[in Japanese]

View full abstract

Present status of III-V channel MOSFET is reported. Progress of Si MOSFET requires successive technology boosters and next candidate as technology booster is high mobility channel. In case of Ge, sufficient contact cannot be obtained in n-MOSFET. Thus, III-V channel became the candidate of high mobility channel in n-MOSFET. At present, interface state density around 1×10¹² eV ^-1cm^-2 by using ALD, high current density > 2 A/mm @V_DD=0.5 V are reported. Formation of InGaAs film on large scale Si wafer is also reported. However, ultrathin channel taking short channel effect into account degrades mobility. Multi-gate FET and improvement of MIS interface are required for realization of high on-current in short channel devices.

View full abstract

This paper demonstrates the impact of tilted Mg ion implantation for the threshold voltage control of GaN MISFETs for the first time. The threshold voltage of the MISFETs by using Mg implantation shifts up to -1 V, whereas that without Mg ion implantation is about -8 V. The GaN MISFET achieved maximum drain current of 165 mA/mm and an extrinsic transconductance of 30 mS/mm. These results indicate a definite availability of our process in normally-off GaN MISFETs for power switching device applications.

View full abstract

This paper describes reliability of AlGaN/GaN heterostructure field-effect transistors fabricated on Si substrates. At an early stage of 50 V high temperature operating life tests, they exhibited a significant lowering of the forward turn-on voltage. The mean time to failure (MTTF) determined by this degradation mode was exponentially increased by decreasing full width at half maximum values of x-ray rocking curves. High-resolution transmission electron microscope and energy dispersive x-ray spectroscopy analyses revealed thinning or partial vanishment of GaN cap layer in a degraded sample. To account for these experimental findings, degradation mechanism assuming diffusion of interfacial materials along dislocations was discussed. An improved device operating at 50 V, where this degradation mode was suppressed, exhibited an MTTF exceeding 1×10⁶ h at a channel temperature of 150℃.

View full abstract

Reduction of thermal resistance of InP-based double heterojunction bipolar transistors (DHBTs) is an important issue for preventing degradation of their reliability at high-collector-current-density operation. In order to reduce the thermal resistance, we fabricated InP-based DHBTs on a highly-thermal-conductive SiC substrate with two kinds of wafer bonding technologies (bonding first and bonding last technologies). In the bonding last technology, a thinned InP substrate with DHBTs fabricated in advance was bonded to a SiC substrate. On the other hand, in the bonding first technology, the devices were fabricated after the DHBT epitaxial-layer structure was transferred onto a SiC substrate by wafer bonding. It was found that electrical characteristics of 0.25-µm-emitter DHBTs fabricated by the bonding last technology degraded when the InP substrate was thinned down to 20 µm. In contrast, DHBTs fabricated by the bonding first technology successfully demonstrated marked reduction of their thermal resistance by 28% without any degradation of their electrical characteristics.

View full abstract

Due to its low insertion loss and the high linearity, the GaAs Junction Pseudomorphic High Electron Mobility Transistor (JPHEMT) is widely used for RF switches in wireless communications. In this paper, we describe a JPHEMT with a p-type capping layer (p-cap JPHEMT) for improving the RF characteristics. The advantage of p-cap JPHEMT is shown by device simulation technology and the device parameters of the p-layer are optimized to enhance the off-state characteristics. It has been demonstrated that BV_gs is 11 V higher, C_off is 25% lower, the power handling capability is better and the 3^rd harmonic distortion is 6 dB lower than those of the conventional JPHEMT. The obtained R_on*C_off product of 100 fsec is very favorable compared with RF switches based upon conventional JPHEMT and SOI technologies.

View full abstract

We report the dependence of I-V characteristics on doping concentration in a GaAsSb/InGaAs Double-Gate Tunnel FET (GaAsSb/InGaAs DG TFET) by simulation. Increase of doping concentration at source region is effective to achieve high on-current. However, it leads degradation of off-current. To suppress off-current, low doping concentration at drain region is effective, although on-current is decreased by high series resistance in the drain when drain concentration is too low. As a result of optimization, we obtain I_ON of 466 µA/µm and I_OFF of 10 pA/µm at V_DD 0.5 V in a GaAsSb/InGaAs DG TFET.

View full abstract

GaN high voltage diodes with high-k dielectrics passivation underneath the filed plate are demonstrated. Simulation results at reverse voltage of 1000V showed that the maximum electric field near the mesa-etched pn junction edges covered with film of dielectric constant k value of 10 was reduced to 2.3MV/cm from 3.2MV/cm (SiO₂(k=3.9)). Mesa structures of pn junction diodes were fabricated by ICP dry etching, and mixed oxide of SiO₂ & CeO₂ dielectric film with k value of about 12.3 was deposited by CVD. I-V characteristics of the diode with a field plate showed the breakdown voltage above 2000 V with an increased avalanche current. This means that the electric field reduces at the periphery of the mesa etched pn junction and was uniformly formed across the whole pn junction. It is clear that high-k dielectric film passivation and filed plate termination are essential techniques for GaN power devices.

View full abstract

The new wide-bandgap oxide semiconductor, gallium oxide (Ga₂O₃), has gained attraction as a promising candidate for power device applications because of its excellent material properties and suitability for mass production. Fundamental halide vapor phase epitaxy (HVPE) technology for Ga₂O₃ thin film growth has been developed for development of vertical Ga₂O₃ devices. Unintentionally-doped Ga₂O₃ layers grown by HVPE were high purity and high crystalline quality, and an electron density in the Ga₂O₃ film can be precisely controlled by Si doping in the wide range of n=10¹⁵-10¹⁸ cm^-3. We fabricated vertical Ga₂O₃ Schottky barrier diodes (SBDs) with HVPE-grown n^--Ga₂O₃ drift layers on single-crystal β-Ga₂O₃ (001) substrates. The SBDs demonstrated promising device performance as next-generation power devices, such as the specific on-resistances of about 3.0 mΩ·cm², the ideality factors of almost equal to unity, and the off-state breakdown voltage of about -500 V.

View full abstract

The kHz band PLC (Power Line Communication) is one of the convincing mediums of the communication system for smart meter reading. Data transmission between MV and LV points to pass the transformer by PLC is interested. In this paper, the PLC signal propagation characteristics between points of MV at substation and LV at customer are calculated and analyzed.  The results show that direct communication between points on MV distribution line and most LV customers is impossible due to large propagation losses at MV-LV transformer and on LV distribution line.

View full abstract

This paper evaluates effect of the overheads caused by 6to4 tunneling on QoE for Web services by experiment. The authors consider Web usability as QoE and treat three actual Web services which support both of the IPv4 Internet and the IPv6 Internet. The results show that the effect of QoS degradation caused by 6to4 on Web usability depends on the Web service. According to Web service, QoE degradation caused by 6to4 does not always occur.

View full abstract

Although advanced age is generally accompanied by a decline in physical and mental attributes related to cognition, decision-making and mobility, few elderly people recognize how far their visual, cognitive and decision performance have declined since their youth. This lack of awareness is a major cause of accidents. Especially, elderly traffic accidents are related to several cognitive functions, such as divided attention, spatial recognition, planning and execution, etc. In this study, we have developed a method to evaluate cognitive functions of the elderly while driving, and have made a simple driving simulator. Elderly cognitive functions, which are divided into 5 levels (level-1: traffic accident, level-2, 3, 4: near miss cases, level-5: safe driving), are evaluated by driver performances in simulated driving. Some experiments by elderly drivers have largely showed the validity of this method and driving simulator.

View full abstract

The final goal of our research is to develop a measurement system for quantitative evaluation of the skillfulness of the lower extremities. There is a strong need for methods of quantitative evaluation of therapeutic effects to be available in rehabilitation clinics. The quantitative evaluation results should also have high reliability and high validity. Previously, we developed a measurement system to assess of the lower extremity function with a circular course, and a star-shaped course. The quantitative evaluation method for the lower extremities in both the circular and star-shaped courses has sufficient reliability. In this study, it was confirmed that the proposed quantitative evaluation method has sufficient validity in both the circular and star-shaped courses.

View full abstract

The purpose of this study is to develop a small hybrid sensor module that is integrated an active electrode and an optical device, and to confirm the basic performances of the sensor module and the capability to measure brain activities. The active electrode is composed of an electrode plate and a preamplifier. The optical device is composed of a laser diode, photo-diodes, and preamplifiers. In order to confirm the basic performances, we measured the frequency response and the noise density of the electrode, the frequency response of the optical detector, the stability of the light source, and the response to hemoglobin concentration change. All of the basic performances were enough to measure brain activities. In order to confirm the capability to measure brain activity, an electroencephalogram (EEG) measurement experiment and a cortical hemoglobin concentration measurement experiment were carried out. EEG signals of alpha wave band were significantly weak during light stimuli (p < 0.05). Cortical hemoglobin concentration signals were significantly strong during Stroop test (p < 0.05). In conclusion, we developed the small hybrid sensor module and confirmed the basic performance and the capability to measure brain activities.

View full abstract

The mathematical simulations are provided for the development of devices and/or methods for medical brain hypothermal treatment. The mathematical model reflecting thermal property of human head was presented as a group of particles consisting of small nodes. By using our mathematical model of head, a new appropriate brain cooling pattern is proposed for the temperature control in selective brain hypothermia, when its temperature rises owing to the increasing productive and accumulating heat by brain injuries. The clinical effectiveness of the new cooling pattern with less consumption of energy is given for the control of brain temperature by using our model of particles of nodes.

View full abstract

Multiple Frequency Continuous Wave (MFCW) is one of the efficient methods having both long range performance and range resolution. In this paper, the transmission sequences and signal processing to expand the observable maximum velocity and maximum range of MFCW are described. At first, we propose Hybrid Sequence MFCW in which frequency steps for the velocity estimation are added to the conventional MFCW to expand the observable maximum velocity. And then we also propose the signal processing using two initial frequencies to expand the observable maximum range. We show the initial experimental results on a general road. The results indicated that both proposed methods worked well on the field experiments.

View full abstract

Vibration sensor is used as a sensor to measure such as pressure, force, acceleration and density. In particular, tuning fork sensor is suitable as a sensor for electronic balance due to high sensitivity to the force and its long term stability. However, tuning fork sensor is affected by the temperature and the air density. Therefore, it is necessary to compensate the effects of them. In the conventional method, we could not compensate adequately for the abrupt change in the air density due to the delay of smoothing filter. In this paper, we propose accurate compensation method of the drift which is caused by changing the air density of tuning fork sensor. Applying Kalman filter, the drift of the tuning fork sensor can be very small even when abrupt change in the air density and there is noise the measured the atmospheric pressure and the frequency.

View full abstract

[in Japanese]

View full abstract

The VBW (Vogel-Breuß-Weickert) model is proposed as a method to recover 3-D shape under point light source illumination and perspective projection. However, the VBW model recovers relative, not absolute, shape. Here, shape modification is introduced to recover the exact shape. Modification is applied to the output of the VBW model. First, a local brightest point is used to estimate the reflectance parameter from two images obtained with movement of the endoscope camera in depth. After the reflectance parameter is estimated, a sphere image is generated and used for Radial Basis Function Neural Network (RBF-NN) learning. The NN implements the shape modification. NN input is the gradient parameters produced by the VBW model for the generated sphere. NN output is the true gradient parameters for the true values of the generated sphere. Depth can then be recovered using the modified gradient parameters. Performance of the proposed approach is confirmed via computer simulation and real experiment. Although it is also possible modify the shape by using regression analysis instead of neural network, it was confirmed that NN performs better accuracy than regression analysis.

View full abstract

Financial markets are connected well these days. One class assets' price performance is usually affected by movements of other classes of assets. However, the relationship between them is hard to trace and predict along with increase in complexity of markets' behaviors these days. Nothing like stock market, money or bond market is an over-the-counter market, where assets' prices are often presented in the form of classes of discrete quotations by trader's subjective judgments, thus are hard to model and analyze. Given concern to this, we define the Type 2 fuzzy random variable (T2 fuzzy random variable) to quantify those bid/offer behaviors in this paper. Moreover, we build a T2 fuzzy random support vector regression (T2-FSVR)scheme to study relationships between these markets, thus form an effective trading strategy to predict the trend of market prices. We use matlab platform to implement and test the effectiveness of the new model, then train and test it with 2014 whole years price data of bond and money markets. We also compare T2-FSVRs prediction accuracy with type-2 fuzzy expected regression(T2-FER) and confidence-interval-based fuzzy random regression model(CI-FRRM). The result shows that T2-FSVR outperforms and has 98% accuracy while CI-FRRM has 81% accuracy and T2-FER has only 70% accuracy. Moreover,T2-FSVR can be developed into a automated trading strategy for practical business use, which is able to learn behaviors of different markets based on mass of available historical and real time data and earn profit automatically.

View full abstract

Skyline cube (SC) is an extension of data cube, where the skyline operation is used to aggregate each cell of tuples with the same values of dimension attributes. To make SC more effective decision making tool, range query regarding SC is potentially promising. This paper discusses the storage structure of SC and range query processing system. R-tree based storage structure is dedicated to multi-dimensional index which associates values of dimension attributes with the corresponding results of extended skyline operation. The sequential processing system and the parallel processing system are built upon R-tree based storage structure to answer range queries. Experimental results show that R-tree based storage structure is allowable, because its storage size is at most 3 times as that of materialized view. Also, both of the sequential processing system and the parallel processing system are superior to the materialized view system in processing range queries. Especially, the parallel processing system takes below 20% times as that of the materizalized view system, even in the case of range query including anti-correlation skyline attributes which entail heavy burdens.

View full abstract

This paper proposes a method for detecting general obstacles on a road by subtracting a present in-vehicle camera image sequence from its corresponding past image sequences using an illumination insensitive feature. We conducted experiments using several image sequences captured under various illumination conditions, and confirmed the effectiveness of the proposed method.

View full abstract