To improve speech recognition performance, feature transformation based on discriminant analysis has been widely used to reduce the redundant dimensions of acoustic features. Linear discriminant analysis (LDA) and heteroscedastic discriminant analysis (HDA) are often used for this purpose, and a generalization method for LDA and HDA, called power LDA (PLDA), has been proposed. However, these methods may result in an unexpected dimensionality reduction for multimodal data. It is important to preserve the local structure of the data when reducing the dimensionality of multimodal data. In this paper we introduce two methods, locality-preserving HDA and locality-preserving PLDA, to reduce dimensionality of multimodal data appropriately. We also propose an approximate calculation scheme to calculate sub-optimal projections rapidly. Experimental results show that the locality-preserving methods yield better performance than the traditional ones in speech recognition.

Due to the development and popularization of high-definition televisions, digital video cameras, Blu-ray discs, digital broadcasting, IP television and so on, it plays an important role to identify and quantify video quality degradations. In this paper, we propose SV-CIELAB which is an objective video quality assessment (VQA) method using a spatio-velocity contrast sensitivity function (SV-CSF). In SV-CIELAB, motion information in videos is effectively utilized for filtering unnecessary information in the spatial frequency domain. As the filter to apply videos, we used the SV-CSF. It is a modulation transfer function of the human visual system, and consists of the relationship among contrast sensitivities, spatial frequencies and velocities of perceived stimuli. In the filtering process, the SV-CSF cannot be directly applied in the spatial frequency domain because spatial coordinate information is required when using velocity information. For filtering by the SV-CSF, we obtain video frames separated in spatial frequency domain. By using velocity information, the separated frames with limited spatial frequencies are weighted by contrast sensitivities in the SV-CSF model. In SV-CIELAB, the criteria are obtained by calculating image differences between filtered original and distorted videos. For the validation of SV-CIELAB, subjective evaluation experiments were conducted. The subjective experimental results were compared with SV-CIELAB and the conventional VQA methods such as CIELAB color difference, Spatial-CIELAB, signal to noise ratio and so on. From the experimental results, it was shown that SV-CIELAB is a more efficient VQA method than the conventional methods.

Silicon-quantum-dots (Si-QDs) with an areal density as high as 1012 cm - 2 were self-assembled on thermally-grown SiO2 by low pressure CVD using Si2H6, in which OH-terminated SiO2 surface prior to the Si CVD was exposed to GeH4 to create nucleation sites uniformly. After thermal oxidation of Si-QDs surface, two-dimensional electronic transport through the Si-QDs array was measured with co-planar Al electrodes evaporated on the array surface. Random telegraph signals were clearly observed at constant applied bias conditions in dark condition and under light irradiation at room temperature. The result indicates the charging and discharging of a dot adjacent to the percolation current path in the dots array.

We investigated the influence of the thickness of the AlN interlayer for InAlN/GaN and InAlN/AlGaN/GaN heterostructures. The AlN thickness strongly affects the surface morphology and electron mobility of the InAlN/GaN structures. The rms roughness of the surface increases from 0.35 to 1.2 nm with increasing AlN thickness from 0 to 1.5 nm. Large pits are generated when the AlN is thicker than 1 nm. The highest electron mobility of 1470 cm2/VS is obtained for a 0.75-nm-thick AlN interlayer. The mobility, however, becomes lower with increasing deviation from 0.75 nm. It is only 200 cm2/VS for the 0-nm thick AlN. Inserting AlGaN between AlN and InAlN suppresses the influence of the AlN interlayer thickness. A smooth surface with rms roughness of 0.35 nm is obtained for all samples with 0-1.5-nm-thick AlN. The electron mobility ranges from 1000 to 1690 cm2/VS. The variation is smaller than that for InAlN/GaN. We fabricated field effect transistors (FETs) with gate length of 2 µm. The electron mobility in the access region affects the transconductance (gm) of FETs. As a results, the influence of the AlN thickness for InAlN/GaN FETs is larger than that for InAlN/AlGaN/GaN FETs, which reduces gate leakage current. The transconductance varies from 93 to 235 mS/mm for InAlN/GaN FETs. In contrast, it varies from 180 to 230 mS/mm for InAlN/AlGaN/GaN FETs. These results indicate that the InAlN/AlGaN/GaN heterostructures could lead to the development of GaN-based FETs.

We have studied transmission property of electron in vertical MOSFET (V-MOSFET) from the viewpoint of quantum electro-dynamics. To obtain the intuitive picture of electron transmission property through channel of the V-MOSFET, we solve the time-dependent Schrodinger equation in real space by employing the split operator method. We injected an electron wave packet into the body of the V-MOSFET from the source, and traced the time-development of electron-wave function in the body and drain region. We successfully showed that the electron wave function propagates through the resonant states of the body potential. Our suggested approaches open the quantative and intuitive discussion for the carrier dynamics in the V-MOSFET on quantum limit.

We fabricated and examined current limiting effect for InP Gunn diodes with stable depletion layer mode operation of diodes for high efficiency Gunn oscillators. Current limiting at the cathode was achieved by a shallow Schottky barrier at the interface. We discussed fabrication procedure, the results for negative differential resistance and rf tests for InP Gunn diodes. It was shown that the fabricated Gunn diodes have the output power of 10.22 dBm at a frequency of 90.13 GHz. Its input voltage and corresponding current were 8.55 V and 252 mA, respectively.

This letter proposes closed form solutions to the L2-sensitivity minimization of second-order state-space digital filters with real poles. We consider two cases of second-order digital filters: distinct real poles and multiple real poles. In case of second-order digital filters, we can express the L2-sensitivity of second-order digital filters by a simple linear combination of exponential functions and formulate the L2-sensitivity minimization problem by a simple polynomial equation. As a result, the minimum L2-sensitivity realizations can be synthesized by only solving a fourth-degree polynomial equation, which can be analytically solved.

Hierarchical constellations offer a different property of robustness to the multiple bits that construct a symbol according to channel errors. We apply the characteristics of hierarchical constellations to a multi-user cellular system that has limited modulation levels, in order to improve cell capacity. We propose an adaptive resource allocation scheme based on the hierarchical constellation in which a symbol is shared by multiple users and each bit in a symbol is allocated adaptively according to the channel condition of each user. The numerical results show that the proposed resource allocation scheme provides mobile users with higher modulation levels so that the cell capacity is improved.

The manner of a person's eye movement conveys much about nonverbal information and emotional intent beyond speech. This paper describes work on expressing emotion through eye behaviors in virtual agents based on the parameters selected from the AU-Coded facial expression database and real-time eye movement data (pupil size, blink rate and saccade). A rule-based approach to generate primary (joyful, sad, angry, afraid, disgusted and surprise) and intermediate emotions (emotions that can be represented as the mixture of two primary emotions) utilized the MPEG4 FAPs (facial animation parameters) is introduced. Meanwhile, based on our research, a scripting tool, named EEMML (Emotional Eye Movement Markup Language) that enables authors to describe and generate emotional eye movement of virtual agents, is proposed.

We designed and fabricated 4H-SiC PIN avalanche photodiodes (APD) for UV detection. The thickness of an intrinsic layer in a PIN structure was optimized in order to achieve the highest quantum efficiency at the wavelength of interest. The optimized 4H-SiC PIN APDs exhibited a maximum external quantum efficiency of >80% at the wavelength of 280 nm and a gain greater than 40000. Both electrical and optical characteristics of the fabricated APDs were in agreement with those predicted from simulation.

In this paper, an efficient radio resource allocation scheme for OFDMA systems is proposed, which follows two steps to take care of real-time traffic characterized with multi-level delay constraints. Urgent packets, those with imminent deadlines, are released first in step 1. After that the remaining channel resources are managed in such a way that overall throughput is maximized at Step 2. In this work, 2-dimensional diversity over multiple sub-bands and multiple users are jointly considered. The proposed scheme is compared with existing schemes designed for real-time traffic such as Exponential Scheduling (EXP) scheme, Modified Largest Weighted Delay First (M-LWDF) scheme, and Round robin scheme in terms of the packet discard probability and throughput. Numerical results show that the proposed scheme performs much better than the aforementioned ones in terms of the packet discard probability, while slightly better in terms of throughput.

Packet classification has become one of the most important application techniques in network security since the last decade. The technique involves a traffic descriptor or user-defined criteria to categorize packets to a specific forwarding class which will be accessible for future security handling. To achieve fast packet classification, we propose a new scheme, Hierarchical Cross-Producting. This approach simplifies the classification procedure and decreases the distinct combinations of fields by hierarchically decomposing the multi-dimensional space based on the concept of telescopic search. Analogous to the use of telescopes with different powers**, a multiple-step process is used to search for targets. In our scheme, the multi-dimensional space is endowed with a hierarchical property which self-divides into several smaller subspaces, whereas the procedure of packet classification is translated into recursive searching for matching subspaces. The required storage of our scheme could be significantly reduced since the distinct field specifications of subspaces is manageable. The performance are evaluated based on both real and synthetic filter databases. The experimental results demonstrate the effectiveness and scalability of the proposed scheme.

Differential-phase-shift (DPS) quantum key distribution (QKD) is one scheme of quantum key distribution whose security is based on the quantum nature of lightwave. This protocol features simplicity, a high key creation rate, and robustness against photon-number-splitting attacks. We describe DPS-QKD in this paper, including its setup and operation, eavesdropping against DPS-QKD, system performance, and modified systems to improve the system performance.

We propose a call admission control scheme in cellular and wireless local area networks (WLANs) integration: integrated service-based admission control with load-balancing capability (ISACL). The novel aspects of the ISACL scheme include that load transfer in the cellular/WLAN overlapping areas is allowed for the admission of originating data calls from the area with cellular access only and vertical handoff requests to the cellular network. Packet-level quality of service (QoS) constraints in the WLANs and other-cell interference in the code division multiple access (CDMA) cellular network are taken into account to derive the WLANs and cellular capacity. We model the integrated networks using a multi-dimensional Markov chain and the important performance measures are derived for effective optimization of the admission parameters. The analytical model is validated by a computer simulation. The variation of admission parameters with traffic load and the dependence of resource utilization on admission parameters are investigated. It is shown that optimal balancing of the traffic load between the cellular network and WLANs results in the maximum resource utilization. Numerical results demonstrate that substantial performance improvements can be achieved by applying the proposed ISACL scheme.

The proportionate normalized least mean square algorithm (PNLMS) greatly improves the convergence of the sparse impulse response. It exploits the shape of the impulse response to decide the proportionate step gain for each coefficient. This is not always suitable. Actually, the proportionate step gain should be determined according to the difference between the current estimate of the coefficient and its optimal value. Based on this idea, an approach is proposed to determine the proportionate step gain. The proposed approach can improve the convergence of proportionate adaptive algorithms after a fast initial period. It even behaves well for the non-sparse impulse response. Simulations verify the effectiveness of the proposed approach.

Owing to their ultra-wideband characteristics, tapered slot antennas (TSAs) are used as element antennas in wideband phased arrays. However, when the size of a TSA is reduced in order to prevent the generation of a grating lobe during wide-angle beam scanning, the original ultra-wideband characteristics are degraded because of increased reflections from the ends of the tapered slot aperture. To overcome this difficulty, we propose a new antenna structure in which parallel-plate waveguides are added to the TSA. The advantage of this new structure is that the reflection characteristics of individual antenna elements are not degraded even if the width of the antenna aperture is very small, i.e., approximately one-half the wavelength of the highest operating frequency. In this study, we propose a procedure for designing the new antenna through numerical simulations by using the FDTD method. In addition, we verify the performance of the antenna array by experiments.

With the process scaling, the leakage current reduction has been the primary design concerns in a nanometer-era VLSI circuit. In this paper, we propose a new lithography process-aware edge effects correction method to reduce the leakage current in the shallow trench isolation (STI). We construct the various test structures to model Ileakage and Ileakage_fringe which represent the leakage currents at the center and edge of the transistor, respectively. The layout near the active edge is modified using the look-up table generated by the calibrated analytic model. On average, the proposed edge effects correction method reduces the leakage current by 18% with the negligible decrease of the drive current at sub-40nm DRAM device.

In this paper, we propose a new subcarrier allocation algorithm for a downlink OFDMA relay network with multicells. In the proposed algorithm, subcarriers are allocated to users and relays to maximize the overall sum of the achievable rate under fairness constraints. Simulation results show that the proposed algorithm achieves higher data rate than the static algorithm and reduces the outage probability compared to the static and greedy algorithms.

Inductive theorem proving plays an important role in the field of formal verification of systems. The rewriting induction (RI) is a method for inductive theorem proving proposed by Reddy. In order to obtain successful proofs, it is very important to choose appropriate contexts (such as in which direction each equation should be oriented) when applying RI inference rules. If the choice is not appropriate, the procedure may diverge or the users have to come up with several lemmas to prove together with the main theorem. Therefore we have a good reason to consider parallel execution of several instances of the rewriting induction procedure, each in charge of a distinguished single context in search of a successful proof. In this paper, we propose a new procedure, called multi-context rewriting induction, which efficiently simulates parallel execution of rewriting induction procedures in a single process, based on the idea of the multi-completion procedure. By the experiments with a well-known problem set, we discuss the effectiveness of the proposed procedure when searching along various contexts for a successful inductive proof.

Graph matching is a NP-Hard problem. In this paper, we relax the admissible set of permutation matrices and meantime incorporate a barrier function into the objective function. The resulted model is equivalent to the original model. Alternate iteration algorithm is designed to solve it. It is proven that the algorithm proposed is locally convergent. Our experimental results reveal that the proposed algorithm outperforms the algorithm in .