This article has no abstract.

View full abstract

Semiconductor light emitting device with a high-power output and a broadband spectrum characteristic is valid as the light source for the optical sensing system. However, a high-power and a broadband spectrum characteristic are in the relation of the trade-off. We have successfully fabricated a broadband emitting diode by multiplying emission areas. This device has the performance of 1.4 times of spectrum width compared with the conventional identical output device. This result shows the device performance exceeds the conventional trade-off.

View full abstract

Deep-ultraviolet (DUV) light-emitting diodes (LEDs) are demanded for a wide variety of potential applications, such as sterilization, water and air purification, medical use, and so on. We have demonstrated 222-351nm AlGaN and quaternary InAlGaN-based DUV-LEDs by developing a low threading dislocation density (TDD) AlN crystal. We achieved an external quantum efficiency (EQE) of about 4% and an output power more than 30mW in DUV-LEDs for use in sterilization applications by developing new crystal growth techniques.

View full abstract

InGaN green laser diodes (LDs) on semipolar {2021} GaN substrates with output powers of over 100mW in the spectral region beyond 530nm is demonstrated. Wall plug efficiencies as high as 7.0-8.9% are realized in the wavelength range of 525-532nm, which exceed those reported for c-plane LDs. The lifetime at a case temperature of 55°C was estimated to be over 5000 hours for an optical output power of 50mW. These results suggest that InGaN green LDs on the {2021} plane are better suited as light sources for applications requiring wavelengths over 525nm.

View full abstract

Growing demands for reduction of energy consumption and environmental impact are rapidly leading to the spread of solid state LED lighting with low power consumption and high reliability. To meet these demands, we are developing high efficiency and high power LED on GaN substrate. High optical output power of 1.23W at 1A was obtained by using the technologies which are epitaxial growth of high quality m-plane InGaN on GaN substrate and low contact resistance electrode. The m-plane InGaN-LED realized the quite small external-efficiency droop of about 20% up to 500A/cm², while large droop seriously occurs in polar c-plane (0001) InGaN LED.

View full abstract

We have developed a frequency-tunable continuous-wave (CW) terahertz (THz) light source using non-collinear phase matched difference frequency generation (DFG) based on excitation of phonon-polariton mode in gallium phosphide (GaP). Distributed feedback laser diode (DFB) and external cavity laser diode (ECLD) are used as excitation near-infrared light sources. THz wavelength can be tuned by changing the wavelength of DFB LD. The wavelength of DFB LD can be tuned from 1058nm to 1061nm by changing the temperature of its diffraction grating structure embedded in DFB laser diodes. The line width of each laser is less than 4MHz. The output power of two lasers is amplified up to 2W by using a polarization maintained (PM). ytterbium-doped fiber amplifier. CW THz wave source has superior features in the frequency resolution because of its narrow line width. As an example of high resolution THz spectroscopy, we have shown our recent results on non-destructive inspection of ultra high molecular weight polyethylene (UHMWPE) deformation.

View full abstract

We developed the compact high brightness white light source based on the nitride semiconductor blue laser diodes (LDs) with yellow phosphor materials. The blue laser light from the LD is focused on yellow phosphor by optical lens. As a result, the white light is obtained by mixing the blue laser light and the yellow light of phosphors excited by blue laser light. The luminous flux and the luminance were 250lm and 270cd/mm² at the operating current of 1.2A. This high luminance is 2.5 times higher than conventional HID lump.

View full abstract

Wall plug efficiency (WPE) of 830nm single mode LD is dramatically improved with AlGaInP material. Conventional near infrared LDs based on AlGaAs have poor temperature characteristics due to small energy gap in conduction band between a p-cladding layer and an active layer, resulting poor WPE at high temperature. An AlGaInP based LD is a most effective candidate for the excellent characteristics because of the large gap. A high power 830nm LD is newly designed based on AlGaInP. The LD shows excellent WPE around 40% at 400mW, CW output, 60°C case temperature. They also show very stable operation at the condition up to 1,100 hours.

View full abstract

We are developing automotive light detection and ranging (LIDAR) sensing technologies based on both CMOS single-photon avalanche diodes (SPADs) and scanning systems and on optical fiber preamplifier. CMOS SPADs are highly sensitive detectors that can detect single photons and can form dense pixel arrays. LIDAR based on optical preamplifier has thumb-sized small sensor head.

View full abstract

A Ge-MIS structure has attracted attention for the candidate of a next generation CMOS device. To date, we have successfully developed Ge-MIS structures with superior interface qualities by ECR (Electron Cyclotron Resonance) plasma techniques. In addition, we have shown that DLTS (Deep Level Transient Spectroscopy) method is useful for evaluating the Ge-MIS structures. In this report, we have evaluated the near-interface traps in GeN_x/Ge structures by DLTS method, and have evaluated the effect of annealing on reducing the traps. The traps observed in the Ge-MIS samples were greatly reduced by 400°C annealing. The origins of the traps are also discussed by comparing with the traps in germanium reported to date.

View full abstract

Plasticity of neuronal networks is widely investigated as the basis of learning and stimulus-induced plasticity is one of the remarkable topics. For inducing plasticity and leading a network to a desired state, effective stimuli have to be applied to the network. However, the effect of stimuli on living neuronal networks is difficult to evaluate because of the unsteadiness of the network property. In this paper, we produce a simulated neuronal network and compare the effects of tetanic stimulus at various sites of networks in silico. Using simulation, the tetanic effects can be evaluated irrespective of network transition. We confirmed that the direction of the network plasticity is determined by the tetanized site in the network, and this can be observed only when the proper test-stimulus site is selected. We conclude that this simulation-based approach effectively evaluates the stimulus effect on neuronal networks.

View full abstract

Vagus Nerve Stimulation (VNS) is a surgical treatment for intractable epilepsy. Although VNS has treated more than 60,000 patients, the precise therapeutic mechanism of action have not been elucidated. Here we hypothesize that VNS modulates the local synchronization of cortical activities and thus inhibits a seizure. In order to verify this hypothesis, we mapped local field potentials (LFPs) with high-spatial resolution using a microelectrode array from the temporal cortices of adult rats and analyzed VNS-induced phase modulation at a local network level. VNS significantly increased phase locking value (PLV), an index of synchronization, specifically at the gamma band. The optimal stimulation pulse of VNS was 0.5 mA and delivered at 10 Hz. This PLV modulation was most effective when the inter-electrode distance was about 1.6 mm. Gamma-band PLVs within the auditory cortex increased more significantly than those between the auditory and non-auditory cortices, while alpha- and low-beta PLV exhibited more synchronization between the auditory and non-auditory cortices. These results demonstrate that VNS modulates cortical synchronization in a band specific manner, and have some implications for the mechanism of action of VNS.

View full abstract

The purpose of this study is to develop an control algorithm for lighting systems which can generate uniform and satisfying illumination environment for office buildings and save considerable energy as well. In our previous research, we developed an energy saving algorithm which can successfully optimize the lighting systems in small sized offices. However when it came to the deep-plan office cases with large amount of luminaires, the algorithm suffered from the weakness of low speed and unstable output. In this paper, we introduce an improved control strategy for lighting systems with large amount of luminaires. The strategy defines a new characteristic value “Importance” for each luminaire and divides the luminaires into appropriate groups according to this value. After the grouping, an improved PSO algorithm is executed to optimize the dimming ratio for luminaires according to the target illuminance in occupied areas. The high speed and effectiveness of the proposed control method for lighting systems is verified by simulations in a lighting system simulator, and the superiority of this improved PSO based control algorithm is indicated in comparison with both traditional control method and prototype PSO control algorithm in our previous work.

View full abstract

For embedded systems, it is important to capture changes in real world and process properly. The changes are represented by external interrupts, and proper processes are implemented by nested interrupt service routines (hereafter ISRs). Here a necessity of mutual exclusion arises. In a software execution environment without any Real Time OS (hereafter RTOS), a traditional mutual exclusion approach was to disable/enable interrupts in a CPU specific manner. However this method typically degrades the real time performance because it defers execution of mutual-exclusion-free part of the system. Considering this situation, we have been studying a Real-Time Embedded Monitor (REMON) which provides a novel mutual exclusion method that can maintain real-time performance without RTOS. For in-service embedded systems, one major runtime fault is ISR stack overflow. It is extremely difficult to test every conditions where ISRs are called from various external conditions. Note that ISR stack holds not only data but also program instruction addresses, consequently a stack overflow may cause a fatal system error. In summary, ISR stack overflow is a significant issue, but it was not addressed by REMON until now. This paper proposes two safety extension methods for embedded systems using REMON. The first method detects ISR overflow and safely stops the system before triggering a system-down or a misbehavior. The second method reallocates ISR stack and resumes the system execution automatically.

View full abstract

In the Machine-to-Machine (M2M) system, which connects devices to the Internet and provides various remote services, there are many applications that need to carry the stock-type data analysis and the flow-type data analysis at the same time. However, because the stock-type data analysis and the flow-type data analysis are provided as independent systems, it is difficult to combine them. In this paper, we propose a method to process stock-type data analysis and flow-data analysis simultaneously by linking the M2M gateways to the M2M server. In this study, we implemented a prototype system using open source software, such as SQLite and R, to evaluate performance and functions of the proposed method.

View full abstract

Many researches have been done as for visual inspection, however, the fundamental viewpoint problem exists that a camera position is low and fixed. As a result, the field of view is greatly restricted by several objects in the environment to be monitored. We have been doing research on the motion analysis to handle aerial images for the purpose of wide area monitoring using the proposed Simplified Aerial Image Acquisition System which consists of four vinyl balloons and a small wireless camera. At first, lens distortion is corrected. Next, SURF feature points are detected and tracked in continuous image sequence. Then, these local features are classified into “rigid static object in environment” and “new appeared moving object” by analyzing locus of each feature points. Finally, the movement of the viewpoint is estimated by calculating the Affine transformation matrix. Experimental results in both indoor and outdoor environments have shown the effectiveness of the proposed method.

View full abstract

In image analysis, it is important to detect lines (straight lines) which represent edges of an object in the image. The most popular technique to detect lines is Hough Transform. In the transform, resulting peaks in the accumulator array which are gotten by a voting procedure in the parameter space represent strong evidence that a corresponding line exists in the image. In the voting procedure, a large number of votes which are far from the peaks are unnecessary and it makes the transform slow. In this paper, we propose a speed-up method by introducing a procedure in image space which removes lines except for the limited directional lines. Owing to the procedure, we can reduce voting range and remove the unnecessary vote. From the experimental results, we confirm that the proposed method has good performance.

View full abstract

In this paper, we propose a robust voice activity detection (VAD) method that uses a density ratio model. For VAD under highly noisy environments, the likelihood ratio test (LRT) is effective. Conventional LRT constructs speech and noise models, calculates the likelihood of each model, and takes the ratio of those likelihoods to detect speech. Although some improved LRT have been proposed, in conventional LRT, it has not been taken into account that the likelihood ratio of speech and noise model is required, not the likelihood of each model. The proposed method directly estimates the likelihood ratio without calculating each likelihood using an density ratio model obtained in advance by density ratio estimation procedure. Moreover, there is the problem of determining thresholds, which are used for VAD and significantly affect its performance. We propose a method that automatically determines thresholds using discriminant analysis. The experiments show that the proposed method is more effective than conventional methods especially under non-stationary noisy environments.

View full abstract

In this paper, we propose a speech intelligibility estimation method using the Support Vector Regression (SVR) with the normalized segmental Signal-to-Noise Ratio in 25 critical bands (cbSNRseg). In the proposed method, estimation was done in the target 32 noise environments which were classified into 3 clusters by the ambient noise clustering method with MIR (Music Information Retrieval) features and the x-means algorithm. Next, We compared cbSNRseg and 1/3 octave bands SNRseg (obSNRseg) and used the cross-validation RMSE in 5 regression methods including SVR. As a result, the weighted sum of RMSE using cbSNRseg was better than obSNRseg with RMSE reduction factor of about 0.8 compared to all other regression methods. Finally, we compared the performance of each regression methods in open tests. As a result, the best regression method was the SVR using the RBF kernel, in which RMSE is reduced by a factor of about 0.7 compared to other regression methods.

View full abstract

We propose an interface for a name input based on speech recognition using a word dictionary and syllable-based N-gram. User first utters a name and then chooses the correct word/syllables by touch input from word/syllable candidates which were obtained from speech recognition. Name utterance is hard to recognize accurately because of the large vocabulary size, so the system uses isolated word recognition with a dictionary containing high frequent words and continuous syllable recognition with syllable-based N-gram. System displays word candidates, syllable sequence candidates and a syllable lattice on the touch panel. Users who were familiar to speech input could find the correct answer from word candidates or syllable sequence candidates at a rate of 87-92%, and they felt that it was useful. Novice users also obtain the same level of accurate candidates after 30-40 minutes' practice and the interface became useful to them.

View full abstract

The menu planning problem deals with determining the dishes on a menu that satisfy some nutritional constraints, preferences of an individual, cost of food, etc. while optimizing the given objective functions. Many researchers have studied “one-period” menu planning problems with linear programming. However, our focus is on a “multi-period” menu planning problem as a combinatorial optimization problem. In this paper, we propose the model of nutrition education as the objective function for this problem. Our model of nutrition education employs entropy based on the menu diversity and order of appearance for the multi-period menu. We apply an estimation of distribution algorithm (EDA) to optimize this problem and then show that our algorithm has better ability to find good solutions.

View full abstract

We study a method to design wireless sensor networks (WSNs) for condition monitoring of railway structures so as to minimize the total cost of the investment and operation of WSNs. For the continuing long-term operation of WSNs, batteries in the network must be replaced before their depletion. However, large maintenance expense might be required if maintenance workers go to the site and replace batteries whenever a battery depletion occurs. On the other hand, a periodic inspection for railway structures shall be usually conducted once during a period of several years. Therefore it is practical for WSNs used for a condition monitoring of railway structures to replace batteries at the timing of a periodic inspection. This can make the continuing long-term operation of WSNs more effective. In this paper, we propose a mathematical model, that is to simultaneously determine the number of relays and gateways and their deployment locations, the transmission power level for each sensor and relay, and the routes for transferring sensor data to a gateway so as to minimize the total cost of the investment and operation of WSNs. Then we verify its effectiveness through computational experiments using real data acquired at the London underground.

View full abstract

There's one reason to utilize cache that mitigates processor performance limitation comes from data transfer bandwidth. Recently, cache size expansion is required in this use because data transfer bandwidth requirement is increasing for recent large data size and multi-core trends. However, cache size expansion is unwelcome because it causes problems come from circuit area and power consumption viewpoint. This paper focuses a data redundancy with the goal of reducing cache size and proposes a mechanism that does not store redundant data into cache. The proposed mechanism divides data into Higher Bit and Lower Bit, that stored into Higher Cache and Lower Cache, respectively. We reduced Higher Cache size to half size by keeping 46% redundant data in Higher Bit area not to store into Higher Cache. The evaluation results show that the proposed mechanism increases IPC by 3.3% on average compared with same circuit area conventional cache under SPECCPU2000 benchmarks.

View full abstract

To lower the burden on the global environment, renewable energy resources are increasingly being integrated to power systems, in particular, a great deal of photovoltaics (PVs) in Japan are expected to be installed on the roof-top of residential houses. However, this will make the supply-demand balance control and the voltage control of a power system difficult. A measure to reduce this impact is consuming excess power in local area where PVs are located.
We have developed a control algorithm for adjusting the operating pattern of multiple heat pump water heaters as a group to maximize the consumption of excess power of PVs in local area. The problem was formulated as a combinatorial optimization problem, and the tabu search method was applied as a meta-heuristics solution. The result of a case study for group control of six consumers that used the same low-voltage transformer showed an operating pattern that has the following features. First, the consumption of excess PV power was maximized. Second, the consumption of commercial electric power was minimized. We also compared the result by controlling a group of six consumers with the result by controlling of six consumers that had been controlled individually, and showed that the consumption of PV excess power was increased by 46% at the same level of commercial electric power consumption.

View full abstract

As cases of accidental water contamination caused by factory disposal and other influences are reported often, early detection of contamination is becoming an important issue in terms of safe water supplies and environmental protection. In recent years, bioassay systems have attracted attention because they enable water quality assessment based on the biological responses of fish, thereby removing the need for chemical examination. In this regard, several bioassay systems premised on the monitoring of swimming behavior or bioelectrical signals as indicators of contamination have been proposed. However, conventional systems can overlook contamination that affects non-monitored indicators. This paper proposes a bioassay system allowing simultaneous monitoring of swimming behavior and bioelectric signals from small fish. The results of experiments on 10 test fishes confirmed that the system can detect contamination within 10 minutes of 0.5% ethanol being added. These outcomes validated the applicability of the proposed system for the detection of water contamination.

View full abstract

As the first step in the development of new image quality assessment methods by using Facial Electromyogram, we examined whether the Facial Electromyogram changes with coded image distortion. First of all, by implementing the subjective assessment test, we examined the relationship between Facial Electromyogram and the conventional questionnaire method. Then, we analyzed the activity of facial muscles from the experimental results and predicted a subjective assessment score by using the stepwise regression analysis. At this time, we also analysed the influence of eye blink for Facial Electromyogram.

View full abstract