博士 / 輔仁大學 / 商學研究所 / 98 / The purpose of this study is to investigate the mean reversion characteristic in firm-specific earning-to-price ratio (E/P ratios), and employ it for investment strategies. Existing studies related to the mean reversion feature in P/E ratio has focus on US and non-US market-wise data, but overlooked individual stocks. The findings and implications might not be generalized insofar as individual stocks. We differentiate our study from other research by analyzing firm-specific time series data. However, P/E ratio will diverge when the denominator (EPS) is close to zero. To overcome the flaw, we adopt E/P ratio in our following tests. The results show that not all E/P time series of individual stocks exhibit a mean reversion feature. Among the 1,156 non-finance firms listed on TSE and OTC in Taiwan in 2006, the E/P ratios of 516 firms (about 45%) exhibit the tendency of mean reversion. In addition, the current study makes an effort to combine the mean reversion theory with investment practice. We design two innovative investment strategies based on the detected mean reversion feature of E/P ratio, and report the dominant investment performances.

碩士 / 國立成功大學 / 微生物及免疫學研究所 / 98 / Patients infected by dengue viruses (DV) may display dengue fever, dengue hemorrhagic fever (DHF) or dengue shock syndrome (DSS). Hemorrhagic syndromes of DHF/DSS include thrombocytopenia, coagulopathy and vasculopathy, which are related with dysfunction of endothelial cells and platelets. Previous studies in our laboratory showed that antibodies (Abs) against DV nonstructural protein 1 (NS1) cross-react with human endothelial cells and induce apoptosis. Platelet dysfunction and bleeding tendency are also induced by anti-DV NS1 Abs. Using sequence homology analysis, we found that the C-terminal region of DV NS1 protein contains cross-reactive epitopes shared with self-antigens. For safety in vaccine development, the cross-reactive epitopes of DV NS1 protein should be deleted or modified. We have generated the chimeric NS1 protein, which consists of N-terminal DV NS1 (a.a. 1-270) and C-terminal JEV NS1 (a.a. 271-352) (designated DJ NS1). The anti-DJ NS1 Abs showed lower binding activity to human endothelial cells and platelets than that of anti-DV NS1 Abs. In the murine model, DV NS1 immunization caused prolonged bleeding time, but DJ NS1 immunization did not. Passive immunization with anti-DV NS1 or anti-DJ NS1 Abs could reduce DV-induced prolonged bleeding time and hemorrhage on the local skin. KU812 basophil/mast cell line was susceptible to DV infection without enhancing Abs. Anti-DJ NS1 Abs induced DV-infected KU812 cells to undergo apoptosis. Moreover, the expression of MIP-1 alpha and MIP-1 beta in DV-infected KU812 cells was also reduced by anti-DJ NS1 Abs. According to these findings, DJ NS1 may provide a strategy for dengue vaccine development.

博士 / 國立交通大學 / 材料科學與工程學系 / 98 / The primary objective of this dissertation is to improve the light output power of light-emitting diodes by mirror structure and surface roughening Process. Effect of thermal annealing of the reflectivity of ITO/Ag mirror during wafer bonding process has been investigated. We presented a new structure of mirror to improve the reflectivity and also the output power of the vertical GaN-based LEDs could be improved. In the experiments of surface roughening process, the p-GaP and n-AlGaInP surface of the AlGaInP-based LEDs would be roughened by inductively coupled plasma (ICP) dry etching and /or chemical wet etching. In addition, the effect of the etching pattern and etching depth on the characteristics of the LEDs would be discussed. In general, GaN epitaxial layers were grown on sapphire substrates. The LEDs epitaxial layer would then transferred to the Si substrate by wafer bonding, since the thermal conductivity of the sapphire is lower than Si. A metal layer usually added between the LEDs and transferred substrate as mirror to avoid the absorption of the generated photons by substrate. In previous studies, because of the high reflectivity, Ag usually was used to serve as mirror layer. And the ITO was used as ohmic contact layer on LEDs. In our experiment, the Ag atoms would diffuse into ITO layer and agglomerated after wafer bonding process. This would decrease the reflectivity of the ITO/Ag mirror. However, the new mirror structure with Ni thin film (2 nm) between interface of the ITO and Ag could improve the reflectivity after the same bonding process. Consequently, comparing with the ITO/Ag mirror, the new mirror structure could enhance the light intensity 87.3% under a 20-mA forward injection current. For the AlGaInP-based LED, vertical AlGaInP-based LED epitaxial films are successfully fabricated on a Si substrate using wafer bonding technology. Furthermore, the new structures of AlGaInP-based LEDs with single (p-GaP) and double roughened (p-GaP and n-AlGaInP) surfaces are fabricated by chemical wet etching and ICP dry etching technologies. It is found that the light output power of the single and double surface-roughened AlGaInP-based LED chips is 2 and 2.56 times higher than that of the conventional AlGaInP-based LEDs (flat p-GaP and n-AlGaInP layer) at an injection current of 20 mA, respectively. Finally, the effects of the different etching-patterns and the etching-depth on the optical and electrical characteristics of the surface-roughened AlGaInP-based LEDs will be investigated.

博士 / 國立交通大學 / 材料科學與工程學系 / 98 / In this thesis, the heterostructure growth of GaN on Si substrate by Low-pressure metal-organic chemical vapor deposition (LP-MOCVD) is studied. The investigation focuses on the growth of GaN on 6 “ Si (111) wafer. There different approaches werw including in this study. In the first approach, high quality GaN film was successfully grown on 150 mm Si (111) substrate by MOCVD method using multilayer AlN combined with the graded AlGaN layer as buffer. The buffer layer structure, film quality and film thickness are critical for the growth of the crack free GaN film on Si (111) substrate. Using multilayer AlN films grown at different temperatures combined with graded Al1-xGaxN film as the buffer, the tensile stress on the buffer layer was reduced and the compressive stress on the GaN film was increased, as a result, high quality 0.5 μm crack-free GaN epitaxial layer was successful grown on 6” Si substrate. In the second aaproach, the GaN film was grown on Si substrate using multilayer AlN/AlxGa1-xN buffer by low pressure metal organic chemical vapor deposition (MOCVD) method. The AlxGa1-xN films with Al composition varying from 0~ 0.66 was used to accommodate the stress induced between GaN and Si substrate during GaN growth. The correlation of the Al composition in the AlxGa1-xN films with respect to the stress induced in the GaN film grown was studied using high resolution X-ray diffraction including symmetrical and asymmetrical ω/2θ scans and reciprocal space maps. It is found that with proper design of the Al composition in the AlxGa1-xN buffer layer, crack-free GaN film can be successfully grown on 6” Si (111) substrates using multilayer AlN and AlxGa1-xN buffer layers Finally, in the third approach, low stress, low defect density GaN film was successful grown on circle array patterned Si (111) substrate using AlN as the nucleation buffer followed by two steps growth of the GaN film. Raman measurement shows a reduction of the in plane biaxial stress for the GaN film grown on patterned substrate. The slight blueshift of the band edge PL peaks further provides the evidence that the tensile stress in the GaN film was relaxed in the patterned Si substrate. It’s believed that the grain boundaries of the polycrystalline AlN buffer layer and the dislocations in the GaN film grown helped to relieve the stress induced by the lattice and the thermal coefficient mismatches during growth.

碩士 / 國立交通大學 / 統計學研究所 / 98 / Comparisons of reference charts for verifying if two populations of subjects have the same growth pattern have received some attention in literature. However, the proposals of comparison are restricted on equalities of regression parameters or regression functions. For public health purpose of comparing growth countries, we consider general theory of equalities of reference charts and establish its relations to equalities of growth model parameters. This approach allows us to display these relations for several interesting longitudinal growth models and these relations show that it is in-appropriate in comparing reference charts by testing equality of regression parameters or regression functions. Finally, we propose an exact test for comparisons of reference charts.

博士 / 國立交通大學 / 資訊科學與工程研究所 / 98 / Traditionally, the production of 2D animation is a labor-intensive artisan process of building up sequences of drawn images by hand which, when shown one after the other one at a fixed rate, resemble a movement. Most work and hence time is spent on drawing, inking, and coloring the individual animated characters for each of the frames. Instead of the traditional animation generated by hand, we introduce a novel method by enhancing still pictures and making characters move in convincing ways. The proposed method is based on the statistical analysis and inference, while minimizing users’ intervention. We adopt nonparametric regression to efficiently analyze the displacements of the pre-sampled data from characters in still pictures and use it to generate 2D character animation directly. Furthermore, 2D character animation is regarded as 3D transformation problem, which consists of a 2D spatial displacement and a 1D shift in time. Hence, we focus on the temporal relationship of different poses of the same character in these still pictures. Time series is applied to analyze the character’s movement and forecast a sequence of the suitable limbs movement of the character. In this dissertation, 2D character animation involves novel view generation, expressive talking face simulation, and limbs movement synthesis. Considering characters in still pictures, we focus on nonparametric regression to generate a novel view and an expressive facial animation synchronized with the input speech of a character. Kernel regression with elliptic radial basis functions (ERBFs) is proposed to describe and deform the shape of the character in image space. Note that the novel parametric representation, ERBFs, can be applied to represent the observations of the shape on the unit ellipse. For preserving patterns within the deformed shape, locally weighted regression (LOESS) is applied to fit the details with local control. Furthermore, time series is used to analyze the limb movement of a character and represent the motion trajectory. Note that a character’s motion could be described by a series of non-continuous poses of a character from a sequence of contiguous frames. According to these poses, we investigate a nonparametric Bayesian approach to construct the time series model representing the character’s motion trajectory. Then we can synthesize a sequence of the motion by using the motion trajectory. Last but not the least, we also investigate how to adopt the proposed statistical approaches mentioned above to animate passive elements. The movements of passive elements involving natural movements that respond to natural forces in some fashion like trees swaying and water rippling could be synthesized. Given a picture of a tree, we make it sway. Given a picture of a pond, we make it ripple. The solutions are developed to animate photographs or paintings effectively. Experimental results show that our method effectively simulates plausible movements for 2D character animation. They also show that the estimated motion trajectory best matches the given still frames. In comparison to previous approaches, our proposed method synthesizes smooth animations, while minimizing unnatural distortion and having the advantages of being more controllable. Moreover, the proposed method is especially suitable for intelligent multimedia applications in virtual human generation. We believe that the provided solutions are easy to use, and empower a much quicker animation production.

博士 / 國立交通大學 / 資訊科學與工程研究所 / 98 / Future optical Metropolitan Area Networks (MANs) have been expected to exploit advanced Optical Packet Switching (OPS) technologies to cost-effectively satisfy a wide range of applications having time-varying and high bandwidth demands and stringent delay requirements. In this thesis, we present a high-performance real-time medium access control scheme for our experimental high-performance OPS metro WDM slotted-ring network (HOPSMAN). HOPSMAN has a scalable architecture in which the node number is unconstrained by the wavelength number. It encompasses a handful of nodes (called server nodes) that are additionally equipped with optical slot erasers capable of erasing optical slots resulting in an increase in bandwidth efficiency. In essence, HOPSMAN is governed by a novel medium access control (MAC) scheme, called Probabilistic Quota plus Credit (PQOC), which is further enhanced with QoS assurance, called Probabilistic Quota plus Credit with QoS Assurance (PQOC/QA). The proposed MAC scheme embodies a highly efficient and fair bandwidth allocation in accordance with a quota being exerted probabilistically. The probabilistic quota is then analytically derived taking the server-node number and destination-traffic distribution into account. Besides, the MAC scheme introduces a time-controlled credit for regulating a fair use of remaining bandwidth particularly in the metro environment with traffic of high burstiness. Moreover, PQOC/QA adopts slot-basis reservation through a simple and flexible marking mechanism to support QoS and to resolve the intrinsic access problem in WDM network. Instead of focusing on estimation of the bandwidth requirements, PQOC/QA sets up real-time connections by employing constant mean rate reservation on each cycle of the ring and effectively accommodates bursty real-time traffic (VBR). Furthermore, we develop a novel approximation to acquire the accurate results of the expected connection setup queueing delay by means of an M/G/m queueing analysis. In the analysis, the maximum admissible quota of real-time traffic is regarded as the number of servers and the service time has a duration that follows an exponential form with an added constant. In M/G/m queueing analysis, the accurate results have only been attained for a limited number of special service distributions, while most of the proposed approximation only maintained a less than 10% relative error for certain properties of service distributions. Our approximation results, which are derived under the particular general service distribution in our system, show that the mean setup queueing time is in profound agreement with the analytic result. Additionally, extensive simulation results show that HOPSMAN with the proposed MAC scheme achieves exceptional delay-throughput performance and remarkable real-time traffic performance (high statistical multiplexing gain for real-time traffic, exceedingly low VBR delay and jitter) under a wide range of traffic loads and burstiness.

博士 / 國立交通大學 / 資訊科學與工程研究所 / 98 / A cloud storage system consisting of a collection of storage servers provides storage services over the Internet for long-term storage. A user can store data into the system and access data from anywhere at any time via the Internet access. However, storing data in a third party's cloud system brings a serious concern on the data confidentiality. We consider a cloud storage system model that has no central authority. A tight integration of public key encryption schemes and random erasure codes is developed. By using this integration, we present a secure cloud storage system, which guarantees the data confidentiality and robustness and supports the secure data forwarding functionality. Hence, in our storage system, a user can not only securely store data but also forward data to other user in a confidential way.

碩士 / 國立交通大學 / 電子研究所 / 98 / 3GPP Long Term Evolution is popular standard of 4G mobile communication technique. But before develop any kinds of communication technique, we must be confirm the channel simulation environment and how to actual modeling channel condition. And the 3GPP SCM standard and WINNER channel model were developed by this purpose. In this thesis, we first present four kinds of popular channel model which were Jakes, SCM, WINNER and Autoregressive (AR) model. We focus on discuss the WINNER channel model, because it was developed by more actual 4G channel environment and could be simulted for twelve secnario. But we realize it would cost a lot of simultion time and complex to generate WINNER channel coefficients. So we discuss the approximation technique to reduce its complex, and we also discuss its statistical properties using the AR model. In system application part, we combined the different block code in IEEE 802.16e PHY layer, in order to fast simulate and verify their code in total system. Besides, we make the simple connection between PHY layer and MAC layer. In other application part, we discuss the reason to affect channel spatial correlation by the WINNER channel.

博士 / 國立交通大學 / 電子研究所 / 98 / The dissertation presents the design of the cascode Class E power amplifier with AM-AM and AM-PM compensation for polar applications. For integration and reliability analysis in CMOS process, the Class E designed with small dc-feed and using cascode topology has been presented. When the Class E is in supply modulation, the AM-AM and AM-PM distortion is introduced at the RF output signal. This work not only analyzes the cause of the distortion but also presents a compensation technique. When modulating the gate bias voltage of the cascode transistor, the transistor operates as a resistance alike to improve the distortion of the amplifier. The experimental result of proposed 2.6 GHz cascode Class E power amplifier shows that when the PA is compensated the AM-PM is reduced from 30° to 6° and the output envelope voltage is linearly to supply voltage in VDD > 0.6 V. The experimental result of the 2.6 GHz cascode Class E power amplifier with self-biased control circuit shows that the voltage slope of AM-AM is 1 V/V and the phase error of AM-PM is 5°. The PA has a output power of 12 dBm, drain efficiency of 17.8% and PAE of 16.6% from a 1.8 V supply and an input driving of 6 dBm. Therefore, the experimental results demonstrate the proposed compensation technique can effectively improve the AM-AM and AM-PM distortion of the cascode Class E amplifiers. In addition, the simulation results of the RF/baseband co-verification platform with OFDM-based signal source show that the EVM is improved from -17 dB to -19.2 dB at 16QAM modulation and from -21 dB to -25.1 dB at 64QAM modulation. Finally, due to the low quality factor of silicon-based inductor causing the degradation on PA efficiency, the PA with multi-metal layer suspended inductors has been presented. The simulation result shows that the PA can have a maximum efficiency improvement of 17%.