Journal of Information Processing
Online ISSN : 1882-6652
ISSN-L : 1882-6652
27 巻

• Nobuo Kawaguchi
原稿種別: Special Issue of Pervasive System and Intelligent Transportation Systems for Future Life
2019 年 27 巻 p. 1
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー
• Mineo Takai, Jay Martin, Shigeru Kaneda, Taka Maeno
原稿種別: Special Issue of Pervasive System and Intelligent Transportation Systems for Future Life
専門分野: Invited Papers
2019 年 27 巻 p. 2-9
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー

This paper presents Scenario Generation and Management Framework for in-Depth Analysis and Extended Control, or simply Scenargie, for the analysis, evaluation and control of various networked systems. It was originally developed as a wireless network system simulator, and has significantly extended its target systems and capabilities for over a decade. This paper gives an overview of Scenargie and its primary modules with several use cases, and describes its simulation models and their fidelity. It also presents its system control capability recently added as part of its evolvement with two demonstrative projects.

• Kosuke Watanabe, Kei Hiroi, Takeshi Kamiyama, Hiroyuki Sano, Masakatsu ...
原稿種別: Special Issue of Pervasive System and Intelligent Transportation Systems for Future Life
専門分野: Invited Papers
2019 年 27 巻 p. 10-24
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー

In recent years, the importance of location information has increased due to the popularization of terminals such as smartphones. Our purpose is to estimate the 3D position of smartphones within several centimeters. This location information can reveal a person's behavior patterns and subject of interest. A method based on dynamic magnetism can estimate the 3D position of a terminal within several centimeters, and it is robust against the environment. However, since the performance of the smartphone's magnetic sensor is limited, this method cannot be used in smartphones. In this research, we propose a smartphone 3D positioning method using a Spinning Magnet Marker (SMM) that spins a powerful neodymium magnet with a motor. Magnetoquasistatic field generated by an SMM is detected by a magnetic sensor mounted on a smartphone, and the 3D position of a smartphone based on the SMM is estimated from the magnetism. We developed an SMM with a stepper motor to improve estimation accuracy. In addition, we studied the magnetism generated by an SMM and derive equations to estimate the 3D position of a smartphone. Furthermore, we improved the estimation accuracy and expanded coverage by introducing a noise reduction method. We evaluate the estimation accuracy of the proposed method. When the azimuth angle is 0°, the elevation angle is 0°, and the distance is 3m, the azimuth angle is estimated with a mean error of 1°, the elevation angle is estimated with a mean error of 4°, and the distance is estimated with a mean error of 9cm.

• Thongtat Oransirikul, Ian Piumarta, Hideyuki Takada
原稿種別: Special Issue of Pervasive System and Intelligent Transportation Systems for Future Life
専門分野: ITS
2019 年 27 巻 p. 25-32
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー

Quality of service is one factor passengers consider when deciding to use the public transportation system. Increasing the quality of service can be done in several ways, such as improving on-time arrival, reducing waiting time, and providing seat availability information. We believe that if passengers have better information for making decisions, that can increase the quality of service. This paper proposes estimating the number of passengers by analyzing signals from their Wi-Fi devices, classifying them as originating from passenger or non-passenger devices using a real-time filtering mechanism. Experimental validation was performed aboard busses of different types taking different routes. Our experimental results show that filtered data can classify passenger device signals from environmental ones with an accuracy of 75 percent, which is a promising basis for providing real-time information to passengers that improves the quality of their service.

• Bondan Suwandi, Teruaki Kitasuka, Masayoshi Aritsugi
原稿種別: Special Issue of Pervasive System and Intelligent Transportation Systems for Future Life
専門分野: Intelligent Robotics and Automation
2019 年 27 巻 p. 33-40
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー

In vehicle dead reckoning or vehicle positioning systems, an inertial measurement unit (IMU) sensor has an important role to provide acceleration and orientation of the vehicle. The acceleration from the IMU accelerometer is used to calculate the velocity of the vehicle, and then it estimates the vehicle's distance traveled to time. However, the accelerometer suffers from external noises such as vehicle vibrations (generated from the engine, alternator, compressor, etc) and road noises. This paper delivers deep analysis and focuses on how to handle the error from vehicle vibrations. A filter method is proposed by using a combination of adaptive least mean squares (LMS) and low-pass finite impulse response (FIR) filters. The adaptive LMS filter is used to cancel the vehicle vibration error frequencies and adapts those frequency changes in several engine rotation conditions. It is then finalized with the low-pass FIR filter which is used to filter high-frequency vibration noises. Several experiments were made and the results show that the proposed filtering method is able to give better signal to noise ratio (SNR dB) and noise attenuation ratio (ATT dB) in comparison with regular low-pass FIR filter and independent adaptive LMS filter in a particular condition.

• Hidekazu Shiozawa
原稿種別: Special Issue of Collaboration technologies and network services towards the society connecting all people and things
2019 年 27 巻 p. 41
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー
• Yunosuke Higashi, Masao Ohira, Yutaro Kashiwa, Yuki Manabe
原稿種別: Special Issue of Collaboration technologies and network services towards the society connecting all people and things
専門分野: Development Environments and Automated Technologies
2019 年 27 巻 p. 42-50
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー

• Yu Suzuki, Yoshitaka Matsuda, Satoshi Nakamura
原稿種別: Special Issue of Collaboration technologies and network services towards the society connecting all people and things
専門分野: Network Services
2019 年 27 巻 p. 51-60
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー

In microtask crowdsourcing, low-quality workers damage the quality of work results. Therefore, if a system automatically eliminates the low-quality workers, the requesters will obtain high-quality work results with low wages. When we consider simple Human Intelligent Tasks (HITs), such as yes-no questions of a labeling task, the requesters have difficulty assessing the worker quality only from the work results. Therefore, we need a method to accurately predict the worker quality automatically from the behaviors of workers, such as working time and the number of clicks. When we accurately predict the worker quality, we are able to prepare many features from the worker behaviors. However, when we submit simple HITs, we can capture only a small number of behaviors of workers, then the accuracy of predicted worker quality will be low. To solve this issue, we propose a method to insert into the simple task of obtaining many features of worker behaviors. We prepared a classification task of tweets as simple HITs. We added a button to the work screen. The workers can browse the target tweets on the work screen during the time the workers are pressing the button, but the workers cannot browse the target tweets when the workers have released the button. Using this button, we can obtain six more kinds of features of worker behaviors. Using our method, we can improve the recall ratio 12% of identifying low-quality workers. However, as the load of workers increases, then the processing time becomes longer, and the motivation of workers decreases. From this result, we also discovered that there is a trade-off between the number of obtained behaviors and the load of workers.

• Abu Nowshed Chy, Md Zia Ullah, Masaki Aono
原稿種別: Regular Papers
専門分野: Information Retrieval
2019 年 27 巻 p. 61-76
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー

In microblog search, vocabulary mismatch is a persisting problem due to the brevity of tweets and frequent use of unconventional abbreviations. One way of alleviating this problem is to reformulate the query via query expansion. However, finding good expansion terms for a given query is a challenging task. In this paper, we present a query expansion framework, where supervised learning is adopted for selecting expansion terms. Upon retrieving tweets by our proposed topic modeling based query expansion, we utilize the pseudo-relevance feedback and a new temporal relatedness approach to select the candidate tweets. Next, we devise several new features to select the temporally and semantically relevant expansion terms by leveraging the temporal, word embedding, and sentiment association of candidate term and query. Moreover, we also utilize the lexical and twitter specific features to quantify the term relatedness. After supervised feature selection using regularized regression, we estimate the feature importance by applying random forest. Then, we make use of a learning-to-rank (L2R) framework to rank the candidate expansion terms. Results of extensive experiments on TREC Microblog 2011 and 2012 test collections over the Tweets2011 corpus show that our proposed method outperforms the baseline and competitive query expansion methods.

• Yoshihiko Kakutani, Yuito Murase, Yuichi Nishiwaki
原稿種別: Regular Papers
専門分野: Special Section on Programming
2019 年 27 巻 p. 77-86
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー

In the stream of studies on intuitionistic modal logic, we can find mainly three kinds of natural deduction systems. For logical aspects, adding axiom schemata is a simple and popular way to construct a system. The Curry-Howard correspondence, however, gives us a connection between logic and computer science. From the viewpoint of programming languages, two more important systems, called a dual-context system and a Fitch-style system, have been proposed. While dual-context systems for S4 are heavily used in the field of staged computation, a dual-context system for K is also studied more recently. In our previous studies, categorical semantics for Fitch-style modal logic is proposed and usefulness of levels is noticed. This paper observes an interesting fact that the box modality of the dual-context system is in fact a left adjoint of that of the Fitch-style system. In order to show the statement, we embed both the two systems, which are refined with levels, into the adjoint calculus that equips an adjunction a priori. Moreover, the adjunction is refined with polarity and the adjoint calculus is extended to polarized logic.

• Yuya Kono, Hideyuki Kawabata, Tetsuo Hironaka
原稿種別: Regular Papers
専門分野: Special Section on Programming
2019 年 27 巻 p. 87-94
発行日: 2019年
公開日: 2019/01/15
ジャーナル フリー

The type class mechanism, which introduces ad-hoc polymorphism into programming languages, is commonly used to realize overloading. However, this forces programmers to write many type annotations in their programs to resolve ambiguous types. Haskell's type defaulting rules reduce requirements for annotation. Furthermore, the widely used Glasgow Haskell Compiler (GHC) has an ExtendedDefaultRules (EDR) extension that facilitates interactive sessions so that the programmer avoids problems that frequently occur when using values like [] and Nothing. However, the GHC EDR extension sometimes replaces type variables with inappropriate types, so that, for example, the term show.read that is determined to have type String -> String under the GHC EDR extension does not exhibit any meaningful behavior because the function read in the term is considered to have type String -> (). We present a flexible way of resolving ambiguous types that alleviates this problem. Our proposed method does not depend on default types defined elsewhere but rather assigns a type to a type variable only when the candidate is unique. It works with any type and type class constraints. The type to be assigned is determined by scanning a list of existing type class instances that meet the type class constraints. This decision is lightweight as it is based on operations over sets without using algorithms that require backtracking. Our method is preferable to using the GHC EDR extension since it avoids the use of unnatural type variable assignments. In this paper, we describe the details of our method. We also discuss our prototype implementation that is based on the GHC plugins, and the feasibility of modifying GHC to incorporate our method.

• Daisuke Sakamoto
原稿種別: Special Issue of Understanding, Technology, and Application of Interaction
2019 年 27 巻 p. 95
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー
• Kentaro Ueda, Tsutomu Terada, Masahiko Tsukamoto
原稿種別: Special Issue of Understanding, Technology, and Application of Interaction
専門分野: Embedded System Technology
2019 年 27 巻 p. 96-105
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

Although wearable interfaces can enable computer operation anywhere and at any time, many current devices are too cumbersome or socially unacceptable to wear in daily life. In contrast, a ridge of cloth produces a wrinkle that forms naturally on clothes, and the shape of these wrinkles can be recognized by their tactile sensations. Hence, they can achieve eye-free input and output functions. We therefore propose an interface that uses the wrinkles on clothes to input and output information. The interface generates several wrinkles on an item of clothing when an application requires input. The user senses the current state and number of choices by touching these wrinkles. The user can then input commands by selecting one wrinkle. The wrinkles disappear after the operation is complete. We implemented a prototype of this interface and investigated the social acceptance of the device position. The results show that the various device positions (excluding the armpit) achieve a moderate level of social acceptance. We mounted the prototype on the front of the right thigh of a pair of trousers and evaluated the recognition accuracy of the wrinkle patterns as well as the learning curve with respect to input accuracy and time. The participants achieved a recognition rate of 89.4%, input accuracy of 95% or more, and an input time of approximately 2s. This device has the potential to provide unobtrusive and eyes-free operation and the current results indicate directions for its future development.

• Ryosuke Takada, Toshiyuki Ando, Buntarou Shizuki, Shin Takahashi
原稿種別: Special Issue of Understanding, Technology, and Application of Interaction
専門分野: User Interfaces and Interactive Systems
2019 年 27 巻 p. 106-115
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

We present BaroTouch, a technique that leverages a waterproof mobile device's built-in barometer to measure the touch force. When an airtight waterproof device is touched, the distorted surface changes the air pressure inside that device and thus changes the built-in barometer value. BaroTouch estimates the touch force with a simple conversion from the air pressure to the touch force; it does not need machine learning. In addition, BaroTouch is a passive sensing technique. Therefore, BaroTouch could be lightweight. To investigate BaroTouch, we conducted two experiments. First, we investigated the relationship between the sensor value and the touch positions or forces using weights with three devices: two smartphones and a smartwatch. Second, in a controlled user study with 15 participants, we examined how the users can use BaroTouch. The results showed that the participants could exert 2-6 levels of the touch force with accuracies of over 96% accuracy under each device using BaroTouch.

• Masashi Yamamuro
原稿種別: Special Issue of Network Services and Distributed Processing
2019 年 27 巻 p. 116
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー
• Takahiro Noi, Takuya Yoshihiro
原稿種別: Special Issue of Network Services and Distributed Processing
専門分野: Basic System Software Technology
2019 年 27 巻 p. 117-124
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

In this paper, we propose a new scheduling problem for WMNs based on slotted-CSMA. Slotted CSMA is a mechanism that divides a single frequency channel into several time slots where CSMA functions. With a schedule that matches links and slots, each node transmits frames in the assigned slot to avoid collision of frames. CATBS (CSMA-Aware Time-Boundable Scheduling) is a slotted-CSMA-based WMN architecture with a scheduling algorithm. However, it suffers from large end-to-end delay due to its long slot time that allows transmitting several frames within a single slot. This paper extends the scheduling problem of CATBS to consider inter-slot collision to reduce the overhead of collision at slot boundaries even if using short slot time. Evaluation results shows that the proposed scheduling problem reduces the overhead that arises at slot boundaries, and improves communication performance when using short time slots.

• Junji Takemasa, Yuki Koizumi, Toru Hasegawa
原稿種別: Special Issue of Network Services and Distributed Processing
専門分野: Network Architecture
2019 年 27 巻 p. 125-134
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

This paper addresses how to design a cache algorithm for achieving high cache hit rate and high packet forwarding rate of Named Data Networking (NDN) software routers. Although sophisticated cache eviction algorithms like LFU and Adaptive Replacement Cache (ARC) successfully achieve high cache hit rate, they incur heavy computational overheads to choose a victim Data packet from the cache. In contrast, cache admission is expected to achieve high cache hit rate with light-weight computation since it decides simply whether an incoming Data packet should be inserted into the cache or not. In this paper, we design a frequency-based cache admission algorithm, Filter, with light-weight computation by simply counting frequencies of incoming Data packets in a fixed time window. A simulation-based evaluation proves that Filter achieves high cache hit rate comparable to sophisticated cache eviction algorithms like ARC. By implementing a prototype of an NDN software router with Filter, we validate that the NDN router with Filter improves packet forwarding rate compared to that with a sophisticated cache eviction algorithm like ARC and even that with a simple one like FIFO.

• Hironori Nakajo
原稿種別: Special Issue of Embedded Systems Engineering
2019 年 27 巻 p. 135
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー
• Yang Qin, Gang Zeng, Ryo Kurachi, Yutaka Matsubara, Hiroaki Takada
原稿種別: Special Issue of Embedded Systems Engineering
専門分野: Embedded System Technology
2019 年 27 巻 p. 136-148
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

Energy-aware task allocation of embedded systems is one of the most important issues in recent decades. A classical solution to solve the issue is Integer Linear Programming (ILP). However, given the considerable time consumption, it is effective only to the extent that the scale of the problem is small. How to use ILP to solve large allocation problems on heterogeneous multiprocessor systems to minimize energy consumption is still a challenge. This paper proposes two ILP formulations to deal with it. One complete ILP(1) is used to derive a feasible allocation, and the other simplified ILP(2) is for calculating the desired minimum energy. Then the desired minimum energy can be used as a reference to evaluate the intermediate solution of ILP(1) and decide its timeout. Besides, to find out the best-suited platform for a given workload, a flexible design which presents flexibilities and choices in core assignment, is considered for further energy saving. For example, the optimal core number design and core type design are generated as two independent ILP formulations, denoted as ILP(3) and ILP(4). The experimental results on randomly generated task sets demonstrate that, compared with the fixed platform, automatically synthesizing a flexible core assignment saves more energy.

• Huatao Zhao, Jiongyao Ye, Takahiro Watanabe
原稿種別: Special Issue of Embedded Systems Engineering
専門分野: Embedded System Technology
2019 年 27 巻 p. 149-158
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

Nowadays, on-chip cache scales are oversized in multicore embedded systems, and those caches even consume half of the total energy debit. However, we observe that a large portion of cache banks are wasted, meaning that those banks are rarely used but consume a great deal of energy during their entire lifetime. In this paper, we propose a controllable shared last level cache (SLLC) scheme to dynamically trace cache bank demands for each thread. Thus, energy on useless banks can be largely saved. Specifically, we (1) propose an effective cache bank allocating policy to explore bank demands corresponding to executed applications, (2) discuss the interrelations between application locality change and bank demands for further energy saving and (3) represent a modified PID controller to generate optimal banks for each thread. Experimental results show that our controllable SLLC design can save on average 39.7 percent shared cache access energy over conventional cache, while its performance is slightly improved and additional hardware overhead is less than 0.6 percent.

• Junko Takahashi, Masashi Tanaka, Hitoshi Fuji, Toshio Narita, Shunsuke ...
原稿種別: Special Issue of Embedded Systems Engineering
専門分野: System Security
2019 年 27 巻 p. 159-167
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

We present a method for influencing vehicle behaviors using only informative Controller Area Network (CAN) messages that are used to inform vehicle status. Some recent vehicle attack techniques have been shown to have a significant impact on the automotive industry. Almost all previous studies employ active CAN messages that directly induce actions for the attacks, but there have been no studies that explicitly use only the informative CAN messages. This is the first report of using only informative CAN messages in an attack especially targeting a driving-support system. Through experiments, we show that abrupt acceleration/deceleration is abnormally induced using informative messages regarding the wheel speed while the cruise control system is activated. We also find that the speed limit control of the cruise control system can be disabled and the parking assist system can be canceled using fabricated informative messages. The experimental results reveal that fabricated informative CAN messages can manipulate the vehicle to yield an improper behavior. We evaluate the effectiveness of some countermeasures by applying them to the attacks and clarify the strength and weakness of each countermeasure. We believe that this study will bring a new perspective to the automotive security toward vehicle system design.

• Ryo Masumura, Taichi Asami, Takanobu Oba, Hirokazu Masataki, Sumitaka ...
原稿種別: Regular Papers
専門分野: Speech Processing
2019 年 27 巻 p. 168-176
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

This paper presents a Viterbi approximation of latent words language models (LWLMs) for automatic speech recognition (ASR). The LWLMs are effective against data sparseness because of their soft-decision clustering structure and Bayesian modeling, so LWLMs can perform robustly in multiple ASR tasks. Unfortunately, implementing an LWLM to ASR is difficult because of its computation complexity. In our previous work, we implemented an n-gram approximation of LWLM for ASR by sampling words according to a stochastic process and training word n-gram LMs. However, the previous approach cannot take into account a latent word sequence behind a recognition hypothesis. Our solution is the Viterbi approximation that simultaneously decodes both the recognition hypothesis and the latent word sequence. The Viterbi approximation is implemented as a two-pass ASR decoding in which the latent word sequence is estimated from a decoded recognition hypothesis using Gibbs sampling. Experiments show the effectiveness of the Viterbi approximation in an n-best rescoring framework. In addition, we investigate the relationship of the n-gram approximation and the Viterbi approximation.

• Masahiro Yasugi, Reichi Ikeuchi, Tasuku Hiraishi, Tsuneyasu Komiya
原稿種別: Regular Papers
専門分野: Special Section on Programming
2019 年 27 巻 p. 177-189
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

Scheme implementations should be properly tail-recursive and support garbage collection. To reduce the development costs, a Scheme interpreter called JAKLD, which is written in Java, was designed to use execution stacks simply. JAKLD with interchangeable garbage collectors was reimplemented in C. In addition, we have proposed an efficient C-based implementation written in an extended C language called XC-cube, which features language mechanisms for implementing high-level programming languages such as “L-closures” for legitimate execution stack access, with which a running program/process can legitimately access data deeply in execution stacks (C stacks). L-closures are lightweight lexical closures created from nested function definitions. In addition to enhanced C compilers, we have portable implementations of L-closures, which are translators from an extended S-expression based C language into the standard C language. Furthermore, we have another mechanism for legitimate execution stack access, called “closures”. Closures are standard lexical closures created from nested function definitions. Closures can also be implemented using translators. In this study, JAKLD was reimplemented in an extended SC language (S-expression based C language) that features nested functions to evaluate (L-)closures and their implementations, including translators.

• Chen Zhao, Bin Yang, Yu Hirate
原稿種別: Regular Papers
専門分野: Special Section on Databases
2019 年 27 巻 p. 190-200
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

This paper designs a novel predictive model that learns stochastic functions given a limited set of data samples. Interpolation algorithms are commonly seen in supervised learning applications for function approximation by constructing models generalizable to unseen data. However, parametric models such as regression and linear SVMs are limited to functions in the form of predefined algebraic expressions and are thus unsuitable for arbitrary functions without finite number of parameters. While properly trained neural networks are capable of computing universal functions, the amount of required training data can be prohibitively large in some practical scenarios such as online recommendation. The proposed model addresses both problems based on a semi-parametric graphical model that approximates function outputs with limited data samples through Bayesian optimization. An online algorithm is also presented to show how model inference is used to locate global optima of an unknown function, as the primary objective of making optimal decisions. Comparative experiments are conducted among a set of sampling policies to demonstrate how click-through rates can be improved by optimized recommendation strategy with the proposed model. Empirical evaluation suggests that an adapted version of Thompson sampling is the best suitable policy for the proposed algorithm.

• Senxi Li, Takeshi Iwashita, Takeshi Fukaya
原稿種別: Regular Papers
専門分野: Special Section on Advanced Computing Systems
2019 年 27 巻 p. 201-210
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

Algebraic block multi-color ordering is known as a parallelization method for a sparse triangular solver. In the previous work, we confirmed the effectiveness of the method in a multi-threaded ICCG solver for a linear system with a symmetric coefficient matrix. In this study, we enhance the method so as to deal with an unsymmetric coefficient matrix. We develop a multi-threaded ILU-GMRES solver based on the enhanced method and evaluate its performance in terms of both the runtime and the number of iterations.

• An Huynh, Christian Helm, Shintaro Iwasaki, Wataru Endo, Byambajav Nam ...
原稿種別: Regular Papers
専門分野: Special Section on Advanced Computing Systems
2019 年 27 巻 p. 211-220
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

• Shun Kurihara, Shoki Fukuda, Takeshi Kamiyama, Akira Fukuda, Masato Og ...
原稿種別: Regular Papers
専門分野: Special Section on Consumer Device & System
2019 年 27 巻 p. 221-232
発行日: 2019年
公開日: 2019/02/15
ジャーナル フリー

Some reports stated that the most important issue of smartphones is their large battery consumption. Information on the power consumption of each application is important for users and administrations of application distributing sites. Especially, information on power consumption of each application in the screen-off state is important because understanding the behavior of an application in the state is difficult. Naturally, the power consumption of a device increases and decreases by installing and uninstalling an application, respectively. However, the sizes of increase and decrease in power consumptions depend on the device. We think there are two types of dependencies, which are hardware and software dependencies. The hardware dependency is that the power consumption of an application depends on the hardware elements of the device. The software dependency is that the power consumption of an application depends on the other applications installed on the device. We then argue that consideration of these dependencies are essential for estimation of the power consumption of each application. In this paper, we focus on the software dependency and propose a method for estimating the size of increase and decrease in power consumptions of the device by installing and uninstalling an application considering software dependency. The proposed method monitors starts and ends of functions such as GPS usage and WakeLock, then estimates the parts of the power consumptions of each application separately. We estimate the GPS usage time and WakeLock time for evaluation of the proposed method and show that the proposed method can estimate these more accurately than the standard method of the Android operating system. Our evaluation demonstrated that the proposed method decreased the difference between the estimated and actual sizes of decreases in power consumption by 89% at most.

• Yoshiaki Kitaguchi
原稿種別: Special Issule of the Internet and operation technologies by constructing resilient information systems
2019 年 27 巻 p. 233
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー
• Motoyuki Ohmori, Masayuki Higashino, Toshiya Kawato, Satoshi Fujio, Ki ...
原稿種別: Special Issule of the Internet and operation technologies by constructing resilient information systems
専門分野: Operation
2019 年 27 巻 p. 234-243
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー

Computer security has been getting more attention because a computer security incident may cause great damage on an organization. A quick and correct response against an incident is then important. One of the first possible responses is then locating and isolating a suspicious host. This isolation typically requires a manual operation that may cause a mistake or long delay. In order to solve these issues, this paper proposes a novel system to locate and isolate a suspicious host on an incident response adopting the Software Defined Network (SDN) approach. This SDN approach allows the proposed system to locate and isolate a suspicious host on-demand in a network that comprises different switches and routers of different makers. The proposed system then requires no host authentication configured, no IP address allocation/assignment database, no network topology map and no switch port list in advance. The proposed system, therefore, can reduce human manual operations. This paper then presents that human manual operations actually induce longer delays, more than 3 minutes on average, and also cause mistakes. This paper also presents that the proposed system can locate and isolate a suspicious host within 10 seconds right after an IP address of a suspicious host is given.

• Etsuya Shibayama
原稿種別: Special Issue of Young Researchers' Papers
2019 年 27 巻 p. 244
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー
• Thach V. Bui, Minoru Kuribayashi, Tetsuya Kojima, Roghayyeh Haghvirdin ...
原稿種別: Special Issue of Young Researchers' Papers
専門分野: Information Mathematics
2019 年 27 巻 p. 245-256
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー

The task of non-adaptive group testing is to identify up to d defective items from N items, where a test is positive if it contains at least one defective item, and negative otherwise. If there are t tests, they can be represented as a t × N measurement matrix. We have answered the question of whether there exists a scheme such that a larger measurement matrix, built from a given t × N measurement matrix, can be used to identify up to d defective items in time O(t log2 N). In the meantime, a t × N nonrandom measurement matrix with $t = O \left(\frac{d^2 \log_2^2{N}}{(\log_2(d\log_2{N}) - \log_2{\log_2(d\log_2{N})})^2} \right)$ can be obtained to identify up to d defective items in time poly(t). This is much better than the best well-known bound, $t = O \left(d^2 \log_2^2{N} \right)$. For the special case d = 2, there exists an efficient nonrandom construction in which at most two defective items can be identified in time $4\log_2^2{N}$ using $t = 4\log_2^2{N}$ tests. Numerical results show that our proposed scheme is more practical than existing ones, and experimental results confirm our theoretical analysis. In particular, up to 27 = 128 defective items can be identified in less than 16s even for N = 2100.

• Masashi Nishimoto, Keiji Nishiyama, Hideyuki Kawabata, Tetsuo Hironaka
原稿種別: Special Issue of Young Researchers' Papers
専門分野: Programming Environments
2019 年 27 巻 p. 257-267
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー

Today's application development process depends heavily on the usage of application programming interfaces (APIs) for many kinds of frameworks. Time spent searching for appropriate API members and understanding their usages tends to occupy much of the time required for the whole development process. This paper proposes a new approach for developing application programs based on APIs in a simple way: through code development by iterating a Search-Select-Superpose (SSS) loop. The approach comprises three phases. In the Search phase, the user searches for a way to implement a desired functionality by combining API calls. The search results are shown to the user as a list of outlines (sets of words) attached to code skeletons. A code skeleton, chosen in the Select phase, is then merged with the program at hand in the Superpose phase. The entire process is implemented through the construction of an indexed dataset composed of code skeletons extracted from open-source repositories, and through the use of a tool to control the SSS loop. We have developed a prototype of the proposed system. In this paper, the design and implementation of the proposed system are described. The effectiveness of the system was confirmed through empirical results from experiments with event-driven Android application development.

• Tran Duy Linh, Masayuki Arai
原稿種別: Special Issue of Young Researchers' Papers
専門分野: Machine Learning & Data Mining
2019 年 27 巻 p. 268-277
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー

In the present paper, we propose a deep network architecture in order to improve the accuracy of general object detection. The proposed method contains a proposal network and a classification network, which are trained separately. The proposal network is trained to extract a set of object candidates. These object candidates cover not only most object ground truths but also a number of false positives. In order to make the detector more robust, we train these object candidates using a secondary classifier. We propose combination methods and prove that a combination of two networks is more accurate than a single network. Moreover, we determine a new method by which to optimize the final combination results. We evaluate the proposed model using several object detection datasets (Caltech pedestrian, Pascal VOC, and COCO) and present results for comparison.

原稿種別: Special Issue of Young Researchers' Papers
専門分野: Music Information Processing
2019 年 27 巻 p. 278-286
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー

In this paper, we present melody2vec, an extension of the word2vec framework to melodies. To apply the word2vec framework to a melody, a definition of a word within a melody is required. We assume phrases within melodies to be words and acquire these words via melody segmentation applying rules for grouping musical notes called Grouping Preference Rules (GPR) in the Generative Theory of Tonal Music (GTTM). We employed a skip-gram representation to train our model using 10, 853 melody tracks extracted from MIDI files primarily constructed from pop music. Experimental results show the effectiveness of our model in representing the semantic relatedness between melodic phrases. In addition, we propose a method to edit melodies by replacing melodic phrases within a musical piece based on the similarity of the phrase vectors. The naturalness of the resulting melody was evaluated via a user study and most participants who did not know the musical piece could not point out where the melody had been replaced.

• Hirotaka Suzuki, Haruhisa Ichikawa, Jin Mitsugi, Yuusuke Kawakita
原稿種別: Regular Papers
専門分野: Wireless/Mobile Networks
2019 年 27 巻 p. 287-296
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー

Software-defined radio (SDR) is used for R&D such as cognitive radio. Because sampling devices and personal computers configuring the SDR environment have fixed configurations, some reconfiguration is needed when the SDR application requires different data sources and computational resources. To enable reconfiguration, we present a cloud platform that has scalable computing resources and data sources deployed over a wide area. We use the existing SDR development environment (SDRDE) and implement it on the cloud platform. It is necessary to transfer the SDR environment to the cloud by separating the UI and data management from the existing SDRDE. In this study, we selected the GNU Radio Companion (GRC) as the base platform and implemented an SDRDE for an unspecified number of users by separating the UI. In addition, we used task parallel and distributed computing for the SDR application. In this study, we focus on compatibility with the base implementation and lifting the limits of computational resources. We confirmed the compatibility with GRC in terms of user skill sets and software assets and evaluated the system response time. Further, the relationship between CPU utilization and instructions per cycle during SDR application execution shows that in general, this approach is effective.

• Yuhei Kawakoya, Eitaro Shioji, Makoto Iwamura, Jun Miyoshi
原稿種別: Regular Papers
専門分野: System Security
2019 年 27 巻 p. 297-314
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー

We propose a design and implementation for an Application Programming Interface (API) monitoring system called API Chaser, which is resistant to evasion-type anti-analysis techniques, e.g., stolen code and code injection. The core technique in API Chaser is code tainting, which enables us to identify precisely the execution of monitored instructions by propagating three types of taint tags added to the codes of API, malware, and benign executables, respectively. Additionally, we introduce taint-based control transfer interception, which is a technique to capture precisely API calls invoked from evasive malware. We evaluate API Chaser based on several real-world and synthetic malware to demonstrate the accuracy of our API hooking technique. We also perform a large-scale malware experiment by analyzing 8, 897 malware samples to show the practical capability of API Chaser. These experimental results show that 701 out of 8, 897 malware samples employ hook evasion techniques to hide specific API calls, while 344 malware ones use target evasion techniques to hide the source of API calls.

• Kazuhiro Seki
原稿種別: Regular Papers
専門分野: Knowledge Processing
2019 年 27 巻 p. 315-321
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー

Accurately computing the similarity between two texts written in different languages has tremendous value in many applications, such as cross-lingual information retrieval and cross-lingual text mining/analytics. This paper studies the important problem based on neural networks. Specifically, our focus is on the neural machine translation models. While translation models are utilized, we pay special attention not to the translation itself but to the intermediate states of given texts stored in the translation models. Our assumption is that the intermediate states capture the syntactic and semantic meaning of input texts and are a good representation of the texts, avoiding inevitable translation errors. To study the validity of the assumption, we investigate the utility of the intermediates states and their effectiveness in computing cross-lingual text similarity in comparison with other neural network-based distributed representations of texts, including word and paragraph embedding-based approaches. We demonstrate that an approach using the intermediate states outperforms not only these approaches but also a strong machine translation-based one. Furthermore, it is revealed that intermediate states and translated texts work complementarily each other despite the fact that they are generated from the same NMT models.

• Mihiro Sonoyama, Takatsugu Ono, Haruichi Kanaya, Osamu Muta, Smruti R. ...
原稿種別: Regular Papers
専門分野: Special Section on Advanced Computing Systems
2019 年 27 巻 p. 322-334
発行日: 2019年
公開日: 2019/03/15
ジャーナル フリー

A spoofing attack is a critical issue in wireless communication in which a malicious transmitter outside a system attempts to be genuine. As a countermeasure against this, we propose a device-authentication method based on position identification using radio-propagation characteristics (RPCs). Not depending on information processing such as encryption technology, this method can be applied to sensing devices etc. which commonly have many resource restrictions. We call the space from which attacks achieve success as the “attack space.” In order to confine the attack space inside of the target system to prevent spoofing attacks from the outside, formulation of the relationship between combinations of transceivers and the attack space is necessary. In this research, we consider two RPCs, the received signal strength ratio (RSSR) and the time difference of arrival (TDoA), and construct the attack-space model which uses these RPCs simultaneously. We take a tire pressure monitoring system (TPMS) as a case study of this method and execute a security evaluation based on radio-wave-propagation simulation. The simulation results assuming multiple noise environments all indicate that it is possible to eliminate the attack possibility from a distant location.

• Yoshinao Ishii, Masaki Takanashi
原稿種別: Regular Papers
専門分野: Knowledge Processing
2019 年 27 巻 p. 335-339
発行日: 2019年
公開日: 2019/04/15
ジャーナル フリー

Recently, an unsupervised outlier detection method based on the reconstruction errors of an autoencoder (AE), which achieves high detection accuracy, was proposed. This method, however, requires a high calculation cost because of its ensemble scheme. Therefore, in this paper, we propose a novel AE-based unsupervised method that can achieve high detection performance at a low calculation cost. Our method introduces the concept of robust estimation to appropriately restrict reconstruction capability and ensure robustness. Experimental results on several public benchmark datasets show that our method outperforms well-known outlier detection methods and at a low calculation cost.

• Naoki Kondo, Kiminori Matsuzaki
原稿種別: Regular Papers
専門分野: Game Informatics
2019 年 27 巻 p. 340-347
発行日: 2019年
公開日: 2019/04/15
ジャーナル フリー

Game 2048 is a stochastic single-player game and development of strong computer players for Game 2048 has been based on N-tuple networks trained by reinforcement learning. Some computer players were developed with (convolutional) neural networks, but their performance was poor. In this study, we develop computer players for Game 2048 based on deep convolutional neural networks (DCNNs). We increment the number of convolution layers from two to nine, while keeping the number of weights almost the same. We train the DCNNs by applying supervised learning with a large number of play records from existing strong computer players. The best average score achieved is 93, 830 with five convolution layers, and the best maximum score achieved is 401, 912 with seven convolution layers. These results are better than existing neural-network-based players, while our DCNNs have less weights.

• Ayumi Ohnishi, Tsutomu Terada, Masahiko Tsukamoto
原稿種別: Regular Papers
専門分野: User Interfaces and Interactive Systems
2019 年 27 巻 p. 348-358
発行日: 2019年
公開日: 2019/04/15
ジャーナル フリー

In this paper, we propose a method for recognizing postures and gestures by using foot pressure sensors, and we investigate optimal positions for pressure sensors on soles from the viewpoint of motion recognition accuracy. In experiments, the recognition accuracies of 22 kinds of daily postures and gestures were evaluated from foot-pressure sensor values. Furthermore, the optimum measurement points for high recognition accuracy were examined by evaluating combinations of two foot pressure measurement areas on a round-robin basis. As a result, when selecting the optimum two points for each user, the recognition accuracy was about 94.5% on average. The recognition accuracy of the averaged combinations of the best two combinations for all subjects was classified with an accuracy of about 91.9% on average. As a result of an evaluation to raise versatility, the average recognition accuracy in a three-point evaluation was 98.4%, which was almost the same with the recognition accuracy when using all 105 points. In anticipation of the applicability of this research result, two types of pressure sensing shoes were developed.

• Yoshiaki Matsuzawa
原稿種別: Special Issule of Information Systems
2019 年 27 巻 p. 359
発行日: 2019年
公開日: 2019/05/15
ジャーナル フリー
• Van Cu Pham, Yuto Lim, Antonio Sgorbissa, Yasuo Tan
原稿種別: Special Issule of Information Systems
専門分野: Network Services
2019 年 27 巻 p. 360-368
発行日: 2019年
公開日: 2019/05/15
ジャーナル フリー

ECHONET Lite is a leading protocol for controlling devices in Japan smart homes. However, it lacks interoperability with service platforms that provide ambient assisted living (AAL) services to residents which are actively researched in order to deal with the population aging. This research proposes an adaptation layer for ECHONET Lite protocol which provides the semantic interoperability based on ontology. In order to verify the proposed solution, a service gateway based on the proposed architecture was implemented to integrate ECHONET Lite protocol into the universAAL platform, a leading AAL platform in Europe.

• Wataru Aoyama, Hideya Iwasaki
原稿種別: Regular Papers
専門分野: Basic System Software Technology
2019 年 27 巻 p. 369-377
発行日: 2019年
公開日: 2019/05/15
ジャーナル フリー

A journaling file system is a file system that records information about pending updates to the file system before committing the updates. This mechanism raises the reliability of the system because it enables any inconsistencies to be repaired with minimal loss of data. Since there is a tradeoff between the overhead and reliability, ext3, a journaling file system commonly used by the Linux kernel, offers three journaling modes: speed-prioritized mode, reliability-prioritized mode, and intermediate mode. Unfortunately, in ext3, the journaling mode has to be set individually for each file system. Thus, the granularity of the journaling mode setting is very coarse. In addition, the journaling mode must be determined at the time of mounting the file system and cannot be changed without unmounting it. To resolve this problem, this paper proposes a new journaling file system named dajFS (per-directory adaptive journaling file system) that is able to set an appropriate journaling mode for each directory and to switch the journaling mode of a directory to another on the fly without unmounting the file system. Essentially, the journaling mode that is specified for a directory applies to all files that reside directly under that directory. By using dajFS, the user can determine and set a journaling mode for each directory on the basis of the importance of files under that directory. As a result, the user can enjoy moderate granularity with the journaling mode setting.

• Takeshi Miura, Takeshi Shibata, Madoka Uemura, Katsubumi Tajima, Hideo ...
原稿種別: Regular Papers
専門分野: Body Information
2019 年 27 巻 p. 378-391
発行日: 2019年
公開日: 2019/05/15
ジャーナル フリー

This paper proposes a cartogram construction method to visualize the relevance of the motion-characteristic distribution of Japanese folk dances to the geographic elements of regional communities in which the dances have been passed down. We use motion capture data of the dances to quantitatively extract their motion characteristics. To systematically organize the cartogram construction process, we adopt a hierarchical model representing the relationship among motion capture data, folk dances, and settlements in which the dances have been passed down. Different cartogram types are selected for different levels in the hierarchical model, and, thereby, a hybrid of circle and distance cartograms is provided. We show that an algorithm to locate the constituents of the hierarchical model in the above hybrid cartogram can be obtained by slightly modifying the existing circle cartogram construction algorithm. On the other hand, we develop another new algorithm to locate geographic elements other than the constituents of the hierarchical model in the hybrid cartogram. The results obtained by analyzing the Furyu type folk dances passed down in Akita Prefecture demonstrated the effectiveness of the proposed method.

• Koji Kawamata, Kenta Oku
原稿種別: Regular Papers
専門分野: Special Section on Databases
2019 年 27 巻 p. 392-403
発行日: 2019年
公開日: 2019/05/15
ジャーナル フリー

• Yu Suzuki
原稿種別: Regular Papers
専門分野: Special Section on Databases
2019 年 27 巻 p. 404-410
発行日: 2019年
公開日: 2019/05/15
ジャーナル フリー

A large number of texts is posted daily on social media. However, only a small portion of these texts is informative for a specific purpose. For example, in order to collect a set of tweets for marketing strategy, we should collect a large number of tweets related to a specific topic with high accuracy. If we accurately filter the texts, we can continuously obtain fresh and useful information in real time. In a keyword-based approach, filters are constructed using keywords, but selecting the appropriate keywords is often tricky. In this work, we propose a method for filtering texts that are related to specific topics using a classification method that is based on crowdsourcing and machine learning. In our approach, we construct a text classifier using fastText and then annotate whether the tweets are related to the topics using crowdsourcing. For constructing an accurate classifier, we should prepare a large amount of learning data. However, this process is costly and time-consuming. To construct an accurate classifier using a small number of learning data, we consider two strategies for selecting tweets which the crowdsourcing participants should assess: optimistic and pessimistic approach. Then, we reconstruct the text classifier using the annotated texts and classify them again. If we continue instigating this loop, the accuracy of the classifier will improve, and we will obtain useful information without having to specify the keywords. Experimental results demonstrate that our proposed system is adequate for filtering social media streams. Moreover, we discovered that the pessimistic approach is better than the optimistic approach.

• Akimasa Morihata
原稿種別: Regular Paper
専門分野: Special Section on Programming
2019 年 27 巻 p. 411-421
発行日: 2019年
公開日: 2019/06/15
ジャーナル フリー

Parallel evaluation of tree processing using accumulation parameters tends to be difficult because the accumulation parameters may introduce data dependencies between computations for subtrees. Some proposals have broken these data dependencies by using algebraic properties such as associativity and commutativity, but, none has achieved both the capability of complex tree traversals like attribute grammars and a theoretical guarantee of parallel speedup. This paper proposes a tree processing language based on macro tree transducers and provides a parallel evaluation algorithm for programs written in the language. The language can express complex accumulations like attribute grammars, and moreover, the number of parallel computational steps for evaluation is proportional to the height of the input tree. This paper also shows that combining the proposed approach with function fusion for macro tree transducers leads to improvement in the parallel computational complexity. Although comparable complexity improvement can be obtained from known parallel algorithms, the proof and parallel evaluation algorithm here are remarkably simpler.

• Takayuki Miyazaki, Yasuhiko Minamide
原稿種別: Regular Paper
専門分野: Special Section on Programming
2019 年 27 巻 p. 422-430
発行日: 2019年
公開日: 2019/06/15
ジャーナル フリー

Lookahead is an extension of regular expressions that has been adopted in many implementations and is widely used. Lookahead represents what is allowed as the rest of input. Morihata developed a conversion from regular expressions with lookahead (REwLA) to deterministic finite automata by extending Thompson's construction. In this paper, we develop a conversion from REwLA to deterministic finite automata by extending derivatives of regular expressions. First, we formalize the semantics of REwLA. An REwLA has information about the rest of the input, so the definition of the semantics of REwLA is not languages but structures different from those of regular expressions. Thus, we introduce languages with lookahead as sets of pairs of strings with several operations and define the semantics of REwLA as languages with lookahead. Next, we define two kinds of left quotient for languages with lookahead and give corresponding derivatives. Then, we show that the types of expressions obtained by repeatedly applying derivatives are finite under some equivalence relation and give a conversion to deterministic finite automata. We also show that the semantics of REwLA is a finite union of sets of the form A × B, where A and B are regular languages.

• Keigo Imai, Jacques Garrigue
原稿種別: Regular Paper
専門分野: Special Section on Programming
2019 年 27 巻 p. 431-444
発行日: 2019年
公開日: 2019/06/15
ジャーナル フリー

This paper shows an encoding of linear types in OCaml and its applications. The encoding enables to write correct OCaml programs based on safe resource access guided by linear types. Linear types ensure that every variable is used exactly once, and, thus, they can be used to check the behavioural aspects of programs such as resource access and communication protocols in a static way. However, linear types require significant effort to be integrated into existing programming languages. Our encoding allows the vanilla OCaml typechecker to enforce linearity by using lenses and a parameterised monad. Parameterised monads are monads with a pre- and a post-condition, and we use them to track the creation and consumption of resources at the type level. Lenses, which point at parts of a data type, are used to refer to a resource in pre- and post-conditions. To handle comfortably structured data such as linearly typed lists, we further propose an extension to pattern matching based on the syntax-extension mechanism of OCaml. We show an application to static checking of communication protocols in OCaml.

Top