近年，多くの分野でビッグデータやIoT（Internet ofThings）といった言葉を耳にする機会が多くなってきた。それに伴い，様々な分野でこれまでにない詳細でかつ膨大なデータを利用できるようになった。ビッグデータの厳密な定義は知られていないが，一般に①大量で（volume）

State-space representation is well known as a powerful tool for description of linear systems. This scheme can describe internal states of systems as well as the input-output relationship, leading to versatile applications in the fields of signal processing and control systems. This paper reviews how to use the state-space representation, from the respective viewpoints of signal processing systems and control systems.

In technical committee on control engineering, a committee to strengthen the relations between control and signal processing started from 2015. In this paper, we introduce an application example of ”Kalman filter” in the field of signal processing from recent research in this committee.

Independent component analysis plays a central role in blind source separation, leading to many applications of signal processing such as telecommunications, speech processing, and biomedical signal processing. Although the independent component analysis requires cost functions for evaluation of mutual independence of observed signals, little has been reported on theoretical investigation of the characteristics of such cost functions. In this paper, we mathematically analyze the cost function based on Kullback-Leibler divergence in independent component analysis. Our analysis proves that the cost function becomes unimodal when the number of source signals is two and both of the source signals have uniform distributions. In order to derive this result, we make use of whitening of observed signals and we describe the cost function in closed-form.

This paper introduces a case example of signal processing education for researchers and engineers of signal processing. The example to be discussed here is the education for senior undergraduate student in Kawamata-Abe Laboratory, Department of Electrical, Information and Physics Engineering, School of Engineering, Tohoku University. This education provides not only exercises on software/hardware implementation of signal processing systems, but also opportunities to learn and master fundamental theory of signal processing and architectures of signal processors. This paper shows the details of such signal processing application, with discussions on comparison of our education with other lectures on signal processing.

In big data signal processing system, low-delay and low-operation driven digital filters are required for large amounts of data processing. We introduce a design method for low-delay FIR (finite impulse response) filters with semi-sparse coefficients. The semi-sparse coefficients stand for to have some 0 ± values with real values. The semi-sparse coefficients leads to reduction of number of multipliers. We show the design problem of the filters are formulated in a constraint optimization problem. Also, we propose a design algorithm to solve the design problem. Using the filters, the number of multipliers can be reduced. Finally, we present examples to demonstrate the effectiveness of the proposed method.

Afternoon sleepiness in daily life reduces arousal level, performance, concentration and so on. It is well known that short naps are effective to cancel the sleepiness and sleep stage 2 is one of the important factors about sleeping especially in a short time nap. Therefore, it is desired to the detailed analysis of sleep EEG to clarify the relationship between length and quality of nap. Some methods are proposed to detect sleep spindles automatically which based on the Fourie- or Wavelet-transform. However, they couldn't analyze the ream of waves because of frame processing. In this paper, we propose a sleep spindle detection method for the automatic EEG analysis method based on the method of visual scoring. In the proposed method, sleep spindle waves are detected using the enhanced terms of detecting sleep spindles in Fujimori's method and cumulative sum charts. Through example, the proposed method can detect the sleep spindles with 97.32 % sensitivity compared with a medical specialist.

Low pass filter is effective for noise reduction in control systems. This paper treats an I-PD control system with low pass filter. First, we show a closed-loop transfer function from reference value to the output of the system. The I-PD controller is designed based on the pole placement method. Second, we introduce a several conditions of the controller to be satisfied. Finally, we present that the proposed methods have obtained a good result by a simulation example and an experimental result of speed control for a DC motor.

Robust control is often applied to systems with uncertainties and disturbances. Above all, the H_∞ loop shaping method is known to achieve good control performance and robustness. In this method, the final controller consists of weighting functions and a stabilizing controller. The stabilizing controller is derived for the shaped plant to suppress the H_∞ norm of the transfer matrix consisting of a sensitivity function, a complementary sensitivity function, etc. In addition, the stabilizing controller improves robust stability margin while keeping gain characteristic of the shaped plant if weighting functions are suitable. As a result, the closed-loop system is well-balanced between good tracking and robustness. However, a final controller tends to be high-order. For this problem, reduction techniques are often applied to the final controller. In this case, performance and stability is not always adequately evaluated due to errors by the controller reduction.

This paper proposes a fully-parameterized fixed-order controller design method using frequency responses of the plant. We formulate a design problem for Multi-Input-Multi-Output systems as an optimization problem. Therefore, we can directly design a low-order controller from frequency responses using the iterative LMI optimization. Accordingly, we can avoid to deteriorate the evaluation of performance and stability.

We introduced a student experiment method for education of system control. The objective of our work was to make students understand a practical operation in control system engineering. First, students learn control theory such as first order system and PID control with Simulink on MATLAB. Students consider an experimental device of temperature control. Here, students do system modeling of the experimental device and design PID controller based on the model. Finally, we show the results of the questionnaire and some consideration from the experiment report written by the students. Analyzing the results above, we review the student evaluation and consider the effectiveness of the proposed education method.

Since the big data signal has large amount of data, low-delay signal processing algorithm is required. Digital differentiator is one of the digital signal processing tools, and useful to estimate a velocity from the position or obtain the image outline. In this paper, we introduce finite impulse response (FIR) and infinite impulse response (IIR) digital differentiators. The differentiators can achieve a low-delay and high-functioning processing such as low-pass differentiation.

In order to provide attractive content to users, it is important to design and manage services based on users' Quality of Experience (QoE). In recent years, research in this field has been increasingly focused on the biological information approach. The aim of our study is to develop a Near-Infrared Spectroscopy (NIRS)-based QoE evaluation system. In this paper, using event-related NIRS, we examine the change in the oxygenated hemoglobin (Oxy-Hb) as an indicator of regional cerebral blood flow (CBF) changes, which directly reflect brain activity. First, we measured subjects' Oxy-Hb while they were observing videos at different qualities, then performed a subjective assessment test through questionnaires, in order to identify the preference level for each video. As a result, we found significant differences between preferences and quality variations in the right side frontal area. We also recognized that video preferences had impacted more on CBF than video quality degradation. These findings suggest the possibility of building a NIRS-based QoE assessment model which realize the application of BMI to communication control.

Classifying leukocyte and examining their proportions is a very important for disease examination and diagnosis. This examination needs the knowledge of experts and a lot of time. Therefore, many automatic leukocyte images classification algorithms have been proposed. There is a method to classify 13 types blood cells using 1 vs 1 Support Vector Machine(SVM) in one of them. In the conventional method, leukocyte images are classified with the 26-dimensional feature vectors. However, the classification accuracy, is poor with these feature vectors in granulocytes in this method. In this paper, we propose new feature vectors to improve the classification accuracy of blast cells with low classification accuracy among the leukocyte fractions. That is, we add two feature vectors in the proposed method. And we improve the accuracy of the whole by using a random forest for the classifier.

In recent years, a number of algorithms for controlling parameters of an acoustic signal have been realized according as hardware technologies have been developed. A phase vocoder is one of the widely known methods for time scale modification (TSM) and pitch scale modification (PSM). The phase vocoder converts the length and the pitch of the acoustic signal by increasing or decreasing an instantaneous angular frequency. However, due to the phase unwrapping problem that occurs when the instantaneous angular frequency is estimated, the phase vocoder distorts the waveform of the modified signal. In order to alleviate this problem, we present a new algorithm for the phase vocoder. Our method does not encounter the phase unwrapping problem since we directly use the difference of phases between adjacent analysis frames to avoid estimating the instantaneous angular frequency. Moreover, our approach is effective even when the characteristics of the acoustic signal, e.g., a frequency and amplitude, change rapidly.

The image quality improvement techniques based on Retinex theorem are effective for enhancement of visibility in dark region, while visibility in bright region is degraded. This paper presents an image fusion technique using source image and Retinex processed image in order to implement high visibility in both bright and dark regions. In the proposed method, the statistic that corresponds to the subject's evaluation is used for image fusion. It is shown through some experiments that the proposed method can generate the image with high visibility in both bright and dark regions, with less computational cost compared with a conventional image fusion method.

We analyze the adaptation rate of active noise control using a statistical-mechanical method. Three measures are employed to evaluate the adaptation rate. The first measure is the MSE initial decreasing rate. The second measure is an adaptation constant, which is defined as the negative of the maximum eigenvalue of the coefficient matrix of differential equations that describe the dynamical behaviors of the macroscopic variables. The third measure is the integral of the MSE, which is defined as the integral of the difference between the MSE and the steady-state MSE. The first and second measures focus on only the initial and final stages of the MSE, respectively. In contrast, the third measure considers all stages in a well-balanced manner. Therefore, employing the integral of the MSE, we can determine the step size that optimizes the entire learning curve.

Conventional metal patterns in integrated circuits are known to exhibit drawbacks such as signal delays and heat generation. The use of the silicon photonics technology has been proposed to overcome these drawbacks. Silicon-based materials and processing technologies offer cost-effective solutions for large-scale fabrication in microelectronics, such as for the fabrication of silicon integrated circuits. However, the fabrication of light-emitting diodes (LEDs) using silicon-based materials for the silicon photonics has not been realized so far. We investigated the structural optimization of silicon to obtain a high efficiency of light emission. And we observed the correlation between the high efficiency light emission and alkali impurities such as Na and K.

We proposed a novel organic light-emitting transistors (OLETs) named band conduction based vertical organic light emitting transistor (BCBVOLET). In BCBVOLET, an organic light-emitting diode (OLED) is formed on the vertical organic transistor. The vertical organic transistor we used are originally developed transistor called band conduction based organic transistor (BCBOT). BCBOT allows low-voltage, and high-current operation. As a result of using BCBVOLET, we succeeded to obtain maximum luminance of 600 cd/m² at gate voltage of -2 V. ON / OFF ratio of current density is 10³ at gate voltage of -2 V to 0 V.

In the mobile communication equipment, the clock generator for driving each system is required to have a fast pull-in time, multiple signal of constant pulse interval, synchronization range, low output jitter, and wide lock-in range characteristics.

In this paper, multiple frequency MC-DCPLL is proposed. In this loop, the pulse width error of the multiplied output signal is a time within one phase difference of the multi-phase clock regardless of the multiplication ratio. The output jitter in the steady state is always within one phase difference of the multi-phase clock. Since it is a control method by dividing ratio changeable type, the lock-in range is extremely wide. Also, the initial pull-in time is always completed in one cycle of the input signal without being influenced by the multiplication ratio. It is clarified by theory and simulation that these characteristics can be obtained.

From the above, the versatility of the proposed multiple frequency MC-DCPLL is extremely high, and it can be expected to be used for clock sources etc. in various mobile communication equipment systems.

This paper proposes a novel control method to improve the responsiveness and the robustness of the boost DC-DC converter. The proposed control method is a feedback control incorporating a minor control that works to balance the input and output power. Experiment was carried out to verify the responsiveness and the robustness of the proposed control in comparison with the conventional voltage-mode and current-mode control. As a result, the proposed method revealed both of the fast response and the robust output voltage regulation against the load change.

In this paper, we modeled an aggregation communication server of the Fast Automated Demand Response (FastADR) system using an M/M/1 queuing system. Assuming that a very large number of FastADR participating BEMS gateways are widely-distributed, scalability of aggregation communication performance was estimated by the M/M/1 transient analysis. Validity of our analysis model was confirmed by realistic wide-area communication emulation system in the laboratory. Our scalability analysis has shown that at least parallel 25 aggregation systems will be necessary to realize expected communication time of approximately 30 seconds for an example of FastADR aggregation of 100 MW reduced demand.

Nowadays, touch interface has become popular. Touch interface features an easy to use intuitive user interface; however, having a few input elements is one of operability issues. In this study, our method detects an operated finger using information on the touch screen and values from various sensors when the user sits. This study enables touch interface to increase the number of input elements. We conducted experiments to collect operation data for each finger by 19 participants. As a result, we confirmed that our method, which only used other users' operation data to machine learning, classified four classes (left hand's thumb, other fingers, right hand's thumb and other fingers) with about 60% f-measure. Moreover, our method, only used own operation data to machine learning, classified four classes with about 80% to 90% f-measure. In this paper, we describe consideration about the results, and some characteristics.

Timbre is one of the most important factors for evaluating a perceived sound quality. In playing instruments, the timbre strongly depends on the instrumental shape. On human whistle, the instrumental shape is easily changed unlike hard material instruments which are violin, piano, flute, and so on. Hence, a technique adjusting timbre is very important for the human whistle. In this paper, we develop a timbre evaluation system for human whistle. We firstly investigate spectral shapes of the human whistle sounds obtained from professional whistlers and beginners. Then, we derive a simple model of the human whistle sound as a mixture of the fundamental tone, the second harmonic tone, and a broadband breath sound. Based on the beauty-factor of the timbre, we develop the proposed system which individually evaluates the second harmonic tone and the breath sound of the observed human whistle, and visually gives the evaluation result to the whistler in a real time.

We have proposed an artificial fiber pattern as a technique for embedding electrical information into the real world. In recent studies, we tried to extract embedded information in an artificial fiber pattern using a high-resolution digital camera instead of a scanner. In this paper, we attempted to use a relatively inexpensive and low-resolution web camera as an input device because one of the important applications of our technique is an augmented reality system. In addition, from a design perspective, we examined different colors (cyan, magenta, and yellow) instead of the grayscale, and a specific pattern (a traditional Japanese katagami pattern for a kimono (i.e., Ise-katagami)) instead of a paper fiber pattern.

Many new college or university students use the housing conditional search system for a leasehold house on Web recently. There is a recommendation system for a leasehold house for students with conditional search and decision rule in rough set. This system performs conditional search using housing information, and evaluate desirable houses by two kinds of decision rule in rough set. However, the users cannot sufficiently express their preference, and have a load of judgement about many specimen houses which is prepared in advance, in order to grasps the target user preference.

This paper presents the recommendation system in rough set for a leasehold house for students based on three step of target user's evaluation. This system grasps the target user preference based on the rules which correspond to three steps of judgement “Very Good”, “Good” and “Bad”. These judgments for random samples about houses can make the target user’s load small. Through the experiment, it is confirm that the system has the same degree of satisfaction as the usual system, and the load of the user in the system is less than the one in usual system.