Hosei University Repository
Not a member yet
22008 research outputs found
Sort by
RESEARCH ON NEW MECHANICAL STRUCTURES USING 3D STRUCTUAL OPTIMIZATION METHODS
Get PDFIn topology optimization based on structural mechanics, the SIMP method is often used to define the material density, and optimization is performed to maximize stiffness by imposing area and other constraints. 3D structural optimization methods may be used to derive structures with even greater degrees of freedom. In this paper, we first verify the accuracy of the finite element method and the validity of the sensitivity analysis using the associated variable method, using a cantilevered beam model. Topology optimization by mean compliance minimization is performed on the chair model to maximize the stiffness. The SIMP method is used as the characteristic function, the steepest descent method is used as the optimization method, and the outer point penalty function method is applied to account for area constraints
A STUDY ON DIGITALLY DIRECT DRIVEN SPEAKER SYSTEM USING ISI SHAPER
Get PDFThis paper proposes a new dynamic element matching (DEM) method for simultaneously shaping speaker mismatch and inter-symbol interference (ISI) error in digital direct-drive speaker systems. In a previous study, ISI error in a multi-bit Delta-Sigma DAC was analyzed, and a DEM that simultaneously shapes ISI error in addition to shaping the element mismatch was developed. The proposed method was modified to be applied to the driving method of a digital direct-drive speaker system and analyzed by simulation. As a result, it is confirmed that the proposed method simultaneously shapes the element mismatch and ISI error
APPLICATION OF ELEMENTARY CELLULAR AUTOMATA TO MUSIC GENERATION
Get PDFWe present a simple music generation algorithm based on periodic pattern from the Elementary Cellular Automata (ECAs). Our algorithm may be regarded as a simplified version of the WolframTones. The simplification is important in systematic application of ECAs to music generation and its evaluation. We propose a simple music generation algorithm. As preparation, we explain the scales and the rhythm used in it. We evaluate some typical musical examples and evaluate their characteristics
TAOYAKA-Ⅶ : A MULTI-LEGGED ROBOT CAPABLE OF CLIMBING COLUMNAR OBJECTS AND WALKING ON ROUGH TERRAIN
Get PDFIn recent years, robots have applied not only to known environments such as factories, but also unknown complex environments, such as rescue missions and agriculture. In our previous works, we have developed a six-legged robot that can climb various columnar objects without measuring their shape and size by imitating an octopus-like behavior. In addition, it could walk on a flat horizontal plane. However, its legs were not sufficiently stiff to enable rough terrain such as rubbles and steps. The goal of this research is to improve our previous robot to adapt it to various environments such as steps and rough terrain. Experiments were conducted, and as the results, we confirmed that the robot can climb columnar objects as well as walk on rough terrain and steps
NECESSITY OF USING THE CONVERSION LENGTH DETERMINED BY THE FDTD-CW ANALYSIS IN THE DESIGN OF A WAVEGUIDE-TYPE POLARIZATION CONVERTER
Get PDFA simple method for evaluating a waveguide-type polarization converter with an asymmetric cross-section has been studied with an emphasis on the conversion length. We introduce the conversion length L_c1 obtained with the FDTD-CW analysis instead of the conventional length L_c determined by the eigenmode analysis. We show the characteristics of L_c1/L_c as a function of wavelength and defect ratio. Two models are investigated, i.e., a completely buried type and a model loaded on a Si substrate. It is found that the introduction of L_c1 allows highly accurate evaluation of the wavelength response in terms of the extinction ratio and the insertion loss, confirming a good correlation with the FDTD results
Experimental Measurement of Current and Frequency Characteristics of Saltiness Enhancement in Electric Taste
Get PDFIn this study, the experiment was conducted by applying an electrical stimulus to saline solution. The experimental data were analyzed to investigate the effects of current and frequency in saltiness potentiation. We also investigated the characteristics of frequency and current during saltiness potentiation
RESEARCH ON WPT-ROBOT : BASIC STUDY OF SELECTION SCHEME OF USEFUL ANCHORS FOR HIGH ACCURACY INDOOR POSITIONING SYSTEM
Get PDFIndoor positioning, for which demand has been increasing in recent years, faces challenges for practical application due to many error factors such as multipath and noise. In this paper, we investigate a basic algorithm to select useful anchors and estimate their positions with the aim of improving the accuracy of the indoor positioning system in WPT-Robot. After showing the superiority of the anchor selection method over the all selection method by simulation, we conducted an actual experiment using the proposed method in a simulated office environment. As a result, stable positioning in 3D space was achieved, and the position estimation accuracy was demonstrated to be less than about 0.3 m
A STUDY ON DETECTION MTHOD OF SLOW HTTP DoS ATTACK USING ENTROPY
Get PDFSlow HTTP DoS Attack, a type of low-bandwidth DoS attack, is a threat to services because it requires less resources for the attacker and is harder to be detected than conventional DoS attacks. In this paper, we show the feasibility of an attack detection method based on the entropy of the data arrival interval to the server and its average value. From the results of the verification of the proposed method, it is shown that it is possible to separate the normal state and the attack state in the experimental environment by setting threshold values for the two types of measured parameters, We also show that by narrowing the upper limit of the arrival interval of the acquired data, it is possible to separate the normal state from the attack state even when the ratio of attacks is reduced
Improvement of the Spatial Resolution of a Multi-pinhole SPECT System With a Deep Learning Method
Get PDFA method to improve the spatial resolution of images obtained with a stationary multi-pinhole SPECT system was proposed in this paper. A multi-pinhole SPECT system has an advantage of being able to measure dynamic functions of organs. However, the image quality is very sensitive to a pinhole size. In this work, a deep-learning based method to improve the spatial resolution was proposed, in which the projection images were converted to those measured with an infinitesimal pinhole system. The reconstructed images with the proposed method had significantly higher spatial resolution and reproduced more detailed structures than those with the conventional methods. These results demonstrated the effectiveness of the proposed method
EVALUATION OF PERFORMANCE AND INTERPRETABILITY OF A MALWARE DETECTION METHOD FOCUSING ON A HIERARCHICAL STRUCTURE OF EXECUTABLE FILES
Get PDFMachine learning-based malware detection methods realized fast, accurate, and flexible detection. However, their internal calculation processs tend to be black-boxed so they are unreliable and not used for commercial anti-malware softwares. Although some studies proposed interpretable malware detection methods to tackle to this problem, there still is a problem that there is no quantitative evaluation for the interpretability. In this study, we propose a novel method to evaluate the validity of the interpretability of a model by observing whether another malware detection method diagnoses a malware as a goodware after removing important components in it. Also, We propose a new interpretable malware detection method that utilizes a hierarchical structure in executable files: there are many functions in a file, and each function consists of assembly functions. Our model performs assembly instruction-level analysis, function-level analysis to analyze file using attention mechanism. By analyzing attention layers, our model can show which functions and instructions are suspicious. in this paper, We evaluated its performance and the validity of interpretability by experiments using applications of Android OS and Windows OS. As a result, we confirmed the model has high detection performance on both platforms, and the interpretability is verified on Windows applications