電気学会論文誌C(電子・情報・システム部門誌)
Online ISSN : 1348-8155
Print ISSN : 0385-4221
ISSN-L : 0385-4221
最新号
選択された号の論文の18件中1~18を表示しています
特集:来るべきIoT社会に向けた情報通信技術
巻頭言
特集論文
<情報通信工学>
  • 佐藤 義人, 石山 亮, 武田 茂樹
    2020 年 140 巻 6 号 p. 552-558
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    In this paper, received power levels in 420M/ 920M/ 2.4G/ 5.3GHz bands for sensor networks and WLANs are obtained by ray-trace analysis in an office environment and the detail properties of propagation paths are analyzed. As a result, 25% average power of the total power arrives through ceiling spaces in the rooms where electromagnetic waves are hard to arrive through the walls. This means that the power remains in -6dB even though the propagation paths are blocked by several metallic furniture beside the walls. In this indoor environment, the propagation paths through ceiling spaces can be relatively good for wireless communication. The results confirm that modeling of structures in ceiling spaces is necessary to obtain practical received power levels.

  • 丹野 嘉信, 林 隆史
    2020 年 140 巻 6 号 p. 559-564
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    Various devices such as computer and peripheral device contribute to our lives. Because there is a wide variety of device, the types of device connector standard are increasing. A converter for converting the electrical signal of the pin in a connector is required for communication with different connector standard. However, there is a problem that the number of converters for interconnection continues to increase when the connector standard continues to increase. This paper proposes a communication method for the universal connector that can automatically communicate between different connector standard without the converter. The universal connector negotiates at the beginning of the connection using the electrical signal of the pin in the connector. The negotiation using the electrical signal of the pin requires a single-wire communication. Machine with single-wire communication requires to understand connection partner's performance because available machines with the universal connector do not strictly limit the partner. In this paper, we propose a novel single-wire communication method that each machine understands performance of connection partner for the universal connector. The performance of the proposed method was evaluated by simulation.

  • Gaku Kato, Takuji Tachibana
    2020 年 140 巻 6 号 p. 565-572
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    In this paper, in order to improve throughput and fairness in HetNet environment that consists of two different types of base stations (BSs), we propose a combined resource allocation for fronthaul and backhaul. In our proposed method, we define two optimization problems; one is a bandwidth allocation problem and the other is a frequency allocation problem. At first, in order to accommodate data transmission from user equipments (UEs) effectively, we determine a routing and bandwidth for all BSs in the backhaul by solving the bandwidth allocation problem. Then, in order to improve throughput and fairness of data transmission, we determine an allocation of frequency slots for all UEs by solving the frequency allocation problem. We evaluate the performance of the proposed method with simulation, and we investigate the effectiveness of the proposed method from the performance comparison. Numerical examples show that the proposed method is effective to increase throughput while improving the fairness among UEs.

<情報処理・ソフトウェア>
  • 石塚 康成, 倉田 正, 橋場 浩之, 成瀬 慶亮, 出口 弘
    2020 年 140 巻 6 号 p. 573-582
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    In the market increased diversification and globalization, manufacturing companies are making efforts to improve their business operations to enhance competitiveness. Business management, such as visualizing and analyzing business operations, is indispensable in such activities. Also, management to support manufacturing activities requires the utilization of IT technology such as IoT. However, we could not find any concept and framework of an information system that is for management by using state of manufacturing processes based on domain concepts in a manufacturing company's activity area and cooperates with their existing equipment. Therefore, we propose an easy-to-reconfigure state management system by using IoT based loosely coupled software components. We aim to realize that users of the system can manage and improve their activities by using states defined by their specific concepts, and acquired and visualized from manufacturing tools they have. In this paper, we propose a state management system model, and then we describe a method and results that we have demonstrated the system by using some manufacturing tools at a real manufacturing factory. Furthermore, based on the results, we discuss the usefulness of the state management system.

特集研究開発レター
<情報通信工学>
論文
<電気回路・電子回路>
<情報通信工学>
  • Zolzaya Kherlenchimeg, Naoshi Nakaya
    2020 年 140 巻 6 号 p. 592-599
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    In recent years, a deep neural network has been solving a variety of complex problems of science and engineering fields ranging from healthcare to transportation. Among them, one of the most crucial issues is to protect a network against cyber threats. In this article, we present a two-stage IDS framework based on a single-layer Sparse Autoencoder (SAE) and Long Short-Term Memory (LSTM), to design an effective network intrusion detection. Initially, the single-layer SAE learns new feature representations of the data through the nonlinear mapping, following that, the new feature representations are fed into the LSTM model to classify network traffic whether it is being normal or attack. The proposed framework was evaluated on the benchmark NSL-KDD dataset, where the mean accuracy of the proposed method was achieved 84.8%. The experimental results show that the two-stage IDS framework achieved better classification accuracy than the existing state-of-the-art methods.

<システム・計測・制御>
  • 石井 和行, 佐藤 直, 松井 俊浩
    2020 年 140 巻 6 号 p. 600-608
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    The received signal strength indicator (RSSI) is used to estimate the distance between a transmitter and receiver. We focus on the RSSI to localize the transmitter because it can be used to determine whether a receiver is moving toward or away from a transmitter. However, it has yet to be shown the way without the filtering or calibration to determine the approach or departure by RSSI. To determine it, here we study the method to maintain trustworthy judgement by delaying the judgement whether a receiver is approaching or departing continuously while a receiver moves. This proposal aims to prevent the misjudgment from the fluctuated RSSI and can control the moving direction of the receiver. We proposed the static condition and dynamic condition to respite the judgment, and we evaluated the conditions by a method for searching a signal source by simulation. We have shown that the misjudgment rate of the proposed conditions was better over 25% than the judgment not to use the proposed ones, and the search can be effectively performed due to preventing from the misjudgment. Therefore, these conditions are effective in determining the approach or departure under the environment of fluctuated RSSI.

<メディア情報,ユーザ・インタフェース>
  • 岡本 鉄兵, 神戸 英利, 小泉 寿男
    2020 年 140 巻 6 号 p. 609-619
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    In the user interface design of electric appliances, we propose a method to measure the consistency of the designer's mental model and the user's one, and that evaluate the validity of the user interface based on the results. The mental models are image formed in our head that affects human perception and decision making. As these are built from our experience and learning, individual results may vary. Thus, a gap in the mental model between designer and user is one of the factors in usability problems. In the method of the paper, the mental model is quantified by comparing dendrograms that classified product functions. The designer’s dendrogram is generated automatically from the design information by the clustering algorithm, and the user’s one is generated from the card sorting test. The product functions with a large gap in the mental model are identified based on the quantified results, and the designer can review the validity of the user interface. We applied the method to the user interface design of a digital camera and the results of experiments suggested the properness and effectiveness of the method.

<音声画像処理・認識>
  • 永島 徹, 長崎 健, 松原 仁
    2020 年 140 巻 6 号 p. 620-628
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    LIDAR has attracted attention as a surrounding environment recognition sensor in advanced driving support system (ADAS). The number of scan lines of LIDAR irradiated on objects varies, when setting LIDAR at different height. This variation causes the decrease of the recognition performance. We improved our proposal method with added pooling layer, and enhanced robustness against the variation of LIDAR setting height. And we classified the simulation data and KITTI dataset of pedestrians, cars, bicycles, motorbikes, and other objects. The results showed the our improved proposal method had robustness against the variation of LIDAR setting height.

  • 神谷 卓也, 山口 優太, 西山 正志, 岩井 儀雄
    2020 年 140 巻 6 号 p. 629-637
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    We propose a method for identifying people using body sway measured from head regions under a condition that a top-view camera observes bodies disturbed by self-occlusion. In order to correctly represent identities of body sway, it is necessary that the appearances of people are accurately acquired from the camera. However, the deficits of the appearances are contained when occuring self-occlusion. The deficits cause a problem that the identification performance of an existing method is wrongly degraded because the method uses whole-body regions to represent identities. To overcome the problem of self-occlusion, we represent identities of body sway by using head regions acquired from a top-view camera. Our method computes the silhouette images of the head regions by applying a segmentation technique. To represent identities of body sway, we temporally measure the movements of the head regions by spatially dividing them into local regions. The experimental results show that the identification performance improved from 17.3 % using the whole-body regions of the existing method to 57.9 % using the head regions of our method.

  • 浦添 和哉, 黒木 修隆, 加藤 裕, 大谷 真也, 廣瀬 哲也, 沼 昌宏
    2020 年 140 巻 6 号 p. 638-650
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    This paper proposes an image super-resolution using convolutional neural networks (CNNS) with multiple paths.

    After SRCNN was proposed by C. Dong et al., CNN-based super resolutions are getting larger and deeper. They do not work quickly without accelerators such as GPU any more. For practical use, however, we need to design CNNs with less internal parameters and low computational costs for convolution operation.

    The proposed CNN architecture consists of multiple paths with different depth. While a shallow path generates low frequency components, a deep path generates high frequency ones. Finally, they are synthesized at the last layer. This architecture can reduce the number of parameters relative to its performance.

    Experimental results have shown that the average processing time for the proposed CNN was only 25% of the conventional MCH while keeping high image qualities.

  • 鈴木 千文, 坂野 秀樹, 旭 健作, 森勢 将雅
    2020 年 140 巻 6 号 p. 651-663
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    In our previous study, a distance measure between two vibratos had been developed to evaluate singing style. This paper verifies whether the measure properly represents subjective distance by the following two subjective experiments. In these experiments, synthetic sounds whose vibratory amplitude computed from a target singer's singing is variable were used, and listeners were asked to answer reproducibility of individuality of the target singer for the synthetic sound. In the first experiment, listeners were asked to remember vibrato of the target singer by listening to the singing before the experiment. As a result, it was found that not only synthetic sounds showing small distance to the target singer in the distance measure but also those not showing small distance produced high subjective reproducibility, and the reproducibility score of the latter sounds is up to 1.2 times as higher as that of the former sounds, as long as the latter sounds have large vibratory amplitude, in other words, the sounds are just like the performance with deformation. To confirm whether this tendency is caused by the experimental procedure, the second experiment that listeners were asked to compare the synthetic sound with the reference singing sound of the target singer was conducted. This experiment also showed that subjective reproducibility of the synthetic sounds with large vibratory amplitude is higher than or almost equal to that of the reference singing sounds.

<ソフトコンピューティング・学習>
  • 二垣 良知, 井上 勝文, 吉岡 理文
    2020 年 140 巻 6 号 p. 664-672
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    Deep Neural Network (DNN) models have a great deal of parameters. It allows DNN to obtain good performance, however it also causes some problems. The first one is that learning of huge parameters requires enormous learning data for training DNN. The second one is that high-spec devices are requested because learning of huge parameters is computational complexity. These problems prevent the installation of DNN for any real tasks. To solve these problems, we propose a new learning method of DNN by combining transfer learning and knowledge distillation. The characteristic point of our proposed method is that we learn the DNN parameters by applying the techniques mentioned above simultaneously, i.e., we transfer the feature map of teacher DNN to student DNN, which is smaller than teacher DNN.

研究開発レター
<電子物性・デバイス>
  • Qiang Xu, Hailian Liang, Xiaofeng Gu
    2020 年 140 巻 6 号 p. 673-674
    発行日: 2020/06/01
    公開日: 2020/06/01
    ジャーナル 認証あり

    A novel dual-directional diode-triggered silicon controlled rectifier (DDTSCR) for low voltage electrostatic discharge (ESD) protection was designed and realized in a 0.18-µm CMOS process. Compared to the single-directional diode-triggered SCR (SDTSCR), the DDTSCR has dual-directional ESD protection performance due to the symmetric structure, and its ESD protection efficiency per unit area is about 2 times larger than that of SDTSCR under opposite ESD stresses, while remaining the similar trigger voltage of 1.68 V and the figure of merit. The human body model robustness of the DDTSCR measured by the transmission line pulse system is up to 8000 V with an area of 1400 µm2, suitable for low voltage ESD protection requirements.

部門記事
 
feedback
Top