Department of Computer Science Title:Design and Development of ...

shrewdnessmodernMobile - Wireless

Dec 14, 2013 (3 years and 7 months ago)

242 views

Department of Computer Science


Title

Design and Development of Benchmarking Technologies and Tools for Storage

Systems of

Embedded Network and Communication Devices

Principal Investigator

Li
-
Pin Chang

Sponsor

National Science Council

Keywords

Benchmark
,
S
olid
-
state
D
isks
,
F
lash
M
emory,
F
ile
S
ystem
,
E
mbedded
S
ystem
,
O
perating
S
ystem


Flash memory is widely deployed in embedded computing systems because of its

non
-
volatility, shock
-
resistance, and power conservation. To test the performance of

flash
-
memory
-
b
ased storage systems, currently benchmark suites for disk
-
based
storage

systems are used. The ordinary benchmark suites may identify the commonly
existing

performance bottleneck, the source that causes the performance degradation
is not easy to

identify, t
hough. That is because the physical characteristics and
geometry of flash memory

impose many constraints on its management. For example,
newly written data are dispatched

to garbage
-
collected space. Garbage collection
would involve intensive internal data

movement and then introduce lengthy delay.
Furthermore, garbage collection would wear

portions of the flash memory out. So to
lengthen the overall lifespan of flash memory,

wear
-
leveling activities would then be
performed. By the above discussions we may f
ound

that to test a flash
-
memory
-
based
storage system is very different from to test a disk
-
based

ones. Besides response and
throughput, to test flash
-
memory
-
based embedded storage

systems we have to include
the lifetime issue, the support of real
-
time app
lications, the

computational
-
resource
costs of time and space, flash
-
memory space utilization, and

bandwidth utilization.


The objective of this research project is to design and implement the tools,
procedures and

methods of benchmark for flash
-
memory
-
bas
ed storage systems.
Working items include the

methods to capture and to analyze kernel/firmware events,
design of test procedures and

performance indexes, implementation of batch test tools,
and platform
-
surrogating

techniques. In the first year, we would
conduct large
-
scale
benchmark on existing

flash
-
memory file systems and storage systems. The technique
issue is not only to find under

what circumstances a storage system would deliver
good performance but also to reveal that

under what workloads the stora
ge system
would performs poorly. In the second year, in
-
lab

benchmark tools and procedures are
to be completed. A technical item is to develop an

auto
-
pilot diagnosing procedure. It
is to automatically target and to identify the design weak

points pertaini
ng to
performance degradation. In the third year, complete test tools and

procedures can be
delivered to vendors so that they can independently test their products

without much
consulting us. To speed up the test and benchmark, a technique referred to as

p
latform
surrogating is to be developed. It is to provide abstraction model of hardware and

related software interface. With platform surrogating, vendors may transfer their

firmware/driver source code to the virtual platform without many modifications. By
jointly

using the benchmark tools and the platform
-
surrogating technique, the cycle of
test, debug,

and benchmark of their algorithms can be significantly accelerated.

NSC
97
-
2218
-
E
-
009
-
035
-
(
97R113
)

---------------------------------------------------------
--------------------------------------


Title

Multi
-
function Personal /Group Communication System Design II

Principal Investigator

Ming
-
Feng Chang

Sponsor

National Science Council

Keywords

A
d
H
oc
P
ersonal/
G
roup
C
ommunications,
M
ulit
-
attribute
D
ata
Q
uery,




L
ocation
-
based

C
ommunications, P2P
C
ommunications


The existing personal/group communication services provided on the Internet and

PSTN, such as, telephony, instant messages, and VoIP, requires users to specify the
identifier

of the receiver(
s), such as, telephone number and Skype ID. This
requirement may limit the

applications. For example, it is difficult to contact an old
friend without knowing his or her

phone number. It is also not easy to set up an ad
hoc instant message group for person
s in a

tour group. Our goal is to develop a
communication service platform to support

communication scenarios where the
specific identifier of the receiver(s) is unknown or

non
-
existing.



In the second year of this two
-
year project, we will continue devel
op a
multi
-
function

personal/group communication service system. The system supports
not only

communications

using specific receiver identifiers, such as telephone number
and Skype ID, but also

communications using unspecific identifiers. The receivers are

specified by their

characteristics, such as their attributes and location information. We
will use a structured P2P

architecture to build this communications platform. The key
design issues include 1) the

privacy protection of personal information, 2) eff
icient
matching between call requests and

receivers’ attributes, and 3) efficient support for
location
-
based communications. The

functions of this communication service will
include not only the popular instant message,

and VoIP, but also
looking
-
for
-
a
-
per
sona on the net, ad hoc group communications, and

location
-
based
communications.

NSC
97
-
2221
-
E
-
009
-
059
-
(
97R280
)

-----------------------------------------------------------------------------------------------


Title

Cross Layer Design and QoS in Cooperative

OFDM Systems

Principal Investigator

Hsi
-
Lu Chao

Sponsor

National Science Council

Keywords

C
ooperative
C
ommunication,
C
ooperation
A
ssignment,
C
ross
-
layer
D
esign, QoS



Cooperative diversity has recently emerged as a new and effective technique to
comba
t fading in

wireless networks, and its basic idea is to explore the broadcast
nature of the wireless channel. In a

cooperative communication system, each wireless
user is assumed to transmit data as well as act as a

cooperative agent for another user.




Cooperation leads to interesting trade
-
offs in code rates and transmission power.
Using transmission

power as an example, one may argue on one hand that more

power is needed since each user is transmitting

for other users, when in cooperative
mode. On the

other hand, the baseline transmission power for both users

will be
reduced cause of diversity. However, it’s obvious that not only to a homogeneous
wireless network

(such as an ad hoc network), but also to a homogeneous wireless
network which integrates b
oth WWANs

(e.g., WiMAX and 3G) and WLANs (e.g.,
WLAN), cooperative communication significantly increases total

system throughput.



To develop and implement cooperative communications, several complicating
issues must be addressed,

including the loss of

rate to the cooperating mobile, overall
interference in the network, cooperation

assignment and handoff, fairness of the
system, and transmit and receive requirement on the mobiles. The

topic of our project
is

Cooperative Communication in OFDM Wireless N
etworks
」,
and there are five

subprojects to study the cross
-
layer design and QoS, user
-
cooperative protocol design
and analysis,

space
-
time coding and decoding, cooperative multiuser MIMO
-
OFDM
systems, and reconfigurable platform

design, respectively.



The main objective of this subproject is to support QoS in cooperative OFDM
networks. To achieve our

design goals, the research topic of second year will
investigate cooperative assignment in heterogeneous

OFDM networks, which
includes IEEE 802.11 WLANs, I
EEE 802.16 WiMAX, and IEEE 802.20

FLASH
-
OFDM systems. In the third year, we will study related issues of MAC
protocol design in a

cooperative network, and propose our MAC protocol. We will
further study power saving function and

integrating with the propos
ed MAC protocol.



For the proposed protocol, we will evaluate their performances by means of
simulations. In addition,

we will implement our proposed protocol on

WLAN/WiMAX development platforms.

NSC

97
-
2219
-
E
-
009
-
002
-
(
97R057
)

--------------------------
---------------------------------------------------------------------


Title:

System
MobiTransporting: Information Fusion, Dissemination, Transportation
and Multicasting Techniques

for Vehicle Ad
-
Hoc Networks (I) (II) (III)

Principal Investigator:

Chien Chen

Sponsor:

National Science Council

Keywords:



The objective of this project is to develop the transportation techniques which
provide

the fast, reliable and efficient information deliver
y over Vehicular Ad
-
hoc
Networks

(VANET). The topics of this research include (1) Fast information fusion,

dissemination, replication and storage in VANET. (2) Since VANET inherits some

characteristics of Delay Tolerance Network (DTN), we will discuss the
TCP issues

over DTN and our transport layer solution to achieve reliable end
-
to
-
end information

transportation. (3) Finally, we will design a multicast protocol for VANET to achieve

efficient information forwarding.

NSC
97
-
2221
-
E
-
009
-
050
-
MY3
(
97R212
)

------
-----------------------------------------------------------------------------------------


Title

The
D
esign and
I
mplementation of
I
nterface Interfacing
G
enerator for
I
ntegrating and
B
ridging

F
ront
-
end
R
ecognizers and
B
ack
-
end
A
pplication
S
oftware
S
ystems

Principal Investigator

Deng
-
Jyi Chen

Sponsor

National Science Council

Keywords

See
-
through
I
nterf
ace
,
Software Engineering
,
Parser Generator
,
Interface
G
enerator
,
Recognizers
,
Speech Recognizers


It has been shown that the major effort spent on the design and implementation of
the

application system software is the user interfaces (UI) [31] (or human
-
machine
interface

(HMI)). If UI can be developed in a short time, it will be a great help to
reduce development

time for application software systems. Therefore, many
researchers have been seeking better

solutions to aid UI designers to create UI
systems.


In general, there are two kinds of interface system: human
-
machine interface and

interface for bridging application software as one. The former concerns the GUI
design and

implementation for the application software. The later concerns with the
integratio
n of

recognizer and application software together to form a new application
software that uses the

recognizer as the front
-
end system. In this proposal research,
we layout a three
-
year

integration project that focus on the later interface technology,
calle
d generic Interface

Interfacing system, Figure 1 depicts the detailed components.


Figure 1. The proposed Interface Interfacing System


Basically, application systems that utilize recogniti
on technologies such as speech,

gesture, and color recognition pro
vide human
-
machine inte
rfacing to those users that
are

physically unable to interact with computers through tradit
ional input devices such
as the

mouse and keyboard. Current solutions, however, use a
n ad hoc approach and
lack of a

generic and systematic wa
y of interfacing application syste
ms with
recognizers. The common

approach used is to interface with recognizers through
low
-
level programmed wrappers that

are application dependent and require the details
o
f system design and programming

knowledge to perf
orm the interfacing and to make
any modificat
ions to it. Thus, a generic and

systematic approach to bridge the
interface between recognize
rs and application systems must

be quested.


In the first year of this integration research work, we propose a generic

and visual

interfacing framework for bridging the interface between application systems and
recognizers

through the application system’s front end, applying a visual level
interfacing without

requiring the detailed system design and programming knowledge,

allowing for modifications

to an interfacing environment to be made on the fly and
more importantly allowing the

interfacing with the 3
rd
party applications without
requiring access to the application’s source

code. Specifically, an interfacing script
lan
guage for building the interfacing framework is

designed and implemented. The
interfacing framework uses a see
-
through grid layout

mechanism to position the
graphic user interface icons defined in the interfaced application

system. The
proposed interfacing

framework is then used to bridge the visual interface

commands
defined in application systems to the voice commands trained in speech

recognizers.
The proposed system achieved the vision of interface interfacing by providing a

see
-
through grid layout with

a visual interfacing script language for users to perform
the

interfacing process. Moreover such method can be applied to commercial
applications without

the need of accessing their internal code, and also allowing the
composition of macros to

release int
eraction overhead to users through the automation
of tasks. Figure 1 also indicates

an example that a solitary game or an authoring
system in window system can be played using

the speech recognizer in window
system after the integration using the proposed
approach.


The main contributions of such interface interfacing system include 1)
Productivity is

reasonable good: system developers no need to trace the low level
code (without requiring the

detailed system design and programming knowledge)
while integrat
ion the recognizer with

the application software, 2) Maintenances effort
is low: allowing for modifications to an

interfacing environment to be made on the fly,
and 3) Flexibility is good: allowing the

interfacing with the 3
rd
party applications
without re
quiring access to the application’s source

code.


In the 2
nd
year project, we continue the concept used in the first year to investigate
the

handheld device environment such as PDA or Smart phone. In this case, we use
the remote

control capability in the s
mart phone as the front
-
end recognizer and java
program as the

back
-
end application software. The choice of the java as the
implementation language is

rested on its heterogeneous platform adaptation features.
Specifically, we will propose an

interface gene
rator, similar to the concept of the
parser generator, to automatically generate

remote control programs for a specific
multimedia application in the smart phone. With this

generator, designer does not
need to write the textual remote control programs in t
he smart

phone. This will
simplify the development process and make the control system development

and
modification more flexible. Figure 2 depicts the detailed components. In Figure 2, it

indicates that a interface generator (the interface interfacing sys
tem) can proceed to
perform a

code generation (Java Midlet AP) after the back
-
end application software in
the PC

environment has been integrated with the remote control module using the
proposed approach.

Of course, the remote control module can be replace
d by Wii
-

like recognizer if it is needed.


Figure 2

剥m潴o⁃潮 r潬⁉o瑥rf慣楮朠g祳瑥m 佶敲癩敷


With the quick advance of technology, screen display of digital TV and mobile
system

becomes more and more elegant and is able to present fine and vivid
multimedia contents.

Most of the multimedia contents, such

as advertisement, motion
pictures, messages, etc., can

be displayed on different kinds of platforms. If user can
use some simple instruments (such as

smart phone, PDA, etc.) to remotely
communicate with the multimedia application module in

the display dev
ice (such as
PC monitor, digital TV, etc.), then the control becomes live and

interesting. But there
are various control instruments and display devices, and different kinds

of control
methods. If one wants to write the control program or partially modify
the control

features for the multimedia application module in the display device, then he needs to
know

the software source code in the multimedia application module that will be
remotely

controlled, so that he can custom
-
design a set of remote control pro
grams for
each multimedia

application. However, there is a lot of multimedia application; a
custom design for each of

these applications becomes time consuming and less
efficient.


Once we have built the interface interfacing system for both in the PC
envi
ronment (the

first
-
year project) and smart phone environment (the 2
nd
year
project), we are ready to author

various kinds of multimedia presentation such as
mobile name card template system or

e
-
card presentation and use smart phone to
remotely synchronize

the presentation on top the

big LCD and LED displayers

This
is the major effort spent in the 3
rd
year. A more detailed

elaboration of this part will be
narrated in section 12.

NSC

97
-
2221
-
E
-
009
-
062
-
MY3
(
97R221
)

-----------------------------------------------------------------------------------------------


Tit
le:

A Study on Sprite Generation without Segmentation Masks and Fast Multiple


Sprite Generation

Principal Investigator:

Ling
-
Hwei Chen




Sponsor:

National Science Council

Keywords:

Sprite
G
eneration, Sprite
C
oding, Image
B
lending, Multiple
S
prites,


Reference

F
rame, Sequence
S
plitting, MPEG
-
4.


Sprite coding, which can increase the coding efficiency of backgrounds greatly, is
a

novel technology adopted in MPEG
-
4 object
-
based coding. The sprite generator
introduced

in MPEG
-
4 blends frames by

averaging, this will make some places,
which are ever

occupied by moving objects, look blurring. Thus, providing segmented
masks for moving

objects is suggested. Since perfect segmentation is impossible,
some ghost
-
like shadows will

appear in the generate
d sprite. To treat this problem, in
the first year of this project, an

blending strategy without needing segmentation masks
will be developed. It is based on the

fact that for each point in the generated sprite, the
corresponding pixels in most frames

belo
ng to background and only few belong to
moving objects. A counting schema is

provided to make only background points
attend in average blending.


In the second year of this project, a fast multiple
-
sprite partition algorithm will be

proposed. Due to the ge
ometric transformation applied to each non
-
reference frame in
the

procedure of sprite coding, the generated sprite is distorted and the available view
angles

relative to the reference frame are restricted. These problems can be resolved
by using

multiple s
prites. An optimal multi
-
sprite generator has been proposed by
Farin et. al., but the

optimal method requires huge computation in the sprite coding
costs. To treat this

disadvantage, a fast near
-
optimal multi
-
sprite partition algorithm
will be developed in

this

project based on frame translation and scaling information.
Our fast algorithm will divide a

video sequence into several subsequences. Sprites
will be generated from every subsequence,

and the total quality and size of all
generated sprites will be n
ear those of sprites generated

by Farin

s optimal
generator.

NSC
97
-
2221
-
E
-
009
-
137
-
(
97R394
)

-----------------------------------------------------------------------------------------------


Title

Study and Design of QoS Techniques over Multimode Wireless Ac
cess

Networks with High

Mobility

Principal Investigator

Yaw
-
Chung Chen

Sponsor

National Science Council

Keywords

WiFi
M
ultimedia, Quality
-
of
-
Service (QoS), WiMAX,
M
ultimode


W
ireless
A
ccess, Multiple

Input Multiple Output,
E
nd
-
to
-
end




C
onnectivity, Service Level Agreements (SLAs),
S
oft

H
andoff, QoS


R
outing




The fast technology advance in wireless networks leads to the quick growth in
wireless application

services, such as WiFi multimedia and Internet games. Although
nowadays these kinds of services can

be deployed successfully in wireless LAN
environment, however, these services still could not be

support satisfactorily if the
wireless client is in motion. This is due to the deficiency in coverage of

wireless LAN
as w
ell as the lack of an effective mechanism which can fulfill the complex

requirements

for roaming, especially regarding the Quality of Services (QoS).
WiMAX is a broadband wireless

technology that supports fixed, nomadic, portable
and mobile access, with no
ne
-
line
-
of
-
sight coverage

radius of several miles. Two
versions of WiMAX have been defined: IEEE 802.16
-
2004 and IEEE

802.16e, the
former is optimized for fixed and nomadic access, while the latter will support
portability

and mobility. Regarding the hando
ver across heterogeneous networks, the
IEEE 802.21 standard is

undergoing and it will define the specification.

Since the cell of wireless LAN is too small to achieve as wide coverage as cellular
mobile does, it is

difficult to maintain connectivity once a

mobile client is moving, not
mention the QoS. In this proposal,

we will investigate the support of QoS and
mobility based on a multimode wireless access infrastructure

(WiMax/WiFi/Cellular), with MIMO (Multiple
-
Input Multiple
-
Output) radio
techniques, whi
ch is very

likely to be used in WiMax. With MIMO, mobile client
based on the IEEE 802.16e could perform soft

handoff, which is a make
-
before
-
break
mechanism, to transit from a cell to another one, or even from

one type of network
such as WiFi to another on
e such as WiMax, therefore the end
-
to
-
end connectivity

as well as the service quality, such as Service Level Agreements (SLAs) could be
maintained during

handoffs. We will extend our current study of mobility based on
WLANs to multimode access networks,

an
d investigate the cross
-
network soft
handoff, which may include QoS routing, buffering, and traffic

classification and
control. We will also prototype the system module at the second and third years based

on the components available on the market.

NSC95
-
22
21
-
E
-
009
-
055
-
MY3 (95R452
-
1)

-----------------------------------------------------------------------------------------------


Title:

Brain Computer Interface System using Motor Imagery EEG Recordings

Principal Investigator:

Yong
-
Sheng Chen

Sponsor:

National Science Council

Keywords
:
Biomedical
Si
gnal
P
rocessing,
B
iofeedback,
EEG, Brain Computer
Interface


In EEG
-
based Brain
-
Computer Interface (BCI) systems, potential waveforms
recorded at

subject’
s scalp through EEG electrodes are normally used for feature
extraction and signal

classification. By recognizing EEG recordings measured at
different mental situations of subject’s

brain, computers or robots can execute specific
tasks corresponding to the

mental situations.

Unfortunately, the potential measured at
the EEG electrodes is an ensemble effect of the

bioelectric signals originated from all
of the active regions and conducted within the whole

volume of the head. Brain
activities that are not rela
ted to the thoughts for control or the specific

events will also
contribute to the measurements and thus interfere with the following signal

processing
and classification procedures. To enhance the accuracy, stability, and efficacy of

conventional EEG
-
base
d BCI systems, we propose in this project to analyze various
motor imagery

tasks aiming at the most suitable one for BCI systems. We also plan to
develop biomedical signal

analysis technologies, maximum contrast beamformer, to
filter out temporal waveforms

of the

bioelectric brain activities that are related to the
control thoughts or specific events. BCI systems

can thus utilize the brain activation
signals, instead of the EEG measurements, for recognition

purpose. Furthermore, we
will also construct a BCI

with biofeedback so that both the subjects

and the system
can adapt themselves for better performance.

NSC
97
-
2221
-
E
-
009
-
139
-
(
97R455
)

-----------------------------------------------------------------------------------------------


Title

Multi
-
View Factori
zation Methods for 3D Scene Reconstruction

Principal Investigator

Zen Chen

Sponsor

National Science Council

Keywords

3D
S
cene
R
econstruction,
I
nvariant
F
eature
E
xtraction,
F
eature
P
oint




C
orrespondences,
C
amera

C
alibration,
M
ulti
-
view
F
actorization
M
ethod,

3D
S
cene
M
esh

M
odel,
T
exture
M
apping


The three
-
year research project addresses a series of computer vision techniques
for reconstructing 3D

dense scenes from plenty of im
ages taken by either a large
number of fixed cameras scattering around the

scene or a single camera moving
around the scene. In addition, the images can be either calibrated or

uncalibrated. In
the first
-
year project a dense reconstruction of a 3D indoor s
cene (or a target object)
will be

derived from multiple cameras with a fixture deployed in the shape of a dome.
They aim at a common object,

while capturing the images. In this setting all cameras
are calibrated in advance, that is, their intrinsic and

ext
rinsic camera parameters are
all known and fixed. We propose an accurate dense scene reconstruction

method
based on multi
-
view factorization theory rather than by gluing pairwise stereo
reconstructions. First

of all, affine
-
invariant interest points are ex
tracted by our
proposed elliptical LoG detector which can resist

against the view change caused by
the different camera viewpoints. The correspondence point matching

among the
candidate interest points across the multiple views is conducted in an
affine
-
no
rmalized image

space instead of the original image space. As a result, the
failure common to the standard window
-
based

correlation test on the point
correspondence is avoided, in particular, in the wide baseline stereo setting. Next,

to
make a dense scene
reconstruction possible we propose to obtain an initial boundary
edge model (IBEM)

based on the initial highly accurate interest points extracted.
Then an iterative or progressive refinement of

the IBEM is conducted based on the
multi
-
view photo
-
consistenc
y test and visibility test. A final refined

mesh model
obtained becomes a dense mesh model. The model can be mapped patch by patch
with texture

fetched from a corresponding image patch in the respective properly
selected view.


In the second
-
year project a

multi
-
view dense outdoor scene reconstruction is
conducted using a single

camera, assuming its intrinsic camera parameters are known
and fixed throughout the picture shooting. Since

the camera extrinsic parameters are
unknown in this camera set
-
up, we nee
d to estimate these parameters.

Before the
estimation a sufficient number of reliable corresponding point pairs have to be
obtained first. This

is made possible through our affine
-
invariant interest point
detector in which some uniquely matched point

pairs

can be found. Besides, more
point pairs are obtained through the use of proximity measure on the

interest points
with respect to the already uniquely matched point pairs. The multi
-
view factorization
method

is used to derive the camera extrinsic parameter
s and 3D interest points. To
deal with possible local minimum

traps in the parameter estimation a global optimal
solution to the parameter estimation the above solution is

used as an initial solution in
a stochastic search method making use of the simulate
d annealing procedure

with the
aid of the orthogonal experimental design for the solution space. Consequently, the
reliable eipolar

geometry relationships among the cameras can be obtained and the
relationships will be used to disambiguate

the similar stru
cture patterns abundant in
the outdoor man
-
made scene. Finally, a dense reconstruction can be

done in the same
way as given in the first
-
year project.


The third
-
year project will address the multi
-
view 3D reconstruction using
uncalibrated images. The

unca
librated images are quite common in the nowadays
hand
-
held digital camera shooting during which the

lens zooming function is likely
activated. We shall study the effect of the change in the camera intrinsic

parameters
on the reconstruction accuracy. For th
e group of pictures captured with or without the
change in

the intrinsic parameters we apply the multi
-
view factorization method to
obtain a projective 3D scene model,

and then two possible ways of transforming the
projective reconstruction to a Euclidean
reconstruction will

be proposed: one using
the scene prior knowledge and one using the autocalibration. At the same time, we

shall study how to deal with the image noise, missing points, and outliers as a whole
during the multi
-
view

reconstruction process
so that more accurate reconstruction can
be achieved.

NSC
97
-
2221
-
E
-
009
-
145
-
(
97R395
)

-----------------------------------------------------------------------------------------------


Title

Design and Implementation of An Ultra High
-
Resolution 500
-
1000 Electrodes
Cochlear Prosthesis

System and the Correspondence Stimulating Strategy

(1/3)

Principal Investigator

Charles T. M.

Choi

Sponsor

National Science Council

Keywords



Research shows tha
t while 6 to 9 effective channels are sufficient for hearing English
speech in

quiet environment. It is not enough to listen to speech in noise, music and
tonal language,

particularly

Mandarin Chinese.

This sub
-
project project is attempting
to leap
-
frog th
e current state of the art cochlear implant

technology

to raise the
number of cochlear implant electrodes from 16 or 24 to 500~1000

virtual electrodes,
thus dramatically improve the spectral cues delivered to the auditory nerves.

Stimulating strategies bas
ed on the new virtual electrode system will be developed,
thus,

increasing the temporal cues sent to the auditory nerves. It is expected that the
improvement in

performance for cochlear implant users hearing in noisy environment,
in listening to music, and

tonal languages, particularlyMandarin Chinese, will be
equally significant.


Year 1:

1. Develop and implement a new 250 and 500 virtual electrode system in a TI 32055
DSP chip

and an Analog Device Black Fan DSP chip.

2. Develop and implement new stimulati
ng strategies based on the 250 and 500
electrodes

system

3. Transfer to the cochlear Implant Speech Processor for clinical evaluation for
subproject 4 and

subjective performance evaluation for sub
-
project 5.


Year 2:

1. Develop and implement a new 500 and
1000 virtual electrode system in a TI 32055
DSP chip

and an Analog Device Black Fan DSP chip.

2. Develop and implement new stimulating strategies based on the 500 and 1000
electrodes

system

3. Transfer to the cochlear Implant Speech Processor for clinical
evaluation for
subproject 4 and

subjective performance evaluation for sub
-
project 5.


Year 3:

1. Develop and implement a new 1000 virtual electrode system in a TI 32055 DSP
chip and an

Analog Device Black Fan DSP chip.

2. Develop and implement new stimulat
ing strategies based on the 1000 virtual
electrodes system

3. Transfer to the cochlear Implant Speech Processor for clinical evaluation for
subproject 4 and

subjective performance evaluation for sub
-
project 5.

NSC9
5
-
22
21
-
E
-
009
-
366
-
MY3
(9
5R687
-
1
)

-----------
--------------------------------------------------------------------------------------------


Title

Sensor Fusion for Intelligent Visual Surveillance

Principal Investigator

Jen
-
Hui Chuang

Sponsor

National Science Council

Keywords

Sen
sor Fusion
,
Visual Surveillance


This project is based on the basic framework of visual surveillance wherein
various

types of cameras are deployed in an indoor environment. These cameras are
capable of

capturing
visible as well as inferred lights, and they can be fixed in space,
but may have

pen
-
tilt
-
zoom (PTZ) capability, or mounted on a mobile platform. The
motivation is to

achieve more critical visual surveillance tasks through multi
-
sensor
fusion. To that end,

the

project will study the communication and information sharing
among heterogeneous

cameras, and the system integration issues including those
associated the dynamic

surveillance functions of the mobile platform, so as to achieve
the ultimate goal of

act
ive/passive surveillance around the clock.


For the proposed research areas of sensor fusion, they are dealing sensors of
somehow

homogeneous (image
-
based surveillance cameras) as well as heterogeneous
(visible and

inferred image data) nature. Moreover, an
alyses of sur veillance data
obtained from both

fixed and mobile platforms need to be taken into account together.
Thus, the related

research topics and the associated attributes include:


1.
Around the clock surveillance


with the help of visible
-

light,

near
-

inferred, and

mid
-

inferred cameras for normal, low, and no lighting conditions, respectively.

2.
Range
-
extended visual surveillance


through camera self
-
motion (as with PTZ

camera), and with the introduction of a mobile platform.

3.
Multi
-
view ima
ge acquisition


for different viewing angles, and visual

information, of a scene associated with stationary as well as moving cameras.

4. More diversified design of surveillance functions


from information embedded

in different sensory data, e.g., fire o
r high temperature warning based on

temperature information associated with mid
-

inferred image data.


According to these research topics, the research plan of this project will include,
based

on the goal of the construction of an environment under visual
surveillance with
multiple

stationary and moving sensors, the development of surveillance modules for
individual

image sensors and navigation module for the mobile platform, followed by
the development

of sensor
-
fusion sub
-
systems with image registration a
nd coordinate
transformation

techniques, and some capabilities of object detection and tracking.
More information sharing

and integration among heterogeneous camera systems will
then be studied to aim at human

surveillance including gesture and motion anal
ysis.
The development of such an integrated

surveillance system will provide a platform
which is essential for system integration and

testing, and serve as foundation for
further, long
-
term research and development for more

real
-
world applications.

NSC96
-
2
2
21
-
E
-
009
-
125
-
MY2 (96N298)

-----------------------------------------------------------------------------------------------


Title

A Study on Realistic Synthesis Techniques for Motion, Skin
-
Muscle and Cloth


of Human

Characters

Principal Investigator

Jung
-
Hong Chuang

Sponsor

National Science Council

Keywords

Character
A
nimation,
M
otion
S
ynthesis,
F
acial
A
nimation,
B
one/skin


D
eformation and
R
endering,
C
loth
D
eformation and
R
endering,
R
eal
-
time

R
endering



Human characters play an important

role in t
he game production. A game will

not be a game without ruling characters. Famous animation or game companies,

such
as Electronic Art (EA), have put great effort to develop their character

animation
systems aiming to exhibit realistic human charact
ers in games. For

example, both
“Half
-
Life 2” and “NBA Live” have impressive human character

animations that
catch player’s eyes. Realistic human character synthesis

techniques, including smooth
motions, realistic skin and cloth deformation, and

realistic
skin and cloth rendering
are challenging problems that continue to

attract attentions from top research groups.



The main purpose of this project is to deve
lop a realistic human character

animation system. It contains three sub
-
projects. Project “A stud
y on motion

synthesis
of body and facial animation for human characters” studies the

synthesis of motion
and facial expression; project“A study on the simulation

of skin
-
muscle and cloth
deformation for human characters” studies

physically
-
based muscle/ski
n and cloth
deformation; project “A study on

real
-
time skin and cloth rendering for human
characters” develops realistic

real
-
time skin and cloth rendering enhanced by
real
-
time self
-
shadowing.



This project is responsible for the
planning and integrat
ion of the

development
and resulting systems of three sub
-
projects, including the design

of data structures and
interfaces, the design and implementation of the common

character animation
platform, and the integration and testing. This project

spans three
years, which is a
quite unique research team on human character system.

We expect to see the result of
this project will be applicable to local

game/animation industries and able to enhance
the competitiveness towards the

world market.

NSC95
-
22
21
-
E
-
009
-
195
-
MY3 (95R486
-
1)

-----------------------------------------------------------------------------------------------


Title

A
S
tudy on
R
eal
-
time
S
kin and
C
loth
R
endering for
H
uman
C
haracters

Principal Investigator

Jung
-
Hong Chuang

Sponsor

National Science Council

Keywords

Re
al
-
time
R
endering,
A
mbient
O
cclusion,
R
ealistic
R
eal
-
time
C
loth
R
endering,
R
ealistic
R
eal
-
time
S
kin
R
ende
ring



A successful animation or game consists of many important features, and one of
the most

important features is the character. How to create memorable characters to
attract audience or

players' attention is always the main goal in the game and movie

industries. To this end, lots

of time is spent on modeling characters with realistic and
meticulous appearances. For

efficiency reasons, most real
-
time applications render
characters with colored textures

designed by artists, which restricts the ability t
o
adjust detailed features of characters in

rapidly changing motions and affects the
quality of character rendering. In this project, a

realistic character rendering module
suitable for characters with highly dynamic motion will

be developed. We focus on
t
hree important features of character rendering: self
-
shadowing

effects, skin rendering,
and cloth rendering.



This project spans three years. In the first year, we will focus on developing
methods

for self
-
shadowing effects and real
-
time skin rendering.

A method of ambient
occlusion

suitable for highly dynamic motion will be proposed, able to capture
detailed features of

characters. The skin rendering method will focus on a
non
-
physically
-
based texture blurring

method with adaptive abilities. The skin
re
ndering method will be enhanced in the second

year. By analyzing the physically
-

based reflection model, the real
-
time skin rendering

method will be enhanced based
on this physically
-
based model. For the third year, we will

focus on realistic cloth
renderi
ng methods, and propose a module for real
-
time clothrendering.

NSC95
-
22
21
-
E
-
009
-
196
-
MY3 (95R487
-
1)

-----------------------------------------------------------------------------------------------


Title

Displacement
-
Map Based Mesh Representation and Its App
lication to Animated
Meshes

Principal Investigator

Jung
-
Hong Chuang


Sponsor

National Science Council

Keywords

D
isplacement
M
ap
,
M
esh
R
epresentation
,
M
esh
S
implification
,
M
esh

P
arameterization
,
R
eal
-
time

rendering


In this project, we will develop a new m
esh representation in which the global shape
of the mesh will

be represented by a simplified base mesh and the geometry
difference between the original mesh and the

base mesh will be stored in a
displacement map. Such representation method will enable the
real time

rendering of
the given mesh using graphics hardware, and, in the meantime, achieve the realistic
rendering

by recovering the geometry details using displacement map. To derive the
proposed representation for a

given mesh, we will develop a simpli
fication
mechanism that takes into account the requirements for a valid

displacement map and
is supported by a new simplification error metric. Since the quality of the

displacement map directly affects the quality of rendering, we will develop a new
param
eterization scheme

based on texture adaptation and level
-
of
-
detail modeling.
Moreover, signal of geometry deviation shall be

taken into account in the
parameterization, and user assisted parameterization will be considered to further

improve the quality of

displacement map. Finally, the real time rendering of the
displacement mapping will

be greatly enhanced or modified to meet the need of the
proposed representation. We will also apply the

proposed mesh representation to the
animated models. In the simplif
ication stage, we will take the

animation sequences of
the model into account and produce an optimal base mesh for the animated model.

And during the run
-
time, the distortion and self penetration due to the displacement
mapping will also be

dealt with.


Th
is is a three
-
year project. In the first year, we will develop a simplification
framework for

constructing the proposed mesh representation. We will consider the
constraints necessary for generating a

valid displacement map and develop a new
simplification

error metric. Moreover, we will study how to

extend current real time
rendering of displacement map to account for the more complicated requirements

for the proposed representation. In the second year, we will focus on how to improve
the quality of the

di
splacement map. Firstly, a new parameterization based on texture
adaptation and LOD modeling will be

proposed. Moreover, signal of geometry
deviation will be included in the parameterization, and

user
-
assisted parameterization
will be attempted. Secondly,
a new simplification scheme that fully aims to

produce a well
-
parameterized displacement map will be studied. In the third year, we
will extend the

proposed representation to deal with the animated model.
Simplification scheme will be adapted to the

animat
ed sequence and aims to generate
a good base mesh. We will also try to remove the surface distortion

and self
penetration due to the displacement mapping for animated model.

NSC
96
-
2628
-
E
-
009
-
160
-
MY3(96N456)

------------------------------------------------
-----------------------------------------------


Title

Research of Key Design Techniques for High Performance 3D Graphics

Processing Unit

Principal Investigator

Chung
-
Ping Chung

Sponsor

National Science Council

Keywords

3D
C
omputer
G
raphics
R
endering, GPU, Compilation, Processor


A
rchitecture, and

Data

B
uffering/
S
torage
.



3D computer rendering techniques are progressing very rapidly. The 3DMark06
benchmarks have

used extensive HDR

high dynamic range techniques and test
items revealed in Microsoft Shader Model

3.0. All these indicate that GPU (gr
aphics
processing unit) architecture and its related processing will

continue to be important
research issues. With users’ increasing demands for finer details and more

realism,
functionalities of a shading system must keep expanding. On the hardware side,

the
rendering

system must be able to process more vertices and fragments, complex
multi
-
threaded/multi
-
pass rendering

programs, more HDR data and computations, and
more intermediate values; furthermore, floating
-
point

support for HDR and

conditional instr
uctions for Shader Model 3.0 are becoming necessary. On the

software side, it must face issues such as thread/pass partitioning, instruction
-
level
parallelism exploitation,

register usage, and data compression. This proposal bases on
the rendering pipeline

flow and explores the

important software/hardware 3D graphics
rendering techniques, in a three
-
year research effort.



The objective of this three
-
year research proposal is to explore key 3D graphics
rendering techniques.

We categorize such techniques i
nto three domains: compilation
techniques (before
-
program
-
execution

processing), processor architecture techniques
(during
-
program
-
execution processing), and data

buffering/storage techniques. Each
contains a number of research topics, as described below:

Domain 1: Compilation techniques

Conventional difficulties faced with a compiler
for 3D rendering

include limited hardware resources, fluctuating code or instruction
execution times, and poor register usages.

We intend to study the following topics:

Topic
1: Instruction scheduling techniques for efficient pixel shader (most of its
contents)

Topic 2: Register usage techniques in pixel shader (most of its contents).

Domain 2: Processor architecture techniques

Features observed during the rendering
process inc
lude vast

amount of computation, vast amount of data, sharply varying
needs for hardware resources, and sharply

varying operation latencies. We intend to
study the following topics:

Topic 1: Instruction scheduling techniques for efficient pixel shader (som
e of its
contents)

Topic 2: Register usage techniques in pixel shader (some of its contents)

Topic 3: Reconfigurable shader design and vertex/pixel load balancing techniques

Topic 4: Reconfigurable texture filter design and load balancing techniques

Topic
5: Study of data inconsistency due to pre
-
Z test and use of hierarchical Z buffer.

Domain 3: Data buffering/storage techniques

In a rendering pipeline, data storages
that typically exist

include processor register file, Z buffer, texture memory, frame
buff
er, and stencil buffer. Furthermore, to

facilitate multi
-
threading, multi
-
pass
execution, deferred shading, etc., large amounts of intermediate data

emerge.
Additional storages (such as F
-
buffer and G
-
buffer) are hence mandatory, require
delicate

managemen
t, and must be integrated into the system properly. We intend to
study the following topics:

Topic 6: Fast indexing technique for variable
-
length data compression

Topic 7: Data and intermediate value management techniques in graphics rendering
pipeline


To
pic 8: Design of a sharable and stretchable register file


We hope to extend our expertise in computer architecture and compilation
techniques to explore these

key software/hardware techniques in 3D graphics
rendering. Achieving the goals set forth in thi
s proposal

will enable 3D rendering to
be faster, more realistic, less costly, and more flexible. It is expected that upon

successful completion of projects of such kind, our domestic capacity in graphics
processor designs and

business revenues can be grea
tly upgraded. The techniques to
be developed will also facilitate the

realization of digital life and medical caring. This
proposal complies with the both essential industrial and

research trends, and its
execution will certainly produce a good number of u
rged quality researchers in

digital design and SoC development.

NSC95
-
22
21
-
E
-
009
-
066
-
MY3 (95R461
-
1)

-----------------------------------------------------------------------------------------------


Title

Waterfall Execution Processor Model: Achieving Out
-
of
-
Order Execution

Performance with

In
-
Order Control Circuit Complexity

Principal Investigator

Chung
-
Ping Chung



Sponsor

National Science Council

Keyword

Processor Execution Model, Control Circuit Complexity, Instruction
-
Level


Parallelism, Clock R
ate


Out
-
of
-
order execution model achieves high instruction
-
level parallelism but
control

complexity impedes performance. Major performance hurdles include large
issue
-
window

size, quadratic complexity growth of data dependency check, and
increased wiring
delays due

to difficult floor
-
planning. We propose a waterfall
execution processor model that achieves

out
-
of
-
order performance but only with
in
-
order control complexity.


Waterfall execution model concatenates a series of single
-
layer data
-
driven
executio
n

stages, number of stages being arbitrary. A basic block is sent down the
stages, batch of

available instructions are executed in current stage, unleashing further
instructions to be

executable in subsequent stage. Early stages are accelerated because
the

waiting instructions

are skipped, and late stages are accelerated because some
instructions have been executed in

early stages. With simple,
always
-
advance
-
one
-
stage instruction dispatching rule, no waterfall

stall will ever
occur. Ideally, with sufficien
t stage depth, a basic block can be executed in a

number
of cycles equal to its critical path depth.


Current semiconductor technology offers abundant circuit capacity, encouraging

architectures that utilize more transistors. And the beauty of waterfall ex
ecution
model lies in

its extremely regular circuit structure, and very simplified and
streamlined control. This

three
-
year project will study waterfall model and related
design issues. The study will include

formal description of waterfall execution model
,
parameter and design considerations,

implementation issues, performance evaluation
and calculation, software environment, and

possible extensions. We will develop
open
-
source simulator and tool chain, which will serve

as the evaluation and
development pl
atform and will promote the deployment and following

research of the
model. This model, if proved working, will be the simplest design providing

the
equivalent performance. The modularized design concept can apply to both

high
-
performance and embedded proc
essors, and will be the best processor model in

many
-
core era.

NSC

97
-
2221
-
E
-
009
-
058
-
MY3
(
97R218
)

-----------------------------------------------------------------------------------------------


Title

The
S
tudy of
I
mage
R
etrieval by
N
eural
N
etworks (V): The

C
onstruc
tion of
Web 2.0 Image Query and

Retrieval System

Principal Investigator

Hsin
-
Chia Fu


Sponsor

National Science Council

Keywords

Neural Network,
D
istributed
I
mage
R
etrieval,
I
mage
C
lustering,
D
istributed
V
ideo
R
etrieval,
Image
T
agging


The growth o
f the Internet and services has caused a corresponding explosion in the
amount

of media data that needs to be archived. The most important issue of
web
-
based search is

how to retrieve the desired media efficiently and correctly. In our
previous related pro
jects,

we proposed a novel image index called
Visual keyword
and
built an image search engine to

retrieve the Corel gallery images. The experiment
results show that the hit rate of the

retrieved images is about 90%. Then, we built a
web
-
based image search
engine, which

contains 20 PC based servers, to retrieve about
45,000 Internet images. The hit rate of the

retrieved images is about 60%, and each
query and retrieval needs 6 seconds in average.

Since WWW is very dynamic, where
users may come and go anytime
, and users’ interests

may change as time went by.
The accuracy of image query varies largely without any rules

or regularity. In this
proposal, we plan to develop Web 2.0 based image tagging technologies

to
incorporate the current visual keyword based ima
ge retrieval prototype system for

accurate and faster image search. Tag based image retrieval has been available for
long time

before the CBIR was proposed and developed. However, traditional manual
tagging is a

label and time consuming task. Thus, we plan

to develop a Web 2.0 based
system, so that

Web browsers can assist the tagging and labeling of their favored
images. Then, we will

develop a machine learning scheme to propagate the “manual
tagging” results to labeling

more images on the Internet and/or W
WW. When more
and more images are tagged with

proper labels, then more content based image
retrieval work can be replaced by the tag based

image retrieval work. Thus, the image
retrieval speed can be greatly improved. On the

system development part of this

proposal, we plan to extend the current distributed image

retrieval prototype system to
a more realistic system, which will contains 50 PC based

servers, and optimally
reaches an average retrieval time to be less than one second.


NSC

97
-
2221
-
E
-
009
-
146
(
97R
396)

-----------------------------------------------------------------------------------------------


Title

Adaptive P2P
-
CDN Layered Streaming over Wired/Wireless Networks

Principal Investigator

Hsu
-
Feng Hsiao

Sponsor

National Science Council

Keywords

C
ontent Distribution Network, Peer
-
to
-
Peer Network, Source Coding,
Video Layered

Streaming, Channel Coding


Multimedia streaming, such as interactive IP television (IPTV), is a natural

consequence of global

convergence of video/audio/data coding and network

technology. However, the massive resource

requirements of streaming server and
network for a large
-
scale streaming application impose great

challenges for a system
designer with quality
-
of
-
service in mind.


Moreover, in a wireless

communication network,
dynamic channel fluctuation
often causes bit errors and subsequently leads to

packet loss, which has distinct
characteristics from the packet loss due to network congestion. We have

been
proposing that different strategies are required for different types
of multimedia
packet loss that

results in performance degradation for efficient streaming. Scalable
coding, on the other hand, is a

popular research direction to deal with heterogeneous
client environments but it certainly complicates

the design of streami
ng system
architecture.


The objective of this research project is to study and

develop a hybrid streaming
system based on extended content distribution network and peer
-
to
-
peer

network to
provide load balancing and address scalability issue for a real
-
ti
me multimedia
streaming

system. We also plan to study joint source coding, including scalable
coding, and channel coding so that

each client may acquire appropriate multimedia
streaming service according to the dynamic

heterogeneous wired/wireless channels
.

NSC95
-
22
21
-
E
-
009
-
077
-
MY3 (95R469
-
1)

-------------------------------------------------------------------------------------------------------


Title

Joint Video
-
Network Coding Techniques over P2P Streaming Platform

Principal Investigator

Hsu
-
Feng Hsiao

Sponsor

National Science Council

Keywords

Joint Video
-
Network Coding, Network Coding, Scalable Video Coding,


Channel

Coding, IPTV, P2P Network


In a peer
-
to
-
peer live media stream
ing platform over wired and wireless
environments, it is

rather challenging to fulfill the needs for Quality of Service and
Quality of Experience

requirements due to frequent topology updates of P2P overlay
network and dynamic channel

fluctuation. The requ
ired techniques are also the key to
the success of such entertainment

platforms. Media streaming if entertainment
systems, such as IPTV which is being

standardized by ITU
-
T and social networked
streaming, like YouTube, have been an emerging

research and in
dustrial emphasis
due to the great progress of network communications and

joint multimedia/channel
coding technologies. The major objective of this research project is

to study and
develop Joint Video
-
Network Coding technologies in order to increase effect
ive

streaming throughput, to minimize the streaming delay, and to mitigate packet loss
over

heterogeneous networks. The proposed network coding is in particular to be
combined with

scalable video coding such that every client can download and upload
networ
k coded blocks

of appropriate streaming contents based on localized channel
conditions with the

consideration of quality of service and quality of experience, even
when a client surfs between

video channels.

NSC
97
-
2221
-
E
-
009
-
054
-
(
97R279
)

-----------------
------------------------------------------------------------------------------


Title

The
S
tudy of 8x8 MIMO
-
OFDM Anti
-
Impulse Jamming System with
All
-
Digital Synchronizations

and Doppler Channel Compensations

Principal Investigator

Terng
-
Yin Hsu


Sponsor

National Science Council

Keywords




Large amounts of high speed and low power mem
ories are indispensable for

multi
-
core platform and multi
-
system emerging. These memories should be able to
support

diverse systems, therefore, an on
-
demand memory system is proposed. This
on
-
demand

memory system provides high bandwidths and low power acce
sses for a
multi
-
core

platform by the memory management unit. On
-
demand memory system
includes

distributed memories, shared memories, memory bus, and three
-
layer
memory

management unit. Besides, we will establish e
-
Home II (eH
-
II) platform
which is a

multi
-
core platform with sub
-
project 2. We focus on the on
-
chip
communication, data

transfer, interfaces and protocols. With circuit and architecture
designs, we will propose

dynamic scheduling mechanisms for memory allocation,
bandwidth, and on
-
chip data

commu
nication. In order to have better power control,
we will provide novel power

management unit and integrate it with the memory
management unit. We will have further

investigation on low power multi
-
thread
register file and will provide novel circuit

structu
re to reduce the area and power
consumption of register files. Finally, we will use

System C to construct our proposed
on
-
demand memory system hardware model, so that

to reduce integration and
verification period and increase reusability. In the first year
, we

will develop key
elements of memory units and complete the structure of multi
-
level

memory
management unit. In the second year, we will complete the multi
-
thread register

file
to provide multi
-
core units to access register files in high speed, and com
plete

communication between memory management units and memory bus. In the third
year, we

will reference protocols of other sub
-
projects to complete the final detailed
structure of

memory system, and a built
-
in
-
self
-
test circuit and a built
-
in
-
self
-
repair
circuit will be

included. Moreover, we will focus on different memory demands for
heterogeneous

systems and construct memory compiler, so to complete the memory
system.

NSC
97
-
2220
-
E
-
009
-
016
-
(
97R269
)

--------------------------------------------------------
---------------------------------------


Title

Secondary Structure Predict
ion and Cluster Analysis of RNA

Principal Investigator

Yuh
-
Jyh Hu

Sponsor

National Science Council

Keywords

RNA,
S
econdar
y
S
tructure
E
lement,
C
lustering



It has recently become clear that RNA molecules play significant roles in diverse
cellular

processes such as protein translocation, splicing, translation quality control,
etc. Most of the

biological activities are accompl
ished by the interaction of specific
RNA sequences and

structures with enzymes and regulatory proteins. In this project,
we propose a new topic for

RNA research. For a given set of unaligned RNA
sequences, our goal is to simultaneously

cluster the RNAs int
o families and predict
the signature structures for each family. We

hypothesize that each family shares
common and characteristic secondary structure elements,

and thus we use these
elements as our basis to discover distinguishable families within a

given
set of RNAs.
Besides, in order to maintain the flexibility, we do not presume the

number or the size
of families, but instead we adopt an iterative approach to find them. Using

a
supervised learner, we first predict the structure elements. Then we separate

out the
RNAs

that contain the elements. With a carefully
-
designed self
-
adjusting method for
family size

estimation, the same procedure can be applied to the remaining RNAs till
the given set of

RNAs is emptied. In the course of the process, we can not onl
y
identify the families in a set of

RNAs, but also the signature structure elements that
characterize these families found.

NSC
97
-
2221
-
E
-
009
-
131
-
(
97R
390)

-----------------------------------------------------------------------------------------------


Titl
e

Study on Efficient and Energy
-
Conserving kNN Query Processing Systems in


Mobile

Environments

Principal Investigator

Jiun
-
Long Huang

Sponsor

National Science Council

Keywords

NN
Q
uery,
K
-
NN
Q
uery,
S
patial
D
atabase,
M
obile
C
omputing


In most mobile
services, location
-
dependent queries are deemed killer applications of
next generation mobile services.
According to spatial constraints of queries,
location
-
dependent queries can be divided into several categories

including proximity
query and k nearest n
eighbor query (referred to as k
-
NN query). A proximity query to

find all objects within a certain range. An example proximity query is “Find all taxis
with distance less than 300

meters to me
.
” Proximity queries are also known as range
queries or window qu
eries. A k
-
NN query is to find k

nearest objects to a specific
location (referred to the query location of the k
-
NN query). For example, a user may

issue a k
-
NN query like “Find the five nearest hotels”. A nearest neighbor query
(referred to as NN query) w
hich

finds the nearest object to a specific location is a
special case of k
-
NN queries with k = 1.

For example, passengers will query the
nearest hotels for taking a rest. In addition, drivers will query the k nearest gas
stations for refueling. Research w
ork shows that deploying proxies supporting NN
and k
-
NN queries is able to greatly reduce system response time and system workload
at the cost of slightly deployment cost. In this project, we aim to provide a scalable
and energy
-
efficient system architectu
re supporting NN and k
-
NN queries. In the first
year, we will focus on NN queries. Specifically, we will first develop a 3
-
tier system
architecture for NN services. We will develop an effective algorithm to speedup the
growth of estimated valid region. In
addition, cache management algorithms will also
be developed. In the second year, we will concentrate on the issues of k
-
NN queries.
We will first extend the proposed architecture for supporting k
-
NN queries. In
addition, the devised cache management algor
ithm will also be extended for k
-
NN
queries. We will also address some problems dedicated for k
-
NN queries. Then, we
will develop an experimental platform. Finally, we will implement the devised
algorithms on the experimental platform to perform performanc
e

evaluation.

NSC
96
-
2221
-
E
-
009
-
161
-
MY2 (
96N
308)

-----------------------------------------------------------------------------------------------


Title:

Simulated Annealing Parameter Detection System for Object Detection and


Seismic Applications(II)

Principal Investigator:

Kou
-
Yuan Huang

Sponsor:

National Science Council

Keywords:

Simulated
A
nnealing,
G
lobal
O
ptimum,
N
e
ural
N
etworks,
Hough
T
ransform,
S
eismic patterns.


Recently the development of theory and application of neural networks become

increas
ingly important in international community. And the areas of the applications
are quite

wide spread. In this year (2007), our project is supported by the National
Science Council. We

propose the pattern detection system using simulated annealing
that can d
etect the parameters

of the lines, circles, ellipses, and hyperbolas. The
system can not only detect parameters with

a higher precision, but also a globally
optimal solution rather than a locally optimal solution

by the Hough transform neural
network. But
it can not synchronously detect ellipses and

hyperbolas due to the
limitation of the used formulas of patterns. So we propose the

synchronous system to
detect ellipses and hyperbolas. We also propose the hierarchical

system to reduce the
number of paramete
rs in computation, and to detect the large number of

patterns.
Traditionally, Hough transform needed the large memory space, but the two

proposed
systems of simulated annealing can reduce the requirement of large memory. In the

first year we use the synchr
onous system to detect patterns. In the next year, we use
the

hierarchical system to detect patterns. After the success of both systems in image
pattern

detection, we will apply it to detect the parameters of the line of direct wave
and the

hyperbola of re
flection wave in the simulated one
-
shot seismogram and real
seismic data. The

detection results of both systems will improve the seismic
interpretations and the further

seismic data processing.

NSC
97
-
2221
-
E
-
009
-
149
-
(
97R398
)

-------------------------------
----------------------------------------------------------------


Title

Interactive Online Test and Judge System for Program Development

Principal Investigator

Shih
-
Kun Huang

Sponsor

Nationa
l Science Council

Keywords



This project will integrate research results of programming languages and

software engineering, for an education platform to support learning process

of
program development. Due to the declined programming skills of ordinary

st
udents,
during the lectures, we lack necessary training platform to satisfy

the three different
roles of teachers, teaching assistant(TA), and students.In general, teachers must design
exercise to fix the lecture contents, and the

TAs must be able to judge

easily and
grading policy must also be consistent

(even to differentiate if the programming
assignment is original). Moreover,students can only catch the classroom contents,
with clear problem descriptions,and input/output specification (also support visu
al
I/O), to pass the homework

examination within some time period.


We will refer to the ACM programming contest for online judge, along with
wargame

systems for computer security, and former development of automatic testing
and

systematic debugging, integ
rating with program similarity tools to be an

interactive programming learning platform. The platform will support teacher

to issue
problems with template, online judging the submitted homework, not

only reply pass
or failure message, also assist in testin
g and provide debugging

information and
visualize the run
-
time information, to proceed the homework

interactively.

NSC
97
-
2221
-
E
-
009
-
074
-
(
97R282
)

-----------------------------------------------------------------------------------------------


Title

Key Tec
hniques for Mobile Search over Heterogeneous Networks(I)(II)(III)

Principal Investigator

Rong
-
Hong Jan


Sponsor

National Science Council

Keywords

Web 2.0,
C
ommunity
M
ining and
S
equential
P
attern
M
ining, VANET,
DSRC, MAC,

C
lustering,
P
ower
A
ssignment,
C
arry

and
F
orward,
Wireless Sensor
N
etwork,
M
obility,
L
oad

B
alancing,
F
ault
T
olerance,
GPS,
P
ersonal
N
avigation
S
ystems,
I
ndoor
P
ositioning,
P
edometer,

A
ccelerometer,
M
agnetic
C
ompass,
G
yro,
W
alking
P
ath
T
racking.


The mobile web search technologies have drawn
continuing attention in recent
years.

However, due to the limit capability and recourses, the mobile devices can not
perform well in

heterogonous networks environment. For this reason, in this project,
we intend to investigate

key techniques for the mobile

search application and network
platform, including five

subprojects 1) MobiConnectivity; 2) MobiTransporting; 3)
MobiSensing; 4) MobiPositioning;

5) MobiSearching. These techniques will be
integrated in our main project.


In Subproject I, we intend to dev
elop Media Access Control (MAC) and
connectivity

mechanisms to support the Vehicular Ad
-
hoc Networks (VANET). It is a
three
-
year project: In

the first year, we will design the MAC protocols based on the
Dedicated Short Range

Communication (DSRC) standard,
which is expected to be
wildly used in the near future. In

the second year, the design issue on clustering and
power assignment will be investigated for

very high speed vehicle. Based on these
grounds, in the last year, we will propose the carry

and forwar
d protocols to overcome
the long latency due to the intermittent connectivity in

vehicular networks.


In Subproject II, the objective is to develop the transportation techniques which
provide

the fast, reliable and efficient information delivery over VANET
. The topics
of this research

include (1) Fast information fusion, dissemination, replication and
storage in VANET. (2)

Since VANET inherits some characteristics of Delay
Tolerance Network (DTN), we will

discuss the TCP issues over DTN and our
transport la
yer solution to achieve reliable

end
-
to
-
end information transportation. (3)
Finally, we will design a multicast protocol for

VANET to achieve efficient
information forwarding.


In Subproject III, the goal is to provide and maintain the connectivity of wire
less
sensor

networks (WSNs) for mobile positioning and services. Corresponding research
items of this

subproject include (1) dynamic WSN configuration and maintenance,
and (2)

powered

node
-
aware WSN configuration and maintenance. The first research
item co
nsists of

four sub
-
items: mobility
-
aware ID assignment and routing
mechanism, single node and group

mobility support, dynamic load balancing scheme
and automatic failure recovery. The second

research items comprise three sub
-
items:
powered
-
node
-
ware router

delegation,

powered
-
node
-
ware path discovery, and
powered
-
node
-
ware mobility support and failure

recovery.


In Subproject IV, we would like to use sensors embedded in hand held devices
like

accelerometers, gyroscopes, magnetometers, etc., to collect movem
ent
information. By

combining the analysis results of sensor readings and indoor or
outdoor positioning systems,

we would like to achieve following four objectives: (1)
Improve the accuracy of GPS by

utilizing low cost sensors; (2) Improve the accuracy
of
indoor positioning systems; (3)

Develop group navigation systems by applying the
wireless communication; (4) Based on the

technology developed for personal
navigation systems, we would like to develop equipments

that can record sensor
readings such that we

can reconstruct the movement of cars.


In Subproject V, we intend to develop an intelligent search services for mobile
devices.

Our primary goals include (1). discover Geo
-
Entity (referred to Geographical
Entity) from a

huge Web 2.0 repository. (2). mine
popular co
-
cited communities in a
blogspace. (3). Develop

mining algorithms to mine sequential patterns across multiple
attributes
for

NSC
97
-
2221
-
E
-
009
-
048
-
MY3
(
97R210
)

---------------------------------------------------------------------------------------
--------


Title

MobiConnectivity: Medium Access Control and Connectivity Supports for
Vehicle Ad
-
Hoc

Netowrks(I)(II)(III)

Principal Investigator

Rong
-
Hong Jan


Sponsor

National Science Council

Keywords

VANET, DSRC, MAC, Clustering, Power Assignment, Carry and
Forward


Du
e to the dramatic cost down on electronic components and the advances on
wireless

technologies, the development of Intelligent Transportation System (ITS) has
drawn intensive

attention in recent years from many countries. The Vehicular Ad hoc
Network (VANE
T) is a

promising approach for the future ITS. In this novel network
architecture, a communication is

either enable between the vehicle and road side unit
(RSU) or between any two cars in

multihop fashion. In other words, the deployment is
flexible and the

network coverage can be

easily extended to wild range.


In this subproject, we intend to develop Media Access Control (MAC) and
connectivity

mechanisms to support the VANET. It is a three
-
year project: In the first
year, we will design

the MAC protocols b
ased on the Dedicated Short Range
Communication (DSRC) standard,

which is expected to be wildly used in the near
future. In the second year, the design issue on

clustering and power assignment will
be investigated for very high speed vehicle. Based on

thes
e grounds, in the last year,
we will propose the carry and forward protocols to overcome