to see some of us graduating

without being able to tell the difference between a violin and a guitar (not only acoustically, but also visually). I ache to see some fail to differentiate Picasso’s ‘Dora Maar’ from Da Vinci’s ‘Mona Lisa ’. One could not come to admire an engineer who has never opened a book for a reason other than checking a formula or noting a reference. Perhaps it is time books are read more than newspapers, and libraries are visited in occasions other than the reading period. Perhaps it is time students’ political awareness and activism (since the only other is academic) is paralleled – not replaced – by interests ranging from appreciating sculpture to playing ping-pong… Yet, if I were to voice this, I will get the unanimous reply: “We would love to, but there are too many curricular commitments for us to squeeze in the extracurricular.” I believe a viable solution would be ‘relaxing’ the system (assuming we must abide by the 5-year curriculum requirement dictated by the Lebanese Order of Engineers), and spreading it over its rightful 5-year duration. The other alternative would be ‘putting up’ with the 4-year hustle, while exerting a superhuman effort to change the old adage to: “A master of all is not necessarily a master of none.” – More precisely, I preach: mastery of one with versatility in all. This is ‘so that we may have a life, and truly have it more abundantly!’

ACKNOWLEDGMENTS

I would like to acknowledge Dr. Mounir Mabsout for his unowed encouragement and support.

99

Author Index

Abboud, Antoine, 84 Abdel Baki, Raya, 84 Abi Haidar, Chucri, 56 Abou Ghantous, Fadi, 56 Abou-Rjeily, Joe, 45 Asmar, Daniel, 76 Atallah, Paula, 47 Ayoub, George, 5 Azar, Nabil F., 11 Azar, Sima, 58 Azarian, Michel M., 62 Baalbaki, Nadim, 88 Bissat, Jade, 45 Chaker, Chadi, 84 Chalhoub, Ayman, 54 Choueiter, Ghinwa, 41 Dandach, Sandra, 58 Daoud, Rima, 49 Daouk, Carine, 29 Doumani, Mounir M.C., 37 Doumit, Imad, 71 Doummar, Jouanna, 88 El-Chidiac, Tarek O., 64 El-Esber, Lina, 29 El-Mir, Georges, 71 Fadel, Hicham, 52 Geha, Ibrahim, 81 Ghanem, Jean, 49 Ghibril, Mark, 54 Ghulmiyyah, Milhim J., 12 Hachache, Fadi, 81 Haddad, Habib, 52 Haddad, Wissam, 54 Hajj, Ibrahim N., 5 Hamadeh, Nayla, 49 Hamzeh, Najwa, 47 Hani, Alain M., 62

Hani, Rawad, 90 Harik, Mario, 45 Hobeika, Selim, 73 Ibrahim, Jihad, 33 Iliya, Raja, 5 Itayem, Nael, 73 Kailath, Thomas, 16 Kamal, Hassane, 33 Kammoun, Fouad, 29 Kanafani, Adib, 13 Karaki, Houda, 33 Koulakezian, Mossine, 56 Kuran, Albert, 5 Layoun, Francois, 86 Makhoul, John, 14 Malkoun, Joseph, 68 Moussa, Hady, 52 Naamani, Loai, 92, 94 Nahra, Elias, 47 Naous, Youssef, 41 Nehme, Makram, 71 Ohannessian, Nesrob, 41 Rifai, Hussein, 15 Saad, Roland N., 62 Sabbah, Jawad T., 64 Salameh, Maher, 86 Sarkis, Marianne, 68 Sarkis, Michel, 68 Sayegh, Raji, 86 Tabchy, Marc, 73 Younes, Abdul Karim, 81 Zouheiry, Adel T., 37

faculty of engineering and architecturestaff.aub.edu.lb/~webfea/feasc/previous/proceedings.pdf ·...

Documents