The phenomenon referred to as stochastic resonance may improve the performance of signal detection devices. In digital signal processing, the effect of stochastic resonance depends on the sampling frequency. Thus, when applying stochastic resonance to digital signal processing, the sampling frequency to achieve high-performance signal detection must be determined. This paper elucidates the sampling frequency dependence of the stochastic resonance effect. It is found that the signal-to-noise ratio obtained by stochastic resonance becomes almost saturated as the sampling frequency is increased.

抄録全体を表示

ZCZ-CDMA, which is a quasi-synchronous CDMA (code division multiple access) system using a set of sequences with a zero correlation zone called ZCZ code as a spreading code, is useful for short-range wireless communications because of its excellent properties such as co-channel interference-free performance and fast frame synchronization capability. In this paper, we propose a ZCZ code with good aperiodic autocorrelation to be used as a synchronization symbol and training signal for channel estimation. The ZCZ-code is derived from an even-shift orthogonal sequence called an E-sequence, whose out-of-phase aperiodic autocorrelation function takes a value of zero at any even shift. A ZCZ code of length 2ⁿ is discussed expressed by both a logic function and matrices including the Sylvester type Hadamard matrix. The logic function gives a code generator consisting of 2ⁿ counters, and feedforward logic, and the matrices provide a matched filter bank that can take the correlation between input signals and any sequence in, the ZCZ code simultaneously.

抄録全体を表示

The present paper proposes a new method for constructing polyphase asymmetric zerocorrelation zone (A-ZCZ) sequence sets using discrete Fourier transform (DFT) matrices and orthogonal codes. The proposed method can generate A-ZCZ sequence sets that cannot be obtained using known methods. The newly obtained A-ZCZ sequence sets include optimal ZCZ sequence sets. In addition, two arbitrary sequences that belong to different sequence subsets become uncorrelated sequences. The proposed method is expected to be useful for reducing or avoiding inter-cell interference from adjacent cells in approximately synchronized code-division multiple-access (AS-CDMA) systems.

抄録全体を表示

In order to expand the conventional binary-type gene-coding method based on the quantum bit representation, we propose a novel gene-coding method that can represent an integer-type gene as multiple quantum bits, and clarify its characteristics from experimental analysis using three classes of the integer knapsack problem (IKP). In this paper, the proposed gene-coding method is called the integer-type gene-coding method with multiple quantum bits. From the experimental analysis of the IKP, we have confirmed that the proposed integer-type gene-coding method can discover a quasi-optimal solution in a shorter search time than the conventional binary-type gene-coding method in the quantum-inspired evolutionary algorithm (QEA) and the QEA with a pair-swap strategy (QEAPS) using genes represented by quantum bits. Actually, a quasi-optimal solution that is not a strict optimal solution is practical enough for many real social problems. Furthermore, we have confirmed that the gene-coding method suitable for discovering the strict optimal solution in the IKP depends on the algorithm used, namely, the integer type is suitable for the QEA and the binary type is suitable for the QEAPS.

抄録全体を表示

This study is based on the utilization of noise as the main approach to overcoming the problems associated with spike transmission in neuromorphic excitable media, such as receiving weak signals as inputs. The main objective of this study is to consider the dependence of noise on membrane potential. This dependence allows proper spike transmission, discriminating weak signals from parasitic fluctuations that could generate a fake response. Inspired by this idea, we investigate how noise enhances the performance of spike transmission in serially connected electrical circuits receiving subthreshold inputs. Electrical simulation results for a small number of excitable circuits connected with each other showed that with the assistance of noise, spike transmission is activated even in the presence of subthreshold signals. Moreover, dynamical noise proved to be an efficient technique for preventing unwanted spikes.

抄録全体を表示

The particle swarm optimization (PSO) algorithm has been used in various applications because of its powerful ability to tackle complex optimization problems. The optimization process of PSO has to satisfy constraints according to the requirements of actual nonlinear systems. An adaptive finite impulse response (FIR) filter design based on the PSO algorithm was proposed in our previous work. The adaptive FIR filter is a nonlinear constrained system. However, the performance of the adaptive filter is unsatisfactory. A high side lobe level and noise leakage is the main drawback of previous adaptive filters based on PSO. This paper proposes an improved co-evolutionary strategy based on a reinitialization method as a constraint technique for side lobe suppression. Two PSOs have been included in the co-evolutionary strategy to improve the constrained optimization. One is for self-adaption and the other is for side lobe suppression. The reinitialization method can overcome the decline of accuracy due to the difference between the two PSOs. The simulation results illustrate that the co-evolutionary strategy with reinitialization further enhances the accuracy in comparison with the original co-evolutionary strategy. With the improved co-evolutionary strategy, the adaptive FIR filter considerably strengthens side lobe suppression.

抄録全体を表示

This paper presents a novel hardware architecture of particle swarm optimization (PSO) for various kinds of applications based on field-programmable gate array (FPGA) technology. The conventional PSO hardware architecture cannot meet the requirements if the PSO algorithm and its numerical system need to change for different applications. The proposed flexible hardware architecture offers a resourceful approach to change the PSO algorithm and its numerical system without hardware modification. The proposed flexible hardware architecture employs two distinctive features. One is the combination of equations of various PSO algorithms and their integration into a new block called a generic particle calculation block (GPCB). The other feature is the use of two different fixed-point formats to meet the requirements of different numerical systems. Our experimental results prove that the proposed hardware architecture can support 10 different types of PSO algorithm with minimal chip cost. Moreover, it can drastically reduce the redevelopment time of PSO hardware implementation by dynamically changing the type of PSO algorithm without redesigning the hardware architecture.

抄録全体を表示

A hybrid MAC-based multipoint relay with energy awareness is proposed for a system data sharing protocol in a wireless sensor network (WSN), which is abbreviated to the HMAC-EA-MPR-SDS protocol. It adopts a system data sharing concept that combines broadcasting routing with data fusion so that data exchange between wireless sensor nodes can be easily realized in the real world. In order to reduce the redundant broadcasting in the conventional system data sharing protocols such as the on-demand dynamic multihop data sharing (On-Demand DMDS) protocol, we propose a multipoint relay with energy awareness (EA-MPR). In EA-MPR, a subset of nodes (called MPR nodes) is selected to forward the data to be shared throughout the network, and the power consumption of the MPR nodes can be efficiently balanced by energy awareness. Therefore, the network lifetime can be extended drastically. A hybrid MAC based on EA-MPR is introduced to improve the transmission delay and scalability of the time division multiple access (TDMA) method adopted in the conventional system data sharing protocols. It allocates a short-timeslot TDMA for each node to broadcast its data to be shared as well as provides a high-speed access method (token chain) for the MPR nodes to exchange data. The simulation results prove that HMAC-EA-MPR-SDS can reduce the redundant broadcasting and the corresponding power consumption in comparison with On-Demand DMDS. Furthermore, it can achieve a much shorter transmission delay and a higher network scalability than other conventional system data sharing protocols.

抄録全体を表示

Real systems are often composed of many elements interacting with each other and show complex behavior. To predict these complex systems, we can refer to their past behavior, but all of the observed elements do not always compose the same system. Thus, we must detect some essential elements from the observed elements so as to improve the prediction accuracy of learning data. Moreover, if we apply Takens's embedding theorem to reconstruct an attractor with only a single element, we need not select elements, but we must optimize embedding parameters. In any case, because we must solve the above optimization problems, we applied the genetic algorithm (GA) as an example of metaheuristic techniques. Moreover, real systems might be nonstationary and their own mechanism changes dynamically. Therefore, we reiterated the GA for each optimization with simple algorithms to embed and to predict time-series data to save numerical costs. Through some simulations, we confirmed that our dynamical optimization can improve the prediction accuracy of multivariate nonlinear systems, even financial markets, and can help us to examine whether the structure of a complex system dynamically changes or not.

抄録全体を表示

This paper presents Max-plus algebra-based Morphological wavelet Transform (MMT) watermarking, which is a novel mobile-dedicated information data embedding method. The performance of watermarking processing is considered a very important factor in embedded processors, which are applied to mobile devices. This is because, in the case of embedding information data in digital contents for mobile application, real-time processing speed is more important than data security and image quality. The proposed MMT watermarking method can be defined by nonlinear operation (maximum or minimum search) and standard sum. Thus, while the proposed watermarking method only uses integer representation for all variables with embedding operation, the conventional watermarking method has to calculate complex numbers, trigonometric functions, and floating-points for restricting watermarked-image distortion. As a result, the proposed method can reduce the number of clock cycles to drastically less than those of conventional transform-based watermarking. The processing time of the MMT watermarking is about 1/28, 1/2720, and 1/4364 shorter than those of Haar Wavelet Transform (HWT), Discrete Cosine Transform (DCT), and Fast Fourier Transform (FFT), respectively. Furthermore, the practical-use processing time of the MMT watermarking with the evaluation board for the massive-parallel Single Instruction Multiple Data (SIMD) matrix processor is found to be about 96% shorter than that of the HWT watermarking method on the BeagleBoard-xM, which is used the ARM Cortex-A8. To compare the watermarked and extract image distortion, the PSNR values of MMT, DCT and HWT watermarking are calculated. The PSNR value of the proposed MMT watermarking is suitable for practical use with 5-bit watermark embedded information, which can keep maintain image quality for both watermarked and extract images. Consequently, the proposed MMT watermarking algorithm can achieve an efficient real-time watermarking scheme for mobile device applications.

抄録全体を表示

This paper is concerned with a formal linearization based on Chebyshev interpolation for nonlinear dynamic and scalar-measurement systems with Gaussian white noise and its application to a filter design. A linearization function that consists of the Chebyshev polynomials up to the higher order is defined, and a given nonlinear dynamic system is transformed into an augmented linear one with respect to this linearization function by applying Chebyshev interpolation. Further more, an augmented measurement vector that consists of polynomials of measurement data is also defined and a measurement equation is transformed into an augmented linear one with respect to the linearization function in the same way. To these augmented linearized systems, a linear estimation theory is applied to design a new nonlinear filter. With this method, the formal linearization is easily incorporated into many practical systems by simple calculation using a computer, and a nonlinear filter with higher accuracy than those using conventional methods can be designed.

抄録全体を表示

This paper is concerned with single nonlinear feedback control for nonlinear systems with noisy measurement. A given nonlinear system is linearized piecewise to design the linear optimal controllers, which are then smoothly united into a single nonlinear feedback controller by an automatic choosing function. The state estimation is carried out with a modified nonlinear filter, which includes the 1st-and 2nd-order filters. This is called an augmented automatic choosing control of modified filter type (AACCMF). Simulation results show that the new controller improves the stability of electric power systems well.

抄録全体を表示

We describe an improvement of the design of the phoneme recognizer that is based on the articulatory feature (AF). Several strategies for designing the optimal parameter set in AF-based Hidden Markov Model (HMM) are investigated. They include subword units, number of HMM states, vowel group separation, tuned insertion penalty, and HMM topologies. The proposed AF-based phoneme recognition with 5-state HMMs, separated vowel, triphone subword, Bakis topology, and optimal insertion penalty provides the best accuracy among the experiments, i.e., 81.38% for the JNAS speech database. This result surpasses the accuracy of the standard MFCC-based phoneme recognition for triphone subword, 3-state HMMs, and 16 Gaussian mixtures.

抄録全体を表示

People with hearing loss are likely to have sensorineural hearing losses, which are characterized by the ability to hear loud sounds as clearly as people with normal hearing, but not weaker sounds. To remedy sensorineural hearing losses, many methods have been commercialized with confirmed effectiveness. However, these methods basically include linear amplification, and their effects of increasing the volume of soft sounds may not be sufficient. Therefore, we have developed the amplitude-bandwidth compression/expansion (ACE) method, which is based on the nonlinear characteristics of human hearing. The characteristics of sensorineural hearing losses have been examined. Thus far, we had set the ACE parameters on the basis of the results of Hallpike and/or Moore and had tried psychometric tests. As a result, the characteristics of ACE compression tended towards superabundance. Therefore, we set new ACE parameters in this study, on the basis of the optimal speech level (OSL), which is a well-known method for speech audiometry based on the hearing level of the pure tone average. This report shows how to calculate the ACE parameters for a hearing aid based on the OSL, as well as presents the results of the evaluation by a speech audiometry and sensitivity test (KANSEI test) targeting sensorineural hearing losses.

抄録全体を表示

In this paper, we present a modification method of indistinguishable colors on the basis of the Color2Gray algorithm for people with color vision deficiency (CVD). Compared with people with normal color vision, people with CVD have difficulty in distinguishing between certain combinations of colors. To address this problem, we modify saturation and lightness components of the original image to preserve visual detail when perceived by people with CVD. The proposed method can automatically construct a transformation to maintain details for people with CVD while preserving naturalness for normal viewers. In addition, high-speed processing of recoloring is achieved by clustering on the basis of color quantization using the median cut algorithm.

抄録全体を表示

This paper proposes a method of finding scale-invariant corner feature from full-view image based on discrete spherical model. A full-view image is first mapped to a discrete spherical image. Then, Harris corners are detected in the discrete spherical image. After determining the scale of the found corner points, the descriptor of corner feature is generated from the patch determined by the position and the scale. In comparison with detecting feature from the full-view image directly, the proposed method can find more stable features and achieves higher rate of feature matching for full-view image pairs independent of the feature position at image.

抄録全体を表示

This paper presents a pixel restoration for a still frame using statistical measurements with optical flow. The reconstruction of lost pixels is an important issue in image restoration. As statistical properties of optical flow are already measured for neighboring pixels of a lost pixel, the most selected flow is determined as the optical flow of the lost pixel. Then a pixel of the initial point for optical flow becomes the target pixel of the lost pixel. Our approach demonstrates that high-quality restoration can be realized for pixels by utilizing the measured flows statistically, since we redefine the optical flow that analyzes a moving image to be applicable to still frames. In order to show that our approach is effective in quality, experimental results are presented in comparison with those of the conventional technique.

抄録全体を表示

H.264/AVC has been successful in the past 10 years owing to its wide use in many fields from high-definition digital TV (HDTV) to low-resolution one-segment mobile TV. In recent years, with the high demand for high-resolution applications, 4K resolution (UHDTV) which has four times the resolution of 1080p HDTV, has become highly required for video streaming and broadcasting. However, the computational complexity for 4K-resolution applications is still the bottleneck preventing the realization of real-time decoders. In this work, a motion vector management architecture that is designed for total motion vector management and calculation is proposed. The proposed architecture can calculate predicted motion vectors, select reference motion vectors for the direct mode, and calculate the motion vectors for interlace decoding. The proposed architecture can perform 176 cycles/MB decoding in the worst case. The simulation result shows that when working at 200 MHz, the proposed architecture can be implemented byabout 99k gates.

抄録全体を表示

We present a nonnegatively constrained iterative method formulated by discretizing nonlinear differential equations in a continuous-time image reconstruction (CIR) system for computed tomography (CT). The method of using the discretization has a simple structure, and is based on the continuous approach for an ill-posed inverse problem; therefore, we expect that the method of using the discretization produces better-quality images quickly and easily against the conventional methods. We give proof of the convergence of a desired solution in the discretized CIR system, theoretically. The theory is illustrated through experiments with a simulated phantom and projection data acquired from an X-ray CT scanner.

抄録全体を表示

Auditory search experiments were conducted to investigate the detectability of target signals with regard to movements in temporal envelopes. The movements in the temporal envelopes of signals were defined as the mean value of slopes, i.e., the first-order approximation of the temporal envelopes. The movements were systematically controlled by band-pass filtering on the modulation spectrum of the signals. The movements of background signals in each target signal were set to be different from the movements between target and background signals. Stimuli were composed of different temporal envelopes with a noise carrier. 1/2-octave band-noise as a carrier was used in which the center frequency of the carrier was 1380 Hz. Participants were required to determine whether they heard the target signal with the background signals. The results obtained from auditory search experiments demonstrated that the detectability of each target signal could be improved when the movements of the target and background signals differed. The results revealed that the difference in the movements of the signals affected the detectability of the signals.

抄録全体を表示

Restoring an observed image suffering from blur and noise simultaneously is a challenging problem that may cause a large estimation error of blur and noise parameters. In this paper, a novel blind image deconvolution approach based on noise variance estimation is presented. This method first performs noise variance estimation from a noisy blurred image. Then, using the property that a certain type of blur may lead to a specific frequency component distortion of the image Fourier spectrum, the blur type can be reorganized. After this, according to the reorganized blur type, the image and blur model coefficients can be computed more efficiently by minimizing an objective function based on the autoregressive moving average (ARMA) model, where the maximum-likelihood (ML) method is used. The restored image is obtained with a least-squares filter. Experiments on images are presented, which show that the proposed method is capable of yielding good restoration results.

抄録全体を表示

English rhythm instruction materials (RIM) encourage users to learn English rhythms. However, when one repeats the same lesson many times, it is typical to become bored. What happens to brain waves in the event of boredom - do some brain waves change? To clarify this issue, we acquired electroencephalograms (EEG) of learners' brain waves as they repeatedly completed the same RIM lesson. We also asked them to raise their hands when they felt bored. Time-frequency analysis using fast Fourier transforms of learners' EEGs showed that the power of their α (8-12 Hz), β (8-30 Hz), and γ (30-50 Hz) waves first held constant, then decreased at many electrode regions over the course of the lesson. After the decreases in EEG power, the subjects raised their hands to indicate boredom. On the other hand, the power in the θ waves (4-8 Hz) did not change. These results suggest that the decrease in α, β, and γ power across wide regions of the skull can serve as signs of boredom among learners.

抄録全体を表示