Synchronization is the phenomenon of two or more autonomous oscillators oscillating with the same frequency due to mutual interactions or common external forces. In this study, we focused on crystal oscillators, which are utilized in a wide variety of electrical circuits. When independently oscillating crystals in each circuit are synchronized in-phase, it is conceivable that simultaneous information processing is possible. To observe synchronization, we designed a circuit containing two Pierce oscillators with a branching path for mutual interactions. Then, we show that in-phase oscillation of the Pierce circuits can be induced by controlling the coupling strength.

抄録全体を表示

Recently, machine learning has been attracting attention. Machine learning is mainly realized by the learning of artificial neural networks. Various learning methods have been proposed; however, the learning methods are based on gradient methods. On the other hand, swarm intelligence (SI) algorithms have been attracting attention in the optimization field. Generally speaking, SI algorithms have a large computation cost. Therefore, there are few cases of SI algorithms being applied to machine learning. In this paper, we propose a novel learning algorithm for an artificial neural network which applies our proposed nonlinear map optimization (NMO) method. NMO consists of some simple particles which are driven by a simple nonlinear map. NMO can be classified as an SI algorithm. However, it has only a small computation cost. Therefore, NMO can be applied to a learning algorithm for an artificial neural network. In this paper, we introduce NMO, and a small learning simulation is carried out to confirm the performance of our learning method.

抄録全体を表示

To find the shortest tour of a transporting vehicle in a bike sharing system (BSS), a bike sharing system routing problem (BSSRP) has been proposed. In the BSSRP, a single vehicle restores the number of bicycles in stations. However, in a real system, the restoration of the number of the bicycles in each station is carried out by multiple vehicles. To decide the shortest tour of the multiple vehicles, we proposed a mathematical optimization model called the multiple-vehicles bike sharing system routing problem (mBSSRP). We present a construction method and local search methods to solve the large-size mBSSRP. The result of numerical experiments shows that the proposed method can find good solutions in a short time.

抄録全体を表示

Typically, the receiver of wireless power transfer (WPT) systems uses batteries for their energy storage. The power transfer efficiency depends on the load impedance of the battery, which varies according to the charging condition. The efficiency depends on the impedances of both the transmitter and receiver. However, for practical WPT systems, total load optimization is to be executed without directly monitoring the receiver load. This paper proposes a practical and effective method for an adaptive efficiency control of the transmitter with indirect monitoring of the receiver load.

抄録全体を表示

In this work, we focus on digital color shift keying (DCSK), which is one of the modulation schemes of visible light communication (VLC). DCSK is an extended version of IEEE 802.15.7 color shift keying (CSK) that transmits data through the intensity ratios of red, green, and blue. Digitally controlled LED drivers of DCSK can also reduce the nonlinearity effect caused by the shift of the intensity amplitude. DCSK supports lighting functions such as flicker mitigation, target color control, and dimming control. For the dimming control method of DCSK, two schemes have been considered. One is pulse width modulation (PWM), which changes the duty cycle of optical transmit signals, and the other is digital pulse amplitude modulation (DPAM), which changes the number of red, green, and blue LEDs (RGB-LEDs) being used. In a previous analysis, a hybrid scheme of PWM and DPAM was proposed that could achieve both a wider dimming range than DPAM and a higher spectral efficiency than PWM. Unlike in theoretical analysis, with actual LED drivers, the dimming level (DL) obtained by PWM, DPAM, and hybrid PWM/DPAM may cause errors due to individual differences in the dimming method and LED drive current. In this work, we experimentally evaluate the accuracy of DL and the symbol error rate (SER) of DCSK with PWM, DPAM, and hybrid PWM/DPAM.

抄録全体を表示

In this paper, we propose a secure large-scale multiuser multiple-input multiple-output (MU-MIMO) system based on a common-key encryption scheme in uplink. This is a newly proposed large-scale system for a chaos MIMO (C-MIMO) transmission scheme with channel coding and physical layer security effects. However, the calculation complexity for demodulation exponentially increases with the number of transmit antennas in a straightforward conventional large-scale C-MIMO system. Therefore, in this paper, we propose the application of Gaussian belief propagation (GaBP), which is an iterative signal detection method with low calculation complexity, to the C-MIMO scheme. Consequently, we can construct a large-scale MIMO system which has a physical layer security effect with a low calculation complexity, and it is shown that a large channel coding gain can be obtained compared with unencrypted transmission with the same transmission efficiency.

抄録全体を表示

The low communication quality at a cell boundary is a long-standing problem in mobile communication systems. This is because the reception power level decreases and the fading margin becomes insufficient at the cell boundary. Coordinated multipoint transmission/reception (CoMP) technology has been introduced to overcome this problem. In this paper, we propose a CoMP system with Alamouti transmission diversity and a multiple input and multiple output (MIMO) configuration to further improve the average bit error rate (BER) at a cell boundary and the channel capacity. We show the effectiveness of our proposed system in terms of the coverage probability and channel capacity.

抄録全体を表示

Information-centric networking (ICN) is an emerging research area. Moreover, nowadays ICN is also applied to mobile ad hoc networks (MANETs). In a MANET, each node has limited resources. However, in many protocols, a node sends the interest packet by flooding. This increases traffic in the entire network. Furthermore, it causes duplicate content transmissions. To solve these problems, we propose an information-centric architecture using a mobile agent (MA). In this method, the MA manages location information and a content list of all nodes and constructs a route. The use of an MA can realize information acquisition without any flooding of interest packets. The performance evaluation results show the effectiveness of the proposed method in terms of power consumption, packet delivery rate, and content acquisition delay.

抄録全体を表示

A cooperative system of ultra-wideband (UWB) radar and an inter-vehicle communication (IVC) scheme was proposed in a previous study. However, in this system, when the time synchronization is not accurate, the ranging cannot be well performed. To solve this problem, a system that realizes time synchronization between the transmitter and the receiver by using the two-way ranging (TWR) method is proposed in this paper. This scheme is evaluated by computer simulations in an environment where the number of interference vehicles is increased. However, the reception characteristics deteriorate when the number of interference vehicles increases. In order to reduce the interference from other vehicles, a repetitive detection scheme using a successive interference cancellation (SIC) technique is adopted. As a result, better reception characteristics are obtained up to a distance of 90m without using SIC. When SIC is used, good reception characteristics are maintained up to 90m even when there are seven interference vehicles.

抄録全体を表示

In this paper, we present a nonparallel voice conversion (VC) approach that does not require parallel data or linguistic labeling for the training process. Dictionary-based voice conversion is a class of methods aiming to decompose speech into separate factors for manipulation. Non-negative matrix factorization (NMF) is the most common method to decompose an input spectrum into a weighted linear combination of a set comprising a dictionary (basis) and weights. However, the requirement for parallel training data in this method causes several problems: 1) limited practical usability when parallel data are not available, 2) the additional error from the alignment process degrades the output speech quality. To alleviate these problems, we present a dictionary-based VC approach by incorporating a variational autoencoder (VAE) to decompose an input speech spectrum into a speaker dictionary and weights without parallel training data. According to evaluation results, the proposed method achieves better speech naturalness while retaining the same speaker similarity as NMF-based VC even though unaligned data is used.

抄録全体を表示

This study proposes a loudspeaker system called the head-enclosed back surround loudspeaker array (HEBLA) that creates high realistic sound sensations. The proposed system can be combined with video to reproduce a sound field around the listener's head, but it is difficult to reproduce sound distance cues because the power ratio of direct sound to reverberation at the point of the listener's ears is different from that in an actual environment. The power ratio is an important cue for perceiving sound distance. Thus, this study proposes a sound image reproduction method which controls sound distance cues by restoring virtual front loudspeakers' room impulse responses (RIRs) with HEBLA. Evaluation experiments confirmed that the proposed method allows sound distance perception.

抄録全体を表示

Acoustic scene classification (ASC) classifies the place or situation where an acoustic sound was recorded. The Detection and Classification of Acoustic Scenes and Events (DCASE) 2017 Challenge prepared a task involving ASC. Some methods using convolutional neural networks (CNNs) were proposed in the DCASE 2017 Challenge. The best method independently performed convolution operations for the left, right, mid (addition of left and right channels), and side (subtraction of left and right channels) input channels to capture spatial features. On the other hand, we propose a new method of spatial feature extraction using CNNs. In the proposed method, convolutions are performed for the time-space (channel) domain and frequency-space domain in addition to the time-frequency domain to capture spatial features. We evaluate the effectiveness of the proposed method using the task in the DCASE 2017 Challenge. The experimental results confirmed that convolution operations for the frequency-space domain are effective for capturing spatial features. Furthermore, by using a combination of the three domains, the classification accuracy was improved by 2.19% compared with that obtained using the time-frequency domain only.

抄録全体を表示

This paper deals with a digital image colorization problem and proposes a novel colorization algorithm. To recover the color image from a given grayscale image and several color pixels, we introduce locally linear embedding (LLE). Each patch of an image is considered as a point on a manifold, and the proposed algorithm colorizes the grayscale image by LLE using the formulation of a block Hankel matrix. Numerical experiments show that the proposed algorithm achieves high recovery performance.

抄録全体を表示

This paper describes a prediction method for wind speed using a neural network and an investigation of the structure of the network. Generally, wind speed is observed as time series data, and the current wind speed is related to the past wind speed. Therefore, we propose a prediction model using long short-term memory (LSTM) and a one-dimensional convolutional neural network (1DCNN) in order to consider the past information for prediction. The prediction results of these networks and a fully connected neural network are compared for evaluation. The prediction accuracy and time delay are found to be improved by using LSTM and the 1D-CNN.

抄録全体を表示

This paper describes a safe driving support system of an electric wheelchair expected to be used by elderly people. However, accidents involving electric wheelchairs tend to occur when they are incorrectly operated. Improving safety and utility is a very important problem. To solve the problem, we propose a new safe driving support system, which controls a wheelchair semi autonomously based on the degree of danger. The degree of danger is calculated by fuzzy reasoning from both the manually controlled input from a joystick and the time to collision. Experimental results using our developed electric wheelchair demonstrate the high control performance of the proposed system.

抄録全体を表示

This paper describes an autonomous human-following control algorithm for a crawler-type mobile robot that works with an agricultural worker. To support an agricultural worker harvesting and transporting crops, the robot must follow the worker. The robot reduces the burden on agricultural workers by transporting crops. Therefore, we have developed a crawler-type mobile robot. In order to detect the worker and identify its own position, the robot senses around itself using only a laser range finder as an environmental recognition sensor. We propose an autonomous human-following control algorithm that can follow the worker automatically. The usefulness of the proposed control algorithm is demonstrated by experimental results using our developed crawler-type mobile robot.

抄録全体を表示

Many methods of sports video analysis have been proposed in the computer vision field. However, the analysis of swimming videos is a challenging task. This is because there is a lot of noise, such as water splashes, making it difficult to see the swimmer's motion and detect body parts. Thus, it is difficult to automatically estimate a swimmer's motion, especially the stroke. In this paper, we introduce a novel approach to automatically estimating the stroke in such situations. Firstly, we detect the swimmer from a swimming video using a projective transformation, background subtraction, and a Kalman filter. We next create a model that learns a mapping from a window of frames to a point on a one-dimensioned (1D) target signal, which represents a swimmer's stroke (we call a 'stroke signal'). We use a convolutional neural network (CNN) and multi long short-term memory (Multi-LSTM) which is an expanded model of LSTM. Finally, we estimate swimmer's stroke from the stroke signal. In a dataset including various environments, the outputs of our system showed higher accuracy than previous ones.

抄録全体を表示

Recently, most of the research on the skill assistance has been aimed at improving individual skills. However, a few assistance techniques consider cooperative skill in conjunction with others in a group. In cooperative work, the operator is affected by other operators. Thus, teamwork is needed in cooperative work. We propose assistance by force feedback in cooperative work using concern for others (CFO) which is one of the factors in teamwork. Each operator is given a force sense similar to a spring by a virtual impedance model that is changed on the basis of the estimated CFO. Experimental results suggest that CFO can be estimated and it was possible for force feedback to assist in cooperative work.

抄録全体を表示