# Dept

Document Sample

					           Department of Electrical and Control Engineering

Title：(Subproject II) A Study of Human Activity, Emotion, and physiological
Signal Monitoring System for Health Care Applications (II-III)
Principal Investigator：Chang Jyh-Yeong
Keywords：Action recognition, Emotion recognition, Fuzzy classifier, Feature
selection and transformation, Fuzzy ID3

It is known that human activity, facial expression, and physiological signals
explicitly shed light on the health and comfort status of a person. Based on
recognizing video-based human activity, facial expression, and physiological signals,
the purpose of this project is to build an automated health monitoring system to
determine the physical and mental comfort of a person so as to predict one’s health
condition [1-4]. With an objective signal conditioning, we will develop a machine’s
ability to recognize human affective health state by watching through a CCD camera
over a person’s action and face. By the use of various physiological signals as well,
such as skin conductance, blood pressure, respiration rate, and electromyogram (EMG)
measurement, they can potentially aid in assessing the health condition [22] of a
person. Machine learning and its intelligence algorithms will be developed and can
hopefully predict the short-term, and very likely extend the time duration, the health
condition of a person. The predicted health condition is automatically generated and
transmitted and thus is helpful for notifying the corresponding family or medical
treatment if necessary. Our effort is to advance the state-of-the-art in pattern
recognition of affect from video streams and physiology by proposing a large space of
reasonable features and systematically evaluating subsets of it and transformations
thereof. For health monitoring purpose, the recognition of human action is expected to
be classified into “normal walking,” “slow walking,” “falling down,” and
“crouching,” etc. As for the same purpose from facial expression, expression is
expected to be differentiated into “normal face,”“pale face,” “slightly pain face,” and
“pain face,” and the like. It is easy to obtain the monitored video streams in an
unconscious way and the classified categories for these video streams are described in
terms of symbolic form. Moreover, physiological signals including EEG, ECG, blood
pressure, heart and respiration rates, and skin conductivity, whose measured outputs
are described in numerical format, can be integrated to access the physical comfort of
a person. This health condition accuracy can be enhanced by fusing the recognized
one’s action, facial expression and physiological signals. Heath condition
determination and prediction methods will be relied on machine learning techniques,
which include two modules. The first module facilitates the classification algorithmic
routine and the second is constructed with a knowledge rule-based system to predict
one’s health condition [24]. Heath state determination through SVM, BP and
statistical approaches are appropriate and a better generation can be reached. ID3
technique is well known for prediction or summarization the status of a set of data. In
this study, we will investigate to extend the ID3 method to a more flexible fuzzy ID3
version [56], which can handle symbolic data, like painful state, and continuous
feature, such as blood pressure. Through the development of this automated health
care and monitoring system, the digital machine can constantly and automatically
detect and monitoring one’s health condition. A quick response from telemedicine
objective can be reached, which as a whole improve the quality and safety of our daily
life through human-centered and automated human-machine interface investigation.
NSC94-2213-E-009-097（94N530）
-------------------------------------------------------------------------------------------------------

Title: The systematic design and analysis of topologies for single-phase single-stage
AC/DC converters
Principle Investigator: Chang Lon-Kou
Sponsor: National Science Council
Keywords: Single-Phase Single-Stage AC/DC Converter, Input Current Shaping,
Soft-Switching, Energy Flow Path

To constraint the undesirable harmonic interference from the power converter
onto power line and other electronic equipment, stringent current harmonic
regulations have been defined by some international standards, such as IEC 61000-3-2.
Thus, the single-stage input current shaping (S2ICS) techniques have been proposed
and intensively studied recently, in order to comply these regulations with minimal
additional component count and cost.
The objective of this two-year project is to develop deep techniques and
innovative solutions for the advanced integrated S2ICS AC/DC converters, which
target at the low power applications (from 70 to 200 W), for power adapters, battery
chargers, and various communication equipments.
In the 1st year, this project will provide a systematic study of the S2ICS
mechanism, circuit topology generalization and modification, bulk capacitor
voltage-stress and switch current-stress, converter design and optimization, and the
evaluation of the S2ICS techniques applied to universal-line input. We will also
develop a novel S2ICS technique to improve the performance and achieve the
cost-effective consideration for universal-line applications. Another focus of this
project is to design and implement an efficient S2ICS converter based on
soft-switching techniques and the overall topology simplification.
In the 2nd year, the project will develop a systematic method for deriving S2ICS
converter configurations in which the charging and discharging paths of the ICS
inductor can be possibly unitized. Based on this method, many topologies known
previously can be analyzed to develop improved ones. The second systematic
approach is to analyze and evaluate the performance of S2ICS converters by the
concept of energy flow path. It provides a useful design guideline for the
efficiency-related S2ICS techniques.
All the proposed concepts or techniques will be analyzed and verified in practice,
and the results will be used as a reference to the realization on the future electronic
equipment.
NSC 94-2213-E-009-147（94N409）
------------------------------------------------------------------------------------------------

Title: Analysis of Direct-Drive in-Wheel Motor and the Design of Sensorless Sliding
Mode Controller (2/2)
Principle Investigator: Chen Yon-Ping
Sponsor: National Science Council
Keywords: In-wheel motor, Sensorless drive, Sliding mode controller

Traditionally, high-speed motor is adopted as the power source in an EV
(electric vehicle). It uses the transmission mechanism to deliver power to the shaft of
the wheels. Recently, the technology of Servo-motor has been well developed, which
can be applied in such procedure. However, the loss resulted from the transmission
mechanism could reduce the overall efficiency seriously. The objective of this project
is to analyze the in-wheel motor with FEM and to construct its analytical model. A
driving method based on the in-wheel motor’s dynamic model is then proposed to
improve the overall efficiency. Furthermore, sliding mode theory is employed for
controller design to suppress the external disturbance. Most significantly, the
sliding-mode control for low speed operation is implemented via the senseless
technology, which is achieved by using a sliding observer.
NSC 94-2213-E-009 -047（94N518）
-------------------------------------------------------------------------------------------------------

Title：The technology of constructing the indoor map for robot’s activities
Principal Investigator：Chen , Yon-Ping
Sponsor：Industrial Technology Research Institute
Keywords：Robot, Indoor Map Building, Navigation, Route Planning

This project will realize the robotic indoor navigation. As the robot wanders
inside the building, the information of environment measured by robot will be utilized
to build the indoor map. Then the robot can walk inside the building conveniently. In
this project, the development environment is based on C# and ActivMedia software
respectively for the indoor map, navigation, and positioning.
The robot in ITRI is adopted as the development platform. There are sonar arrays,
laser, odometer, and gyro mounted on the robot. The information measured by these
sensors could be integrated by intelligent theory. After extracting the useful features,
the surrounding environment of robot could be constructed. While the robot wanders
inside the building, the relative indoor map can be built well. As a result, the robot can
walk conveniently and know the environment.
After the indoor map is built up, we would like to research the indoor
navigation of robot. The user can select a point on the map and send the command to
the computer of the robot with WLAN. Then the robot would walk to the assigned
point. As shown in figure A, the robot would walk from room1 to room2. In this
process, the intelligent algorithm would be adopted to assist the navigation. For
instance, if there is a dynamic object in the room, the robot would detect the object
and keep out the way intelligently. Then, the robot would plan the route again and go
to the assigned point.
94C155 (94.06.01-94.11.30)
---------------------------------------------------------------------

Title：Smart power management for ultra low power systems
Principal Investigator：Chen Ke-Horng
Keywords: Power management

Wireless communication and potable devices are popular in today’s technology.
Low power designs become more and more important. Owing to the growth of low
power designs, we even can use the ambient energy sources for our system. As we
know, it is so-called green energy. It also means that this system is a self-sustention
operating system using green energy as energy source. Thus, how to convert and
manage this tiny energy becomes a major topic in this system’s design. Our research
aims at a smart DC-DC converter and adapts a suitable power management theory in
the self-sustention system. Based on this architecture, this self-sustention can get
stable supply energy under the optimal energy transferring efficiency.
The generator of microelectromechanical systems can generate tiny energy
about hundreds of μW. We can use rectifier and DC-DC converter efficiently convert
the unstable voltage to a specified and stable output voltage. According to the
industry’s latest round of DC-DC converters’ technology, we can use an
inductor-based switching regulator or a regulated charge pump converter. In this
self-sustention system, we need high conversion efficiency, low quiescent current,
high system integration, and low noise. Thus, a lot of systems with green sources use
an inductor-based switching regulator as the DC-DC converter architecture. In order
to reduce the disadvantage of higher noise, we can use some low-noise technologies
to reduce the influence of the effect. In other words, our research will be an
inductor-based switching regulator with ultra-low noise. Based on this architecture,
we can develop our power management for this self-sustention system.
This project proposes a smart power management theory. The feedback information
from the frequency deviation between the operating frequency of the digital signal
processing system in this self-sustention and modulation frequency, utilization of the
energy from MESM generator, the deviation of loading, and the capacity of the
storage device determine which of the modulation technology we should use. The
modulation skills contain PWM (pulse width modulation) and PFM (pulse frequency
modulation). According to the loading current, swap the modulation skill between
PWM and PFM. When the load is light, the system uses PFM. When the load is heavy,
the system uses PWM. It makes the efficiency of the system have the optimal value.
Using buck/boost converting skills, we can step-down or step-down the output voltage
to the range from 0.9V to 1.6V. Make sure the energy consumed in the system and the
energy stored in the storage devices to the best converting efficiency.

MEMS
Generator
Rectifier       Storage
Feedback
Terms
Power
Management
Drive and                  L
VOUT
Self-Sustention
0.9~1.6V        System
Control

Basic Buck/Boost DC-DC Converter

智慧型直流電源轉換器與功率管理架構示意圖

NSC 94-2215-E-009 -059（94N582）
-----------------------------------------------------------------------------------------------

Title：Designing the reader, and building the communication system between reader
and tag
Principal Investigator：Chen Fu-Chuang
Keywords: RFID, Reader, Tag

An RFID system is always made up of readers and tags. Tags are located on personnel or
objects, in order to receive the transmitted signal and energy from readers and return the
relevant information of tags. Besides being responsible to supply energy to tags, readers
are responsible for communication, data security, data reliability between readers and tags.
In addition, readers must deal with data collision problem. Readers are usually fitted with
an additional interface to connect with other systems, so that the data which was read can
be analyzed and processed further. Because, RFID systems have advantages, such as
conveniently, fast, saving time and diversification, etc., the fields of application for them
are wider and wider.
This total plan that our whole research groups want to carry out is: “The research and
development of RFID-based personnel and objects tracking and controlling systematic
platform”, has divided into five sub plans altogether. We are responsible for the sub plan
two “The research and manufacture of readers, and build the transmission system
between readers and tags”. In order to finish whole total plan, we fix the three-year goal.
Goal of the first year is “The plan of transmission system between readers and tags, and
design the system of readers”, we will select the best systematic parameters, transmission
methods according to the demand, and do the preliminary design to the system of readers.
Goal of the second year is “To finish the prototype of readers, and do the transmission
test with tags”, we will finish the prototype machine of reader, and do the transmission
test with the group which builds tags. Goal of the third year is “To connect readers with
network, and the integration and test of system”, we will integrate all systems developed
by sub plans and test systematic exactness.
Before the research and development of actual circuits of RFID systems, we must
understand whole architecture of RFID systems. We need to understand the behavior and
performance under different systematic parameters, so we can select systematic
parameters carefully to achieve the balance point between optimal performance and
actual applications; it is also the main goal of the first year. And up till now, our research
results of the first year have already made our systematic parameters according to the
standard of Auto-ID class 1, including adopting passive RFID system, HDX procedure,
electromagnetic backscatter coupling, the UHF frequency range 922~928MHz. And data
rate, data encoding and modulation are different on the basis of transmission direction.
According to relevant documents, we also standardize the communication interface
between readers and tags, the communication between readers and network and the skill
of anti-collision. In addition, we discuss some fundamental architecture of circuits,
including quartz oscillator, frequency synthesizer, modulator, power amplifier, directional
coupler, band pass filter, low noise amplifier and mixer. We will do more detailed
description in the report of the research results of the first year at present.
With the progressive goals of three years, cooperate with groups which are responsible
for different sub plan, come to finish the whole total plan step by step. We hope that
through the completion of plan, we can develop a perfect RFID system, and accumulate
relative knowledge, experience, and technology to promote the levels of relevant
NSC 94-2213-E-009-101（94N534）
------------------------------------------------------------------------------------------------

Title：Research on the Joint Design of Cascaded Subsystems with Applications in
Communications and Signal Processings
Principal Investigator：Cheng Mu-Huo
Keywords：Cascaded subsystems, Alternating coordinates minimization (ACM),
Generalized Benders decomposition (GBD), Synchronizers,
Interpolators, Equalizers

This project aims for two purposes: one is to present a unified formulation,
investigate differences between existing algorithms, and develop new algorithms for
the joint design of cascaded subsystems; the other is to apply these algorithms for new
applications in digital communication systems and signal processings. When
designing a system with a large number of parameters and complicated relations
between these parameters, one often decomposes the system into several cascaded
subsystems such that each subsystem can be designed separately. This approach yields
several advantages: the number of required design parameters is reduced and their
related functions are often simplified such that the design can be straightforward, the
design algorithm is easily realized, and the optimal design for each subsystem can be
usually obtained. The disadvantages, however, are that even the design for each
subsystem is optimum, the total solution often is not optimal because the joint
capacity is not explored. Hence, the performance is often degraded. The purpose of
the joint design for cascaded subsystems, therefore, is to develop algorithms for
designing each subsystem such that these algorithms are simple to realize as well as
able to explore the joint capacity for performance improvement.
The idea of the joint design of cascaded subsystems has been used in many
applications of signal processings, digital communication systems, and system
sciences. For example, a digital IIR system can be formulated as a joint design of an
all-zero subsystem cascaded with an all-pole subsystem. This approach has been
applied for designing IIR filters or channel model identification. Another design
example is the joint design of an interpolator with an equalizer in a digital
communication system. These designs, however, focuses only on solving the
considered problem, the adopted object function, therefore, varies with applications
and the developed algorithms are also different. These algorithms, in essence, can be
unified in the class of the joint design of cascaded subsystems. Therefore, the first
purpose of this project is to develop a unified formulation for describing the joint
design of cascaded subsystems, and to investigate the realization complexity and
convergence property of the developed algorithms under various objective functions.
We begin by investigating the two famous algorithms in optimization, the alternating
coordinates minimization (ACM) and the generalized Benders decomposition (GBD).
Then we shall study to develop the algorithms, prove the conditions for convergence,
and evaluate the convergence rate under various measures such as the 2-norm,
$\infty$-norm, the maximum likelihood function, or the bit-error rate (BER). Hence,
the joint design of cascaded subsystems and its associated algorithms with their
application scope and constraints are investigated at this stage. The unified view also
enables us to have a clear understanding of the joint design and its design algorithms.
The second purpose of this project is the application of the developed joint design
methods. For the application in communications, we focus on the joint design of the
interpolator and equalizer. We have successfully used the 2-norm as the object
function to design jointly a better interpolator and equalizer. We plan to investigate
further the developed algorithms under different object functions for the design of
this problem for improving the performance. For the application in signal processings,
we design the IIR filters as the joint design of two cascaded subsystems. Then the
filter performance is evaluated and the convergence properties of the design
algorithms are investigated. In summary, this project studies the general joint design
of cascaded subsystems, develops a unified formulation and new algorithms, and
applies them for designing better interpolators and equalizers in a digital
communication receivers and better digital IIR filters. Through this project, we expect
to understand further the fundamental properties of algorithms for the joint design of
subsystems and explore new applications.
NSC 94-2213-E-009-061（94N342）
------------------------------------------------------------------------------------------------

Title: The Study of Smart Micro-Sensor Devices
Principle Investigator: Chiou Jin-Chern
Sponsor: National Science Council
Keywords：MEMS Technology, Micro-sensors, Accelerometer, Optical Gas sensor

The third sub-project entitled “The study of the intelligent microsensors” of the
integrated project plays the role of designing and fabricating microsensors for the
ultra low power wireless sensing module. Within the next three years, the present
sub-project will use MEMS technology to design and fabricate an optical gas sensor
or a novel threshold accelerometer. The abstracts for the next three years are given as
follows:

Abstract of 1st year project:

There are two action items that need to be completed in the 1st year, at first, to
justify our design concepts; we will focus on the pro and con of the operation
principles in designing MEMS components by using simulation software. We will
also develop the needed processes and masks that can be used to fabricate different
components for the optical gas sensor and the threshold accelerometer. Secondly, in
order to verify the design sensing concept, we will utilize the existing
optical/electrical/mechanical components to assemble an optical gas sensor.
This device will be the standard for the future micro optical gas sensor.

Abstract of 2nd year project:

For the 2nd year, with the specifications given by the related sub-projects, we will
use the simulation software to verify and fabricate a series of low driving voltage and
low power consumption optical and accelerometer components. Furthermore, in order
to justify the performance of the fabricated components, we will replace the device of
the previous assembled system one at a time. By discussing with other sub-projects,
we will continue to modify our design and fabrication process.

Abstract of 3rd year project:

The work need to be done in the 3rd year is to integrate different chips fabricated
in 2nd year. Here flip-chip bonding technology will be utilized to complete the
assembly of the ultra low power wireless sensing module. Experiments will be
conducted to obtain system characteristics of the integrated module. If needed,
specifications and designs will be modified according to different requests.
NSC 94-2215-E-009-055（94N578）
-----------------------------------------------------------------------------------------------
Title: The design and fabrication of an ultra low power micro-sensing module for
wireless sensor networks
Principle Investigator: Chiou Jin-Chern
Sponsor: National Science Council
Keywords：Wireless sensor networks, System on Chip, MEMS, Microsensors, RF
Transceiver Circuit, Micro Power Generator, Ultra Low Power A/D
Converter, Mixed Mode Sensing Circuit, Power Manage Circuit

An advanced and highly integrated research project is proposed for the next three
years.The present integrated project consists of six sub-projects that are focused on
the research of Micro Power Generator, Power Manage Circuit, Microsensors, Mixed
Mode Sensing Circuit, Ultra Low Power A/D Converter, and RF Transceiver Circuit.
The main goal of this project isto develop an ultra low power wireless sensing module
with small size, high efficiency andlow price characteristics.
The abstract of the three years integration project is: (1) to achieve defined
performances and specifications of each sub-project. The research items include
designs, simulations and verifications of related devices and fabrication processes.
Encounter problems and solutions will be discussed and if necessary, we will modify
the designs and specifications. (2) to accomplish the goal of ultra low power and
miniaturization. The fabrication, testing and verification are performed to verify
original designs. An optimal integrated solution will be provided to obtain the ultra
low power wireless sensing module. (3) to integrate all the chips that are fabricated by
the sub-projects using the flip-chip bonding technology and successfully manufacture
the proposed ultra low power wireless sensing module. Successive tests will be
carried out and the final performance will be reported at the final stage of the project.
NSC 94-2215-E-009-056（94N579）
-------------------------------------------------------------------------------------------------------

Title：The Development of Packaging Process and Driving Circuits for High Density
Micro Capacitive Device
Principal Investigator：Chiou , Jin-Chern
Sponsor：Industrial Technology Research Institute
Keywords：Tunable CMUT Driving CKT, Capacitance sensing, XYZ-Table, MEMS
Packaging

A tunable CMUT driving and capacitance sensing CKT will be developed. A
XYZ-Table is designed and established to carry the CMUT system for constructed an
integrated measurement application. In addition, a MEMS packaging material and
process for CMUT will be researched and deliver a report for reference in future
works. In the CMUT driving CKT design, a high tunable DC gain is one of the key
design issues. In the others, the tunable frequency and amplitude of AC signal is
designed for satisfy the measurement application. The XYZ-Table with high
resolution servo control is used to precision control the gap between the CMUT and
object for improve the sensing performance. In MEMS packaging, the packaging
material is the main power consumption in CMUT system. A functional packaging
material with well transmission performance is not always compatible to CMUT
process. In a standard/positive packaging process, the passable packaging material is
limited. The material and process will be trade off in this research. Therefore, a new
packaging process will be researched and designed to complete the CMUT packaging.
94C116(94.05.01-94.11.30)
---------------------------------------------------------------------------------------------

Title：MEMS-based electrostatic vibration-to-electric energy converter (II)
Principal Investigator：Chiu Yi
Keywords：Power MEMS, Energy conversion, Vibration, Electrostatic, Smart micro
sensor system

In recent years, there are increasing interests in the power MEMS technology,
where various energy sources, such as chemical and kinetic, are converted to electric
energy using MEMS technology. Such micro power supplies can be applied to
portable 3C products, personal health monitoring system, and other distributed sensor
networks. This project extends the research effort from a 2004 NSC grant to
develop an electrostatic MEMS vibration-to-electric energy converter within 1 cm2
based on a 3.3 V auxiliary power supply. The targeted energy source is the 120 Hz,
2.5 m/s2 vibration that is measured abundant in natural environment. Design and
simulation environment have been set up. Optimal design parameters were found
through both theoretical analysis and numerical simulation. A power density of
2
can be achieved. Final layout is to be completed by the end of 2004 and
device fabrication can be engaged in early 2005. The electrostatic MEMS power
generator can be integrated thermo-electric and photovotaic micro power generators,
power management circuits, signal processing circuits, and sensing circuits to form a
self-sustained smart micro sensor module.
NSC 94-2215-E-009-057（94N580）
-------------------------------------------------------------------------------------------------------

Title：Low-Power, Low-Distortion Image Compression IC Design
Principal Investigator：Dung , Lan-Rong
Sponsor：Chung-Shan Institute Of Science And Technology
Keywords：Video compression, Digital signal processor, Power-aware architecture

The objective of this project is to develop a power-aware video compression
processor for capsule endoscopy or swallowable imaging capsules. In applications
of capsule endoscopy, it is imperative to consider battery life /performance trade-offs.
Applying state-of-the-art video compression techniques may significantly reduce the
image bit rate by their high compression ratio, but they all require intensive
computation and consume much power from battery. There are many fast
compression algorithms for reducing computation load; however, they may result in
distoration of original image, which is not good for the use of medical care. Thus,
this project will first simplify traditional video compression algorithms and propose a
scalable lossless compression algorithm. The scalable algorithm can be used for the
later power-stepping technique. Then, we will develop a power-aware architecture
for battery life extention. The power-aware architecture is an architecture that can
properly reduce the computation load as the battery status changed while the qulity
degradation is little. In the project, we will consider not only the minimization of
average power dissipation but transient characteristics of power dissipation, such as
peak power and power gradient or differential. Therefore, the project will develop a
low-computation, scalable video compression algorithm and its power-aware DSP
ASIC for capsule endoscopy.
94C076(94.01.01-94.12.31)
-----------------------------------------------------------------------------------------------------

Title：Study on Power-aware High-level Synthesis Techniques for Nenometer SOC
Design(3/3)
Principal Investigator：Dung Lan-Rong
Keywords：Power-aware system, High-level synthesis, SOC, CAD

As we get closer to the limits of scaling in CMOS circuits, it is imperative to
consider power/performance trade-offs and to develop appropriate power aware
methodologies and techniques for embedded systems. The use of nanometer
technologies is making it increasingly important to consider transient characteristics
of a circuit's power dissipation (e.g., peak power, and power gradient or differential)
in addition to its average power consumption. State-of-the-art transient power
analysis and reduction approaches are mainly at the transistor- and gate-levels.
However, we believe architectural solutions to transient power problems may
complement and significantly extend the scope of lower-level techniques, as was the
case with average power minimization. This project intends to exploit high-level
synthesis approach to transient power management and reduction in that a
power-aware high-level synthesis can impact the cycle-by-cycle peak power and peak
power differential for the synthesized implementation.
NSC 94-2220-E-009-023（94N467）
-----------------------------------------------------------------------------------------------

Title：Study on Formal Verification for High-Level Synthesis
Principal Investigator：Dung Lan-Rong
Keywords：SoC, Formal Verification, Theorem Proving, Design for Verification

The System-On-Chip (SOC) design encompasses a large design space. Typically, the
designer explores the possible architectures, selecting algorithms, choosing architectural elements, and
constructing candidate architectures. Designing such a complex system is hard; designing such a
system that will work correctly is even harder. Design errors should be removed as early as possible;
otherwise, errors detected at the later stages will result a costly, time-consuming redesign cycles. Thus,
the designer should face two distinct tasks in SOC design; carrying out design process itself and
establishing the correctness of a design. Design correctness is the main theme of this project. This
project aims on the formal verification for high-level synthesis and addresses on three issues: theorem
proving, property reasoning, and design for formal verification. First of all, the project will develop
theorem proving technique for architectural design. Based on Petri-Net models, the theorem proving
verifies the reachability, admissibility, and correctness of task scheduling and resource allocation. In
the property reasoning, the project will focus on metrics of property coverage. With accurate
calculation on property coverage, the notorious property-validation problem can be lessened or even
solved. Finally, a technique on design-for-formal-verification (DFFV) will be developed, driven by
theorem proving and property reasoning. The DFFV technique will aid SoC designers for efficiently
applying formal verification in proof of their design.
NSC94-2220-E-009-039（94N483）
------------------------------------------------------------------------------------------------
Title：Ultra-low power, design-for-digital Testability analog-to-digital converter(II)
Principal Investigator：Hong Hao-Chiao
Keywords: Low power design, Analog-to-digital converter (ADC),
Design-for-testability (DFT)

There are two major goals pursuing by electronic engineers, one is higher speed,
the other is lower power. In many applications such as the bio-sensing systems, the
speed is not a major concern but a longer operating time is very critical. For instance,
an animal tracking system wishes it could sustain several years without changing the
battery since it is hard to replace the battery. It is the case of our major project--- an
ultra low power micro-sensing module applying for wireless sensor networks.
The analog-to-digital converter (ADC) is one of the key components of the ultra
low power micro-sensing module. It is necessary for flexibility of complex signal
processing by digital circuits. It also determines the accuracy of the system and is
very vulnerable to the noises coming from the environment, chip, even the circuit
itself. In addition, since the power provided by the system is very limited, ultra-low
power is an essential requirement for the ADC. At system level, different ADC
architecture has different power requirement. We are currently working on detailed
analysis to find out the power consumption characteristic of the candidates in this
year’s NSC project. At circuit level, special design technique must be developed. One
possible solution is design the ADC by using MOSFETs operating at sub-threshold
region. However, the design techniques have not been well developed yet as far as we
know. Lots of research results may be found.
Another issue in designing the ADC is how to testing it. Traditional way using
an external signal generator to stimulate the ADC then get the digital outputs for final
DSP analysis. Its cost is high and the noisy testing environment often disturbs the
results. Usually, lots of efforts must be paid to debug the testing environment. In the
target micro-sensing module, how to test the ADC becomes more troublesome due to
its highly integration. Alternatively, design-for-testability (DfT) and/or analog
build-in-self-test (BIST) techniques help alleviating the efforts since it accepts digital
stimulus rather than analog one. It isolates any possible noise source coming from the
interface.
This project will develop an ADC with at least 8-bit resolution to be integrated
in the wireless sensor module. It operates at 1V and consumes less than 30 µW. The
target ADC will be equipped with a low cost DfT circuit and being totally digitally
testable to enhance its immunity of environmental noise and interference during
testing, to simplify the testing environment setup, and to reduce the testing cost.
NSC 94-2215-E-009-081（94N604）
-----------------------------------------------------------------------------------------------

Title：Development of the Control Technique for the Omnidirectional Mobil Platform
Principal Investigator： Hsu , Pau-Lo
Sponsor：Industrial Technology Research Institute
Keywords：Omnidirectional mobile platform, Control, Measurement.

To improve the agility of a mobile robot or platform, omnidirectional mobile robots
have been proven to be more capable in high-speed motion operations. Although the
omnidirecitonal wheel can be applied to motion systems to achieve the qualified task,
its motion precision is difficult to achieve. Therefore, this project will develop the
basic control technique for an omnidirectional mobile platform to meet the
specifications as: in a linear motion 180 cm, it will achieve 5 cm accuracy with every
60 degree.

Basically, when the omnidirectional motion system moves in a certain angle, the
slip will occur and it causes the axial force on the platform. Therefore, on-line
measurement, open-loop and feedback control of the omnidirectional system are
difficult compared with motion control in traditional systems. In this project, basic
control techniques for the omnidirectional system will be established through both
theoretical analysis and experimental work. In the future, the present results obtained
in this project will be integrated with other communication and image signal
processing systems under the main project.
94C156 (94.06.01-94.11.30)
------------------------------------------------------------------------------------------------------

Title: Multi-Agent Distributed Monitoring and Control of Traffic Systems
Principle Investigator: Professor Hsu Pau-Lo
Sponsor: National Science Council
Keyword: Traffic light control, Fuzzy decision, Agent, Multi-agent system,
Distributed control, Wireless network

Development on monitoring and control progress significantly in the past two
decades. However, the monitoring system is not compatible in realization with the
control system mainly because of their functions are different. Control needs real-time
computation while monitoring requires less real-time processing but with large
amount of intelligent computation in decision making. Due to the recent rapid
development in networking technology, agent system becomes a potential tool in
distributed monitoring and control systems.
For the traffic lights control issue, it is a standard monitoring system because its
control requires less calculation but how to determine the duration of the signals
requires large amount calculation to achieve an optimized result. In addition, it is a
practical issue in control problem because every intersection in the city is jammed
during the rush hours. In general, one or two polices are assigned to control the traffic
light manually. For a traffic network, it is a typical distributed control system.
However, due to the polices in different intersections do not exchange information;
the control results are not satisfactory. In addition, the traffic flow, speed, density,
queuing time are varied greatly in a day. It is not suitable to apply traditional
scheduling to analyze it.
In this two-year proposals, the fuzzy control as a police is designed for each
intersection. Furthermore, several adjacent intersections are connected by the 802.11b
wireless communication that the local traffic can be monitored by an intelligent agent
to achieve traffic balance. Furthermore, a multi-agent system (MAS) is developed so
that they can negotiate one another to get the most balanced traffic in the whole
system.
NSC 94-2213-E-009-031（94N514）
-------------------------------------------------------------------------------------------------------

Title: On The Study of Acoustic Field Perception and Sound Communication
Interface for Robotic Systems
Principle Investigator: Hu Jwu-Sheng, Cheng Chieh-Cheng, Liu Wei-Han, and Yang
Chia-Hsing
Sponsor: National Science Council
Keyword:Robotics, Hearing System, Array Microphone, Array Signal Processing,
Acoustics, Speech Enhancement

The research on sound was not emphasized in the area of Robotics when the
applications of robots are mainly on industrial automation. Most researches of the
environmental sensing for robots were based on vision, ultrasound, temperature or
touch. However, sound (or speech) is a very important communication mean for
human being. That is to say, if we wish to make robots that can enter household or
become a personal assistant, the sound interface has to be a necessary part of the
system. The communication of sound can be separated into two categories. The first
one is the content of the sound such as speech or music melody. The second one is the
information created by sound wave interaction with the environment. For example,
the interaction can reveal the sound source location or even the acoustic field that is
related to the space where the sound is generated. This is a 3-year proposal to study
the second category mentioned above. The topics include making a microphone array
module to facilitate the research experiment and integration on robots. Secondly, due
to the movement of robots, the microphone array on the robot has the property of
infinite sampling in space. We will study the advantage it brings as well as the
shortcomings. Meanwhile, we will also study the classical sound source separation
and speech enhancement problem as they are very important for the hearing ability of
a robot. This proposal is part of an integrated effort to make personal robots. And this
work will play the role of hearing systems for the robot as we will collaborate closely
with other research groups participated in this project.
NSC 94-2213-E-009-046（94N517）
-----------------------------------------------------------------------------------------------

Title：The Design and Implementation of Embedded M2M Technology
Principal Investigator：Hu , Jwu-Sheng
Keywords：Embedded Systems, Machine-to-Machine, Man-To-Machine,
Machine-to-Man

M2M refers to the communication between machine to machine, machine to
man and man to machine. M2M solution builds up a communication channel via
PSTN, LAN, WAN or WiFi to achieve the goal of instant communication for
enterprises and factories. The researches estimate that M2M solution will be applied
to the public device networks of trasportation, electricity measurement, weather
forcast and so on. It is forcasted to result in another revolution in network
technologies.
One of the key technologies in M2M solution is the integration of hetergeneous
networks. This technology helps to provide a stable solution for distributed
applications with low data rates. In addition, the technology also connects the
embedded devices in a factory to the public, large-scale information network. The
project provides a flexible framework and API for services providers to efficiently
integrate M2M devices and enterprise applications. This M2M solution also brings a
lot of potential oppertunities to create new business.
94C214 (94.12.01-95.11.30)
-------------------------------------------------------------------------------------------------------

Title：Analysis and Design of Automatic Power Monitoring on the Feeder of Power
Systems
Principal Investigator：Liaw , Der-Cherng
Keywords：

In the recent years, the control and monitoring of power systems has attracted lots of
attention in both academia and industry. According to the information from Tai-Power
Company, the company will install monitoring devices on more than 3000 feeder lines
of power systems for automatic power monitoring between the years of 2002 and
2007. Based on the average needs of 5 Feeder Terminal Units (FTU’s) required for
each feeder line, the requirements of the FTU’s and the controller: Feeder Remote
Terminal Unit (FRTU) for the whole project will be around 15000 FTU’s and 300
FRTU’s. There are only few companies on Taiwan can provide part of solution to fit
the requirements of Tai-Power Company. In this project, we will study the required
specifications for the whole power monitoring systems. A detailed survey of the
available commercial products of both FTU and FRTU will be carried out, which will
then provide the required hardware and software specifications of both FTU and
FRTU for product design. The specifications of the system software and the related
application packages software will also be studied to cover the whole design
requirements for fulfilling the function of automatic feeder power monitoring.
94C100 (94.05.01-95.04.30)
------------------------------------------------------------------------------------------------

Title：Blind Identification and Equalization of MIMO FIR Systems
Principal Investigator：Lin Ching-An
Keywords：Wireless Communication, Blind equalization, Blind identification,
MIMOsystem, FIR

The multiple-input multiple-output (MIMO) finite impulse response (FIR) models
arise in many applications, in particular, in describing multiuser communication
channels. The need for blind equalization of such channels arises when multiple
input sequences from multiple users need to be blindly separated and estimated. This
proposed research develop methods and algorithms for blind identification and
equalization of MIMO-FIR systems using second-order statistics. The methods
exploit cyclostationarity of output data, which is induced at the input via periodic
precoding modulation. Part of the research extends our results for the SISO case.
Topics of investigation includes identifiability conditions, identification algorithms,
optimal choices of precoding sequences, equalization based on optimal FIR periodic
deconvolution, effects and optimal choice of deconvolution delay, and recursive
adaptive implementation of the combined identification-equalization algorithms.
Results of this research will provide insight needed for a practical solution to the
MIMO-FIR blind equalization problem.
NSC94-2213-E-009-010-(94N305)
------------------------------------------------------------------------------------------------------

Title：Intelligent Systems in Ovarian Cancer with Microarray Data for Molecular
Evolution and Control
Principal Investigator：Lee Tsu-Tian
Keywords：Ovarian Cancer with Microarray, Intelligent Systems, Prediction and
Classification, Modeling of Gene Network, Clustering of interval
Regression Analysis, Gene Pathway Analysis, Multiple Sequence
Alignment

Ovarian cancer is one of the most common causes of death from a gynecologic
malignancy. The mechanism of transformation from normal cells to malignant cells
remains to be clarified. Microarray provides a tool to study the diseases by a
large-scaled approach. The information of microarray with intelligent system analysis
can assist the physicians on the diagnosis, prediction of prognosis, therapeutic
planning and also shed light on figuring out the pathways of tumor formation.
Computer assisted analysis and simulation of biochemical pathways can improve the
understanding of the structure and the dynamics of cell processes considerably. In this
research, we shall focus on the study of some key technologies of advanced intelligent
system in the ovarian cancer with microarray data for molecular evolution and control.
The group project consists of seven subprojects as:
1. Ovarian Cancer Microarray data preprocessing and analysis.
2. Classification and Prediction of Ovarian Cancer Using Microarray Data based
on Intelligent Systems.
3. Computational Intelligence Approaches on Dynamic Functional Clustering
Algorithm, Clusters of Interval Regression Analysis and Distance-Based
Outliers Detection of Microarray Data for Ovarian Cancer
4. Modeling and Stability of Gene Network based on Computational
Intelligence.
5. A Hybrid Algorithm to Multiple Sequence Alignment for Ovarian Cancer with
Microarray Data.
6. Genome-Wide Pathway Analysis and Visualization Using Gene Expression
Data.
7. Establish the Molecular Evolution Mechanism of Ovarian Cancer via
Computational Analysis and Ex Vivo Test.
In this three-years group project, the effectiveness of the developed subsystems shall
be verified by the simulated and experimental results in the intelligent systems in
ovarian cancer with Microarray data for molecular evolution and control.
NSC 94-2213-E-009-123-（94N539）
-----------------------------------------------------------------------------------------------

Title：Modeling and Stability of Gene Network Based on Computational Intelligence
Principal Investigator：Lee Tsu-Tian
Keywords：Computational intelligence, Ovarian cancer microarray data, Intelligent
Systems, Modeling of gene networks, Molecular evolution and control

For this project, we propose several computational intelligence approaches such
as hybrid genetic-programming-fuzzy-neural, hybrid function Petri nets, Bayesian
neural networks, and graph theory, respectively, for modeling and analysis of ovarian
cancer microarray data via gene networks. Modeling methods of gene networks as a
set of nonlinear differential equations, and as non-differential equations are proposed
independently. Furthermore, the globally exponential and the global asymptotic
stability analyses to the gene network will be investigated. That is, in this project, we
will put forward and develop some novel and innovative approaches and tools for
molecular evolution and control of ovarian cancer with microarray data using gene
networks. Advanced software packages like Matlab will be used to develop computer
programs for gene networks and visual software for bioinformatics.
NSC 94-2213-E-009-124-（94N540）
-----------------------------------------------------------------------------------------------

Title：Advanced Vehicle Control and Safety Systems-Design and Simulation
Principal Investigator：Lee Tsu-Tian
Keywords：Advanced Vehicles, Intelligent Control, Active Suspension, Anti-lock
Braking System, Real-Time Foreground Image Recognition System,
Car-Following System, Distributed Network Monitoring and Control,
Simulator

This is the group project of “Advanced Vehicle Control and Safety
Systems-Design and Simulation”. Due to ever increasing traffic on the roads, the
tendency that the drivers face tough, complicated situations is increasing. Therefore,
built-in driver support systems to assist the drivers in hazardous situations are of
utmost importance. In this research, we shall focus on the study of some key
technologies of advanced vehicle control and safety systems. The group project
consists of six subprojects as:
1. Reliable Controller Design for Intelligent Active Suspension Systems of
2. Dynamic Optimal Slip Ratio Measurement and Intelligent Control for
Anti-lock Braking Systems to Track Dynamic Optimal Slip Ratio
3. A Study and Implementation of Advanced Vehicle Real-Time Foreground
Image Recognition System Based on Moving Object Image Detection
Technology
4. High-reliable Car-Following Control System Design
5. Distributed Network Monitoring and Control for Intelligent Automobile
6. Design and Implementation of a Vehicle Behavior's Simulator Using a
Six-Degree Platform
In this three year group project, the effectiveness of the developed subsystems shall be
verified by the simulated and experimental results in the real car and simulation car.
NSC 94-2213-E-009-125-（94N541）
-----------------------------------------------------------------------------------------------

Title：High-reliable Car-Following Control System Design(1/2)
Principal Investigator：Lee Tsu-Tian
Keywords：Car-following , Fuzzy theory, Neural network, Time-delay, Fault
diagnosis, Reconfiguration
Transportation technology is one of the most influential areas on the human life.
Therefore, researchers have been involving in wide scope of related research activities
aiming to enhance efficiency, comfort, and safety of transportation systems. Due to
the ever growing number of vehicles on the roads, urban highways are congested and
need additional capacity. Though building new roadways sounds like the prominent
solution at the first glance, in practice it causes number of other unsolvable problems
mainly due to scarce of suitable lands. To this end, to increase the traffic capacity by
automating the traffic flow has been identified as a smarter option. Moreover, due to
ever increasing traffic on the roads, the tendency that the drivers face tough,
complicated situations is increasing. Safety issues due to driver weariness when
driving for long hours is another concern. Therefore, built-in driver support systems to
assist the drivers in hazardous situations are of utmost importance.
An automated vehicle following control system that enables the vehicle to
control the engine torque by its own keeping a prescribed safe distance from the
preceding vehicle is proposed. With the inclusion of complementary driver support
systems, the overall control system can mitigate driver’s work load and guarantee
much improved safety. Such a car-following collision prevention system usually
includes radar and vision sensors, micro-processors, antenna, and digital signal
processors (DSPs). This project will last three years.
In the first year, we will consider the problem of longitudinal control of a
platoon of automotive vehicles. We will utilize a fuzzy neural networks based
approach to solve intelligent automated car-following control problem. The control
approach will not use any communication links between the vehicles in the platoon.
An on-line tuning algorithm is derived in the Lyapunov sense and the stability will be
guaranteed even in the face of various road conditions.
Since the automated car-following control problem accounts for lots of efforts
that involve sensor system, antenna hardware programming and design associated
with microprocessors and DSPs, in the second year, we will continue our focus
mainly on the intelligent car-following control problem. Moreover, the problems of
lane changing, merging, and leaving the platoon will be addressed. A novel intelligent
automated car-following scheme using recurrent fuzzy neural networks will be
developed to cope with the inherent time delay problem caused by communication
and actuator delays.
System safety and reliability are the other most demanding issues to be
addressed. In the third year, the purpose of this project will be to develop systematic
design methodologies in order to achieve the designed systems to simultaneously
deliver desired performance, fault detection ability, and recognition capability.
Extensive simulations of automatic steering and throttle/brake performances will be
carried out to demonstrate the effectiveness of the developed tools and controllers.
Since the control algorithms would be model-free techniques and are tuned on-line
using Lyapunov-based adaptation laws, not only the control accuracy but also the
system stability can be guaranteed.
The wide range of applications of the developed technology will also include
flight control, power systems, nuclear power industry, process control, and
transportation systems.
NSC 94-2213-E-009-126-（94N542）
-----------------------------------------------------------------------------------------------

Title: Study on Networked Control Systems
Principle Investigator: Lee Tsu-Tian
Sponsor: National Science Council
Keywords: Networked Control Systems, Stability, H  control, Robust control

By a networked control system (NCS) we mean that a system with a control
loop being closed via the serial communication channel. NCS is widely used in many
complicated control systems, such as manufacturing plants, vehicles, aircraft, and
spacecraft. In NCS, serial communication networks are employed to exchange
information and control signals between spatially distributed system components,
like supervisory computer, controllers, and intelligent input-output devices (e.g.
smart sensors and actuators). The serial communication channel, which multiplexes
signals from the sensors to the controllers and/or from the controller to the actuators,
serves many other uses beside control. Therefore, we always want to design NCS
controllers with minimal network usage to avoid collisions. The stability analysis
and controller synthesis problems for NCSs are more difficult than the traditional
point-to-point connected control systems. This is because for NCSs, there may have
the problems of un-uniform sampling, network induced delay, packet loss, quantized
error, and finite transmission rate.
In this project, we will study some related problems of NCS. This project
focuses on the problem of reducing the data transmission rate in NCS. Three related
topics are studied. The first topic is scheduling-based networked control systems
design. We will study how network scheduling affects the stability and H 
property of a NCS. Then we will find a method to design a scheduling-based NCS
controller such that it will stabilize the closed-loop system (and achieve the H 
requirement) with minimal network usage. The second topic is
transmission-if-necessary-based networked control systems design. We will study
how to design a transmission-if-necessary-based NCS controller such that it will
stabilize the closed-loop system (and achieve the H  requirement) with minimal
network usage, and related problems. The third topic is model-based networked
control systems design. We will study how to design a model-based NCS controller
such that it will stabilize the closed-loop system (and achieve the H  requirement)
with minimal network usage, and related problems. All three topics we will take the
uncertainty into consideration.
NSC 94-2213-E-009-150-（94N546）
------------------------------------------------------------------------------------------------------

Title：Study of Satellite’s Attitude Control Using Variable Structure Control Approach
Principal Investigator：Liang Yew-Wen
Keywords：Reliable control, Active and passive reliable control, Fault detection and
diagnosis, Spacecraft attitude control, Model uncertainties and external
disturbances

Due to the growing demands of system reliability in large-scale or complex
systems such as those in aerospace engineering, nuclear engineering and industrial
process, the study of reliable control has recently attracted considerable attention.
Among these systems, the attitude reliable control issue for satellite is also highly
important and attracted lots of attention since it is very expensive to retrieve a satellite
from space for repair when it malfunctions. In this project, we will study the attitude
reliable control of a satellite via the Variable Structure Reliable Control technique
(VSRC). The reason for employing the VSRC scheme is that it has the advantages of
fast response and small sensitivity to system parameter uncertainties and disturbance.
In addition, the VSRC scheme needs not the solution of Hamilton-Jacobi (HJ)
equation which is essential in implementing the reliable controller through the optimal
approaches such as H  and LQR reliable designs. This project is about to study the
following subjects:
 Set up the dynamical model of a satellite with input redundancy.
 Design the passive attitude reliable control law for the satellite.
 Develop a real-time and on-line fault detection and diagnosis mechanism.
 Use the information of fault detection and diagnosis to design the active
reliable law.
 Create the LQR or H  reliable law for satellite control.
 Use Taylor formula to derive the numerical solution of the HJ equation.
 Compare the performances among the non-reliable and the three reliable
schemes.
 Study reliable performances when system has model uncertainties and/or
disturbances.
NSC 94-2213-E-009-094（94N527）
-----------------------------------------------------------------------------------------------

Title：A study of intelligent design scheme for anti-missile guidance (II)
Principal Investigator：Liang Yew-Wen and Liaw Der-Cherng
Keywords： Orbital trajectory estimation, Takagi-Sugeno fuzzy model, Variable
structure guidance law

According to various reliable reports, the Mainland China is continuously
strengthening their military force in the recent years, especially in the development of
ballistic missile. The increasing investment of China in the force inevitably results in
more and more threat to Taiwan. Meanwhile, because of the distance between Taiwan
and Mainland China is not long, it means that we do not have much time to react once
Taiwan is suffering a ballistic missile attack from Mainland China. Therefore, it is
important and urgent for Taiwan to develop an effective way to fast predict the orbital
trajectory of a ballistic missile, and to guide our interception missile to the place, at a
right time and right direction, where the ballistic missile is predicted to be. In this
project, we will develop the technique of fast orbital trajectory estimation of a ballistic
missile, and study the design of reliable interception guidance law for an anti-missile.
Details are as follows:
 Predict the orbital trajectory of a ballistic missile.
 Analyze and design the interception law for an anti-ballistic missile according to
the prediction of the ballistic missile’s trajectory.
 Study the applicability of the Takagi-Sugeno fuzzy model to the estimation of
ballistic missile’s trajectory.
 Study the applicability of the Takagi-Sugeno fuzzy model to the design of the
interception guidance law.
 Study the applicability of the variable structure control technique to the design of
the interception guidance law
NSC 94-2623-7-009-005（94N009）
-------------------------------------------------------------------------------------------------------

Title：Development of An RFID-Based Platform for Human / Object Surveillance
(2/3)
Principal Investigator：Liaw Der-Cherng
Keywords：RFID, Database, Networking, Monitoring, Tracking

In the recent years, the study of auto-identification (AUTO-ID) or radio
frequency identification (RFID) in industrial or commercial applications has attracted
lots of attention. Several preliminary designs have been implemented in the
supermarket or the gate of the buildings for thief-prevention and security checking.
However, most of these designs may not be reliable and were implemented under
case-by-case study. In order to improve system reliability and provide a systematic
approach of design, in this three-year project we propose to add on an image
processing system and database system for human and object surveillance. The
project is integrated by five sub-projects, namely, the development of image
recognition technique, the development of RFID reader, the application of networking
in RFID application, the development of decision-making strategy and design of
database, and the development of RFID tag. Through the mutual cooperation between
sub-projects, it is expected to achieve the tasks of passive object surveillance, active
object surveillance, active object tracking, and system reliable strategy setup. In the
first year, we will focus on the preliminary concept design of RFID components,
image tracking model and system platform. It is followed by an experimental
verification of system design conducted on laboratory member and equipment
surveillance by using existing off-the-shelf RFID components. Moreover, the circuit
verification of the self-design RFID tag and reader will also be one of the main tasks
in the second year’s study. An overall system test and performance verification will be
carried on in the third year of project, which is proposed to conduct on a floor of
engineering building for human/object surveillance.
NSC 94-2213-E-009-099（94N532）
-----------------------------------------------------------------------------------------------

Title：Development of Decision-Making Strategy and Design of Database (2/3)
Principal Investigator：Liaw Der-Cherng
Keywords：RFID, Database, Networking, Monitoring, Tracking

The techniques of auto-identification (AUTO-ID) or radio frequency
identification (RFID) have been recently applied to the security checking in
supermarket and/or the gate of the buildings for thief-prevention and security
monitoring. However, most of these designs may not be reliable and were
implemented under case-by-case study. In order to fulfill the performance
requirements for system reliability, in this three-year sub-project we will work on the
construction of a networking based database and the development of cooperative
and/or reliable strategy for human/object surveillance. In the first year, we will
focus on the preliminary design of image-plus-RFID based distributive
monitoring platform and the study of reliable scheme for digital monitoring
system. Moreover, we will also study the biologically inspired cooperative
scheme for possible application to object surveillance problem. The proposed
design will be experimentally verified in the second year of project by using
off-the-shelf RFID components. This verification will be conducted on the
surveillance of laboratory’s members and equipments with the help from other
four sub-projects. An overall system performance verification of the enhanced
design will be carried on in the third year of project, which is proposed to
conduct on a floor of engineering building for human/object surveillance.
NSC 94-2213-E-009-102（94N535）
-----------------------------------------------------------------------------------------------

Title: A Study of the Simulation Optimization Algorithm and Software
Implementation (2/2)
Principle Investigator: Lin Shin-Yeu
Sponsor: National Science Council
Keyword:

Simulation optimization is one of the most frontier research areas in optimization.
The main characteristic of simulation optimization problem is the evaluation of the
objective function of an input-variable setting requires lengthy simulation. Therefore
we cannot use the conventional optimization techniques to solve them. There are
various simulation optimization problems such as stochastic optimization problems
with huge input-variable space and large scale optimization problems with decision
and discrete control variables.
In this project,we intend to categorize some classes of simulation optimization
problems and propose algorithms to solve them. Basically, in the simulation
optimization problem ,it is almost impossible to extract structural information of the
system analytically. Therefore, the proposed simulation optimization algorithms will
use simulations as a tool to extract the structural information. The extracted structural
information will be used in the proposed algorithm to reduce the searching space.
Such an iterative simulation optimization technique will use only reasonable
computation time to obtain a good enough solution.
We will use ordinal optimization theory to prove the quality of the solution we
obtain. In addition, we will compare our results with those obtained by the competing
methods such as the genetic algorithms,simulated annealing,and the tabu search
methods.
We will implement our algorithms in the form of commercial software for a more
general purpose.
NSC 94-2213-E-009-044-（94N515）
-----------------------------------------------------------------------------------------------

Title：A study of Integration of the RFID system and Image Processing
Principal Investigator：Lin Sheng-Fuu
Keywords：Radio Frequency Identification (RFID), Tag, Receiver, Facial recognition

Due to the demand of information transmission efficiency, it is a very important
factor nowadays to acquire the information. Automatic identification procedure is
used to apply to people, animals, goods, or industrial products during transmission to
handle information efficiently. The RFID (radio frequency identification) system, as a
kind of automatic identification system, is designed to transfer energy and data
without contact. The transceiver can receive tags’ data simultaneously, even if there
are many RF tags to be detected.
Because the demand of tracking and controlling some specific people or objects,
we propose to use image processing technique here to add human face recognition
system to RFID system. The information of locating and tracking people can be
obtained by adding facial features in face database. Therefore, the RFID system will
have more powerful tracking and recognizing capabilities if the face recognition
technology is integrated with RFID system.
Till now, we have completed the image capturing and pre-processing, finding
configuration of moving object, finding facial image, and post-processing. In the next
year, we will focus on facial recognition, connecting and communicating with main
system, completing the rough system and experiments, comparing with existed
systems. And in the year after the next year, we will concentrate on optimizing the
system and the network, raising the recognition rate and the efficiency, and
completing system integrating and large-scale experiments.
NSC 94-2213-E-009-100-（94N533）
-----------------------------------------------------------------------------------------------

Title: Design and Application of Physiological Signal-Based Human-Machine
Interface—A Novel Approach to Control Machine
Principle Investigator: Lin Chin-Teng
Sponsor: National Science Council
Keywords: Physiological Signal, Human-Machine Interface, Control Application.

Since the EEG was first described by Hans Berger in 1929, there have been
continuing efforts in developing new communication and control schemes, which do
not utilize the normal neuromuscular pathway. Today, along with the development in
the areas of electrical engineering, computer science, and biomedical engineering, it is
no longer a dream to use the physiological signals for control applications. This
multi-disciplined research has received much attention in the international community,
while it just started in Taiwan. In this three-year integration project, we propose to
develop a novel human-machine interface based on the physiological signals, and
intend to apply it for the control of computer, motion simulation platform, robot arm,
and others. As most of the current systems are still limited to be used within the
laboratory, our goal is to develop an interface that can be used in daily lives and also
industrial applications.
During the development of this human machine interface, we first need to have
measurement systems for physiological signals. After the stable, reliable signals have
been acquired, we will then perform signal analysis and interpretation to extract
suitable features. These features are further translated into proper motion commands
according to the given applications. Based on the descriptions above, six sub-projects
are proposed. Sub-project 1 develops a mental workload detection system based on
the virtual-reality dynamic motion simulation platform. It will demonstrate the
feasibility of detecting and modeling, in near real time, via multiple streams of
psycho-physiological information that organize operators’ cognitive states and
responses to task events. Based on physiological signal and facial expression,
sub-project 2 will develop a system to determine and judge the physical and metal
comfort so as to predict the health condition of a person. Sub-project 3 will develop
bioelectric source modeling, estimation, and analysis techniques for brain activation
imaging. These techniques can probe the temporal waveforms of the bioelectric brain
activities that are related to the control thoughts or specific events. Sub-project 4 will
design and manufacture a portable noninvasive measurement system for the
cardiovascular signals. It will be applied to study how the central and automatic
nervous systems can be related to the human machine interface. Sub-project 5 will
first analyze the relationship among the EEG, EMG, and resultant arm movement, and
then establish a model accordingly. The developed system will be used to control
human arm and also robot arm. Sub-project 6 will use optoelectronics and MEMS
technologies to design and manufacture a MEMS based wavelength tunable
multi-wave chips for physiology signals measurement system. With this measurement
system, different and adapted human physiology signals can be used to obtain in
depth physiology information.
The integration project will combine those systems developed in each
sub-project to form a multi-function physiological-signal-based human machine
interface. It will then be used to execute experiments based on the computer, motion
simulation platform, robot arm, and so on. All the sub-projects will join force to
establish the performance evaluation criteria for yielding the same standard in system
development. The setup of the criteria will consider the viewpoints from both biology
and engineering, and also those from the international research communities. This
project is expected to be a successful integration between the fields of biology and
engineering. Meanwhile, it also provides a platform for mutual cooperation and
learning, thus leading to wide extension of their capabilities and applications in both
fields.
NSC 94-2213-E-009-095-（94N528）
-----------------------------------------------------------------------------------------------

Title: Mental Workload Detection in Human-Machine Interfaces Based on VR-based
Dynamic Motion Platforms
Principal Investigator: Lin Chin-Teng
Sponsor: National Science Council
Keywords:Mental workload, Virtual-reality, Physiological signal, Kinesthetic stimuli,
Portable real-time embedded system

In this project, we shall develop a Mental Workload (MWL) detection system
based on the Virtual-Reality (VR) dynamic motion simulation platform for the
next-generation physiological-signal-based human-machine interface. The goals of
this project are to demonstrate the feasibility of detecting and modeling, in near real
time via multiple streams of psychophysiological information that organize operators’
cognitive states and responses to task events. We will also develop and demonstrate a
human-computer interface that can assist operators working in an interactive
monitoring or command and control environment in maintaining a high sustained
cognitive capacity while minimizing performance lapses and errors of interpretation.
In order to apply the developed MWL detection system outside the psychophysiology
laboratory, we shall develop a portable real-time embedded system as the platform of
the next-generation physiological-signal-based human-machine interface.

During the following three years, we will focus our research on three major
topics including (1) Setting up of the VR-based physiological signal measurement
system to collect the physiological signals of the car drivers in various VR driving
environments. This system consists of a VR-car driving simulator to provide
kinesthetic/visual/auditory stimulus and some physiological signal measurement
equipments to record various physiological signals of a driver, simultaneously. We
shall also develop the approach for analyzing the physiological/mental characteristics
under different MWL. (2) Developing mental workload detection technology to
study human performance fluctuations, cognitive-state changes (e.g. drowsiness,
attention, etc) and accompanying correct, incorrect and absent motor responses in
attention-demanding cognitive tasks involving kinesthetic/visual/auditory stimulus
interpretation and decision-making. We shall also study the physiological signals
correlate of perceiving kinesthetic stimuli and construct the dynamic mental workload
model for MWL detection. (3) Designing and developing a portable real-time
embedded system. In order to apply the developed MWL detection system outside
the psychophysiology laboratory, we shall develop a portable real-time embedded
system as the platform of the physiological-signal-based human-machine interface.
This system should have low power-consumption (for portability) and high
computational capability (to process physiological signals). The MWL detection
technology developed in this project will be implemented into this portable real-time
embedded system and we shall continue to perform the final evaluation on the
real-time embedded system.
NSC 94-2213-E-009-096-（94N529）
-----------------------------------------------------------------------------------------------

Title：Advanced Bio-inspired Computer Perception Algorithms/SoC Development
and Their Real-life Applications.
Principal Investigator：Lin Chin-Teng
Keywords：Bio-inspired engineering, Computer perceptual technology,
System-on-a-chip (SOC),Bio-inspired computer visual and hearing
perception,Bio-inspired perceptual neural networks, Home surveillance
system

Living organisms are complex systems exhibiting a range of desirable
characteristics, such as evolution, adaptation, perception and even high-level learning,
thinking and reasoning, that are difficult to realize using traditional engineering
methodologies. Recently, researchers have been allured to devote to study
bio-inspired engineering intending to develop advanced computer perceptual and
intelligent technology. The goal of this project is to construct a complete intelligent
computer perceptual technology and to implement this technology into a
system-on-a-chip (SOC) design based on our research experience on the development
of bio-inspired human perceptual technology in the past few years. In the following
three years, we shall apply the information engineering technology to realizing the
bio-inspired computer visual and hearing perception and combine the artificial
intelligent technology to the perspective developing system to make it more
humanistic and intelligent with self-constructing and learning abilities. Five major
research topics will be performed in this project: (1) Development of the bio-inspired
computer perceptual signal processing technology. (2) Development of bio-inspired
computer visual system. (3) Development of bio-inspired computer hearing system. (4)
Development and design of SOCs for bio-inspired perceptual neural network
hardware implementation. (5) Applying the bio-inspired computer perceptual
technology into the home surveillance system as an application example.
Topic 1 of this project combines human high-level logical thinking and reasoning
mechanism as well as low-level neural network training strategy to develop a series of
fuzzy neural networks (FNNs) as the kernel technology of other research topics for
computer perceptual signal processing. The FNNs also integrates the computer
perceptual signals, performs decision making, and can make the decision system more
humanistic and intelligent with self-constructing and learning abilities. Since vision
and hearing are two most important subsystem of bio perceptual system, Topics 2 and
3 shall firstly study the human visual and hearing subsystems and then develop the
bio-inspired computer visual and hearing processing technology. Topic 4 of this
project aims at designing and realizing the analog and digital circuits of various
bio-inspired computer reasoning strategy and perceptual signal processing technology
developed in Topics 1~3. Development of complete SOC hardware/software system
to integrate the developed multi-level bio-inspired computer perceptual technology is
also included in this topic. Topic 5 of this project is to integrate the bio-inspired
technology developed in Topics 1~4 into a home surveillance system as an application
example and the demonstration platform.
NSC 94-2213-E-009-149（94N545）
-----------------------------------------------------------------------------------------------

Title： Wavelet tree based Digital Watermarking of copyright protection
Principal Investigator：Lin Yuan-pei
Sponsor： National Science Council
Keywords: Digital watermarking, Copyright protection, Tree marking

There has been great interest in embedding watermarks in digital data, e.g.,
image, and audio, for copyright protection, image authentication, proof of
ownership, etc. A reliable watermarking system will effectively discourage the
spreading of illegal copies and accelerate the exchange of digital data.           In this
research project, we propose a wavelet based watermarking scheme for the
application of copyright protection. Ownership can be proved by extracting the
watermark embedded in the host image and illegal copies can be traced. For the
extracted watermark to be more
useful as a proof, the proposed scheme will be a blind watermarking system (original
image is not used in the watermark extraction process). Secret keys will be used in
watermark encoding so that the watermark can not be extracted or removed without
keys. The watermark bits will be embedded in the wavelet domain based on wavelet
trees. As wavelet trees correspond to various frequency bands, the watermarks will
be potentially more robust to frequency based attacks. Also the information of the
watermark bit is spread throughout large spatial regions. As a result, the watermarking
technique is robust to attacks in both frequency and time domains. We will evaluate
the proposed watermarking technique by applying nongeometric as well as geometric
attacks. The results from this research plan will have practical applications in the
development of watermarking systems for images.
NSC 94-2213-E-009-038（94N331）
-----------------------------------------------------------------------------------------------

Title: System-on-Chip of the Driver and Control of a PM Brushless Sensorless
Motor
Principle Investigator: Lin Shir-Kuan
Sponsor: National Science Council
Keywords: Brushless DC motor, Sensorless control, IC design

This project tries to design an IC which implements the sensorless drive and
sensorless control algorithm for a permanent magnet brushless 3-phase motor
(PMBM). There have been several methods and patents in the sensorless drives of the
PMBM. The difficulty is not to use these existing methods, but lies in the
development of new technology to avoid the IP violation. The sensorless drive for a
PMBM can be divided into 4 steps: 1) position detection; 2) open-loop start-up
algorithm; 3) closed-loop sensorless drive; 4) closed-loop sensorless speed control.
This project will develop several new techniques for the first three steps, which will
also be applied for patents. The IC design of such a new sensorless drive for a PMBM
will be worked on an Altera Nios development system. The test of the resulting IC
prototype will also be conducted in this project.
NSC 94-2213-E-009-103（94N536）
-------------------------------------------------------------------------------------------------------

Title: Investigation of the Effect of Zen Meditation based on Bio-Resonance Sound
Diagnostic Mechanism
Principle Investigator: Lo Pei-Chen
Sponsor: National Science Council
Keywords: Zen meditation, Complementary and Alternative Medicine (CAM),
Bioenergetic Medicine, Intrinsic Data Field (IDF),
Electroencephalograph (EEG),Electrocardiograph (ECG), Visual Evoked
Potential (VEP), Intrinsic Data Field nalyzer (IDFA), Bio-Sound
Diagnostics, Quantum medicine

Due to the therapeutic effectiveness, the new area CAM (complementary and
alternative medicine) has drawn the attention of researchers and medical professionals
in the past decades. Researches in biomedical engineering and life sciences should lay
more stress on promoting the human health, in both the physiological and mental
aspects. As a matter of fact, scientists of the West have been reporting substantial
findings of the effectiveness of meditation practice in CAM not only on improving the
physiological and mental health but on treating a number of diseases. Accordingly,
the investigator has been devoted to the study of Zen-Buddhist meditation for the past
years. We investigate, from the viewpoint of biomedical engineering, phenomena of
the human life system under the orthodox Zen meditation practice. We study the
time-varying characteristics and dynamic mechanism during meditation course in
order to further establish the correlation among different electrophysiological signals
and parameters.
At the end of the twentieth century, life science and clinical medicine have begun
striding into the new, insubstantial realm of research --- bioenergetic medicine. It
states that all the living beings and the minerals consist of energy expressed in the
form of various frequencies, called the information field. Our body is the
simultaneous, integrating function of numerous information fields. The
acknowledgement of information field involved in everything in the universe makes
scientists in the pioneering research begin steering a new study, that is, the
exploration of Intrinsic Data Field (IDF). According to the theory of IDF analysis,
diseases may be re-defined as the weakness or deficiency of certain IDF frequency. In
addition, IDF may provide the following indexes of the human life system: spirituality,
creativity, state of mood, stress, stress-counter ability, organ functions, etc.
During the past five years, the principal investigator has been actively engaged in
the research study on the Zen-meditation physiological signals, with the focus mainly
on EEG (electroencephalograph), ECG (electrocardiograph), VEP (visual evoked
potential), etc. Although we have achieved some results of significance, data provided
by these mainstream medical instruments are limited to the physical quantities
reflecting the global variation of electrical potentials. In this research proposal, we
thus hope to inspect the physiological, spiritual, mood, …, or even conscious and
super-conscious states based on the IDF concept of bioenergetic medicine. Two CAM
instruments, ARDK quantum instrument and bio-sound diagnostics, will be applied to
the subjects for collecting references. Significance of this research work is at its
prospect of providing scientific evidence and academic interpretation for the human
life characteristics under Zen meditation as well as the mechanism of health
promotion by the Zen practice.
NSC 94-2213-E-009-136（94N398）
-------------------------------------------------------------------------------------------------------

Title：Design and Applications of Agent-Based Interactive Control for Robotic
Systems
Principal Investigator：Song Kai-Tai
Keywords：Intelligent Agents, Multi-Agent Systems, Robot Control, Sound Direction
Recognition, Personal Robots

In recent years, the research in robotic systems has progressed to a new era, i.e.
robots that come into human life. To enable robots merge into human life, one needs
to develop a reliable, safe and easy to use robot command interface. This interface
must provide a real-time response of the robot in order to guarantee safety as well as
actual functionality. Robots that can learn and adapted to personal inclination of their
user are highly expected. This project involves the development of an agent-based
control system, which fulfills the requirement of the above-mentioned characteristics.
The purpose of this approach is to build an open, structured modulized and
standardized platform for robot control development. Moreover, several interesting
demonstrations such as human robot interaction control, real-time image tracking will
be carried out in order to evaluate the performance of the developed architecture. We
plan to complete this project in three years. In the first year, the main objective is to
develop a platform for agent-based hybrid control architecture. The second year will
forms on the design and implementation of the personal assistant robot. The third year
mainly concerns the system integration of the personal assistant robot and the
reification of the multi-agent architecture. Based on the results of previous project on
advanced home robot, we feel confident to complete the goals of this project in three
years.
NSC 94-2213-E-009 -045 (94N516)
-------------------------------------------------------------------------------------------------------

Title: Intelligent Mixed Signal Sensor Circuit Design
Principle Investigator: Su Chan-Chin
Sponsor: National Science Council
Key words: Sensor Circuit, Analog Circuit, Mixed Signal Circuit, SoC, Bio Sensor

Subproject 4：Smart Mixed Signal Sensor Circuit Design
As revealed in ITRS2001, MEMS and Chemical Sensor are begin to be integrated
into SoC. The application domains include, bio, medical, chemical, mechanical, and
transportation that are closely related to our daily life. It is estimated that in 2010, they
will create another peak for semiconductor industry and integrated circuit industry. As of
today, we are more familiar with digital circuits than the analog sensor circuits. The goal
of this project is to study the low-voltage low-power analog and mixed signal sensor
integrated circuits.
This project proposes smart low-voltage low-power analog and mixed signal sensor
circuits. It amplifies the small signal from MEMS and chemical sensors and sends to

low-power and low-voltage requirement is to meet the critical power requirement in
invasive environment. The smart feature is to activate or shut down certain modules
depending on the signal level. With which, the power can be further reduced.
Smart Mixed Signal Circuit is a distinguishing feature of this project. The function is
to use a simple mechanism to detect the signal level to determine the activated modules
and the biasing current within the module to reduce the power consumption. It can be
regarded as a combination of macro and micro power management mechanisms. A
biomedical signal, such as electrocadiogram and lung sound, is a transient signal. They all
appear in a short period of time with certain frequency. If the circuit can take the
advantage of the signal property, turn of or reduce biasing current while there is no signal,
then the overall power consumption can be significantly reduced.
After detail study and evaluation, the project proposes the sensor interface including
those for resistance, capacitance, and inductance measurement. The micro watt amplifiers
include current, voltage, and charge amplifiers. These include that to be used in the test
vehicle proposed in subproject 3 for measuring the capacitance, resistance, or inductance
change in the MEMS but also for the applications mentioned earlier.
The goal of this project, in addition to the design, implementation, and testing of the
sensor circuits, we would like to adopt the modules in Subproject 5 to integrate a sensor
subsystem, as shown in the figure below. It can be integrated into a platform based sensor
system SoC. Users are able to use software to change the circuit parameters for the target
applications. Or, users can take certain part of the circuits for a specific application IC to
reduce the power consumption.

NSC 94-2215-E-009-052（94N575）
-----------------------------------------------------------------------------------------------

Title：Advanced Test Technology Development for RF Circuits and Systems
Principal Investigator：Su Chan-Chin
Keywords：

After years of development, wireless communication has become essential for
telephony and data communication. In mobile communication, there are many
different standards and systems in use simultaneously, such as GSM, DECT, PHS,
GPRS, WCDMA, CDMA2000. In data communication, Bluetooth and 802.11 are
competing for market acceptance. Therefore, RF circuit design and applications are
wide and extensive. Furthermore, in wired communication, fiber and high speed
copper wired serial links are all fall in RF frequency range. Due to its high added
value, design houses, foundry, and system houses are focus on this sector of
semiconductor industry in order to make high profit margin. However, the testing of
RF circuits and systems are not well developed as design and manufacturing.
Traditionally, RF circuit and system test are achieved in two ways. The first one
is to use specialized instrument to send and receive high quality RF signals to
determine the circuit parameters and go-no-go of the circuits. The disadvantage is the
cost overhead. The instrument in RF range is not only expensive but also very
difficult to operate and maintain. It seems that it is difficult to meet the requirement of
low cost mobile hone handset or wireless local area network. The second type uses
the loop back mechanism for the test. The digital data is sent by the transmitter and
received and recovered by the receivers. It checks the bit-error rate (BER) of the
received data to determine the function of the CUT. It has the properties opposite to
the first method. The test equipment cost is low. But, the test time is very long.
In this project, we would like to propose an RF test methodology which take the
advantage of the RF system architectures nowadays. We utile the built-in
analog/digital (AD/DA) converters and digital signal processors (DSP) as the test
resources for the test. The DSP is responsible for sending digital IF signal to the DAC.
DAC then coverts the discrete signal into analog signal. After that, it is up converted
into RF band by the RF transmitter module. The RF receiver module receives the RF
signal and down converts it into IF band. The ADC convert it into digital form. Finally,
the DSP collects the received digital signal and use DSP techniques to extract the
circuit parameters. The DSP techniques will be based on the “intrinsic response
extraction” we proposed earlier [1-8]. Without the affections of other modules, this
technique is able to extract the intrinsic responses. By this means, the test cost is
minimal because the AD/DA converters and DSP modules are built in already. There
is no external RF ATE needed.
NSC 94-2215-E-009-079（94N602）
------------------------------------------------------------------------------------------------

Title：Design and Implementation of a Digital Programmable PFC-PWM Control IC
(3/3)
Principal Investigator：Tzou, Ying-Yu
Keywords：Digital power control IC, Mixed-mode signal IC design, Power factor
correction control, Pulsewidth modulation control, Switching power
supply

This project makes a research on the design and implementation of a digital
PFC-PWM combination control IC for off-line single-phase switching-mode power
supplies. The combination of the power factor control and pulsewidth modulation
control can greatly simplify the control circuitry. The rising edge switching of the
PFC switch with synchronization of the trailing edge switching of the PWM switch
can minimize the output voltage ripples. The abstract of this three-year project is
described as follows:
The 1st-year: A Research on the Digital Control Architecture for the PFC-PWM
Combination Control
During the first year of this project we will make a study on control architectures
suitable for the PFC-PWM combination control of off-line switching power supplies.
A digital programmable PFC-PWM control scheme will be proposed in applications
to single-phase switching power supplies with a boost PFC input converter and a
flyback output converter. Numerical realization issues for the PFC-PWM control will
be studied. Simulation-oriented analysis will be carried out to study the feasibility of
the proposed control scheme. Experimental verification will be carried out using a
designed CPLD-based programmable digital controller.
The 2nd-year: Design and Implementation of a Digital PFC-PWM combination
Control IC for Off-Line Switching Power Supplies
In the second year of this project we will design a digital PFC-PWM control IC
according to the proposed digital control scheme of the previous research. The
research will focus on the design and optimization of specific circuits used in the
PFC-PWM control IC. In order to apply the proposed control multi output PWM
generator will be designed. A fully-digital control scheme will be developed and
verified using a CPLD-based digital control board. Practical hardware realization
issues will be studied using simulation-oriented analysis.
The 3rd-year: Design and Implementation of a Mixed-Mode PFC-PWM
Combination Control IC for Off-Line Switching Power Supplies
The control of high-performance AC-DC and DC-DC converters usually employ
fast response inner-loop current controller. This research will apply mixed-mode
design techniques for the synthesis of a PFC-PWM combination control IC. The inner
current loop controller will be realized using analog technique and the outer voltage
loop controller will be realized using digital technique. Simulation oriented analysis
and design will be used in the synthesis of this mixed-mode PFC-PWM control IC. A
systematic design procedure will be developed for the design and implementation of
mixed-mode control IC design for switching-mode power converters.
NSC 94-2213-E-009-146（94N408）
------------------------------------------------------------------------------------------------

Title：Design and Implementation of Optimal On-Line Intelligent Control
Systems(3/3)
Principal Investigator：Wang Chi-Hsu
Keywords：Fuzzy Logic, Neural Network, Real-time Control System

This project is to explore the theoretical and practical issues of real-time
applications of FNN (Fuzzy Neural Network) for linear and nonlinear dynamical
systems. This project will firstly to develop the modified dynamical optimal training
for a certain class of FNN. Different kinds of FNN will require different
considerations in the optimal training algorithm. Under real-time environment, there
is very little time for the training of FNN. Therefore the FNN must be optimally
trained with maximum error reduction in minimum time period. The optimally trained
FNN can then be used as a controller or as part of lease-squared identification, all in
real-time environment. Secondly a new practical closed-loop model for real-time
control system will be proposed. This is to consider the unavoidable time delay
incurred in the optimal training of FNN and/or other computational effort. It is hoped
that the maximum allowable computation time for the FNN can be found. This result
will be valuable due to the fact that the maximum allowable computational time can
be used to select the speed of the computer hardware for economical design. The
above issues will be conducted in the first and second years using computer
simulation. The true hardware implementation using several popular plants, such as
the inverted pendulum, the Chua’s chaotic circuit, …, etc., will be conducted in the
third year under VisSim/Pro (Matlab plus DSP card) for real-time FNN-based control.
NSC 94-2213-E-009-014-（94N513）
-----------------------------------------------------------------------------------------------

Title：The Study of Image Processing and Ergonomics Based Intelligent Safety and Comfortable
Driving System (1/3)
Principal Investigator：Wu Bing-Fei
Keywords：Car following,Lane changing,Stop and go,Driving recorder,Mobile
surveillance

The main project will focus on the intelligent safety and comfortable driving
system.
Besides of the integration of signal processing, the human technology is also
considered. The project goals including car following、lane changing、stop and go、
driving recorder、actively mobile surveillance system、noise cancellation device、and
comfortable mechanisms for vehicles are developed by six subprojects to handle the
corresponding studies and achieve the common research target.
This project is a highly integrated study on image processing and ergonomics. In
theimage processing part, three topics are covered：real-time lane detection、obstacle
detection and image compression. In the ergonomics part, three topics are studied:
comfortable cabin car communication system, mechanisms, and cruise safety control
system. Finally, the research results will be integrated to realize the mail project goals.
The execution of this project can promote the research to the international level,
especially the ergonomics studies on the safety driving system.
NSC 94-2213-E-009-062-（94N343）
-----------------------------------------------------------------------------------------------

Title：Subproject 1:Image Processing Based Real-Time Lane Detection and Head
Light Identification with Tough Weather Condition (1/3)
Principal Investigator：Wu Bing-Fei

This project focuses on the development of real time vision processing and
recognition image coming from the CCD cameras installed in vehicles. The image
information includes the lanes, the head and tail lights, and the road signs, which can
provide the other subprojects to achieve the goals of the main project. The
technologies developed in this project play role of the driving assistant to keep the
dangerous driving behavior off. For this issue, the lane, head and tail light, and the
road sign information will be processed and transferred to related subprojects to
realize the scheduled achievements of the main project, an Intelligent Safety and
Comfortable Driving System.
The image detection and recognition will be studied year by year. The real-time
head-light recognition will be completed and integrated with lane detection. The work
in the second year will focus on the tail-light detection and the lane detection on the
rainy day. The road sign detection and the adaptive lane detection study will be
accomplished in the last year.
NSC 94-2213-E-009-066-（94N347）
-----------------------------------------------------------------------------------------------

Title：The study of a high security surveillance system with the integration of DWT
and DCT and video/audio synchronization
Principal Investigator：Wu Bing-Fei
Keywords：Discrete Wavelet Transform, Discrete Cosine Transform, MP3, AES,
Surveillance System

This project focuses on the development of a high performance and high security
Video/Audio surveillance system. In this project, an innovative video compression
method that integrates Discrete Wavelet Transform and Discrete Cosine Transform is
proposed. Furthermore, in order to enhance the functionality, the intelligent image
compression techniques will be developed. The recognizable ability of the important
part of the surveillance images will be better, and the compression ratio will be higher.
Moreover, the low complexity MP3 audio compression method, Real-time Transport
Protocol and the programmable AES cryptography system are merged, too. The users
can not only see the image but also hear the sound of their targets, and the secret
information of the users can be radically protected. Besides, the mobile
communication devices are also combined in order to reach the goal that the users can
view the high quality surveillance images at anywhere and anytime.
The content of the project are separated into three parts: (1) Increase the speed
and the quality of the image compression method, integrated MP3 audio compression
technique, implement the hardware of the programmable AES cryptography system,
and integrate the mobile communication devices. (2) Design the image compression
method that combine DWT with DCT, and develop intelligent image compression
techniques. (3) Implement the whole system by using a low cost embedded system
and a dedicating hardware. The final objective of this project is to accomplish an
intelligent and high speed embedded surveillance system.
NSC 94-2213-E-009-067-（94N348）
-----------------------------------------------------------------------------------------------


DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
 views: 0 posted: 4/29/2013 language: English pages: 32
wang nianwu http://