2012 THE UNIVERSITY OF AIZU ALL RIGHTS RESERVED
Principal Investigator: Ben A. Abderazek
Title: High-Radix Network-on-Chip Architecture for Future Many-Core Systems
Amount: Y=950,000
With the recent advances in semiconductor technologies, that has led to a huge increase in the number of cores on a single-chip, the on-chip interconnection network which connects the on-chip cores and memory together will become more critical in determining the overall systems performance and cost.
The goal of this project is to research and design a scalable NoC architecture based on multi-stage topology that can handle the projected tremendous growth in the number of modules in on-chip Many-core systems. In particular, this research will center on the following themes:
NoC topology design: An NoC topology is a network structure that determines where to put the IP cores and how to link these cores. In our NoC design, buffers are considered as separate cores, and they are made to be fully shared among all other cores. By optimizing the utilization of the shared buffers, the NoC architecture has the potential to need only small buffers to achieve a required network performance (e.g., energy consumption or small delay).
*Flow control:* We will also use network flow control schemes to reduce buffer sizes as well as packet delay in NoC. The flow control schemes aim to control the packet injection rate at traffic sources. The flow control will be based on an end-to-end scheme with some internal node assistance. It does not require internal nodes to store any per-flow information or to do any complex per-packet calculation.
*Energy Optimization*: We will study the impact on energy consumption for high-radix topologies. Then we will develop energy models of the topology and the corresponding architecture. Based on this, we will develop energy efficient design techniques.
Principal Investigator: Incheon Paik
Title: Situation Aware Apps on Smart Mobile Device based on Automatic Service Composition
Amount: Y=1,130,000
Awareness computing aims at our final goal in computer science to simulate human 's awareness and cognition. Technical approaches to model the awareness in human have contributed to the current awareness system together with semantic technology. In this research, as the current awareness system does not consider dynamic and active situation awareness for arbitrary context, streamlined and consistent active situation awareness (ASA) architecture with three layers (Perception, Comprehension, Projection) for situation awareness. Perception of object or data pattern should be well-geared to comprehension of a new situation based on reasoning of rules and ontology. How to getting data from the world at the lowest layer for perception is important for active situation awareness. ASA is to aware a new situation more actively. The existing and passive situation awareness waits for some change of data or information from the world. But ASA requests some data or information to the world to comprehend a new situation or project future in a new situation. Environment to get big data on Web such as social network service data and processing for information on the web freely are important, and we construct Web APIs for getting the information on the Web freely. Our architecture shows a new framework to provide active situation awareness on SNS services.
Principal Investigator: Junya Terazono
Title: Automated data uploading implementation in Web-GIS system
Amount: Y=690,000
Currently, our Cluster is developing Web-GIS based data browsing and collaboration environment, called WISE-CAPS (Web-based Integrated Secure Environment for Collaborative Analysis of Planetary Science). In this system, users can browse, share and (in future) analyze data with the operation in their web browsers, not by downloading locally. However, we have some drawbacks in this system. The current main issue is the lack of easy-to-use data handling system. Currently, we need several tedious step to register the data, including uploading using local application and editing configuration files. Particularly, repeating uploading files are one of major obstacles for general use of this system. Also, administrators needs unnecessary care for system because these process needs opening special ports and have some privileges for individual users (and sometimes, administrators face the many inquiry from users). To eliminate this inconveniency, we created automated data upload system via the web. These system is comprised of PHP, and therefore it is easy to maintain. The feature includes several uploading interface for files required to display at the WISE-CAPS. In this system, users can upload necessary files to display only by clicking corresponding uploading button(s) in the web page. Users do not need to prepare or install specific network copying application or command. Also, administrators do not need to open additional ports for data uploading. Moreover, this system has embedded editing system which enables users to edit configuration files. Users can edit their configuration files until settings in these files are correct for displaying. These functions contribute in increasing of users' convenience and wider usage of the WISE-CAPS system.
Principal Investigator: Igor Lubashevsky
Title: Continuous Markovian description of self-propelled Lévy particles
Amount: Y=1,060,000
Goal: The main attention of the project was aimed at constructing a continuous description of Lvy type random motion, namely, Lvy flights and Lvy random walks. On one hand, up to now it is a challenging problem in the modern theory of stochastic processes because such processes exhibit a number of anomalous properties. On the other hand, a wide variety of different systems from nonequilibrium physical media up to ecological and social systems are described using these notions.
Results: Among the obtained results the following should be noted.
Four publications were stimulated by the conducted research.
Principal Investigator: Paul A. Lyddon
Title: Promoting Second Language Development Online via Asynchronous Video Exchange
Amount: Y=740,000
To employ the OLE System, an open-source online video discussion board, in a semester-long video exchange between English learners at the University of Aizu and Japanese learners at Rose-Hulman and evaluate the role of this type of asynchronous interaction in oral language developments.
Principal Investigator: Shuxue Ding
Title: Nonnegative Matrix Factorization with Sparse Constraint
Amount: Y=880,000
In this research, we investigated Nonnegative Matrix Factorization (NMF) that can be applied for two important applications, 1) sparse representation, or sparse coding, of signal and; 2) blind source separation. In the application 1), we need to involve the constraint for sparseness in NMF. In the application 2), if the sources are sparse, again we need to involve the sparseness constraint in NMF. For these purposes, first, we have worked out a new, effective sparseness measurement, whose minimization will provide a sparse constraint explicitly. Second, we posed the NMF problem as an optimization under the condition that the sparseness measurement reaches its minimization. Third, we have developed an effective algorithm, NMF-SMC, for solving the posed problem. Fourth, we have applied the developed algorithm to blind spectral unmixing problem in remote sensing and image processing.
Principal Investigator: Stanislav Sedukhin
Title: Design and Evaluation of the Parallel Algorithms for the Graph Theory Problems
Amount: Y=820,000
A blocked algorithm for the all-pairs shortest paths (APSP) problem for a hybrid CPU-GPU system was designed and evaluated. In the blocked APSP algorithm, the amount of data communication between CPU (host) memory and GPU memory is minimized. When a problem size (the number of vertices in a graph) is large enough compared with a blocking factor, the blocked algorithm virtually requires CPU≅GPU exchanging of two block matrices for a block computation on the GPU.
We also estimate a required memory/communication bandwidth to utilize the GPU efficiently. On a system containing an Intel Westmere CPU (Core i7 970) and an AMD Cypress GPU (Radeon HD 5870), our implementation of the blocked APSP algorithm achieves the performance up to 1 TFlop/s in a single precision.
Principal Investigator: Tsuneo Tsukahara
Title: A Low-Power Complex Quadrature Modulator for A Multi-band Wireless Transceiver
Amount: Y=830,000
Recently the demand for wireless systems such as sensor networks has been rapidly growing. However, radio-wave resources are limited and invaluable especially in these days. Therefore, software-defied radios (SDRs) and cognitive radios, which is a principal application of SDR, can be the key to greatly improving frequencyspectrum efficiency. SDRs demand flexibility and reconfigurability in RF (Radio Frequency) circuits. Therefore, a spectrum-efficient wireless transceiver architecture is indispensable. In this project, we will propose a multi- band wireless transceiver using a low-power high-precision complex quadrature modulator (HP- CQMOD) and a spectrum-analyzing receiver suitable for sensor networks. As the final goal after 2-to-3 year research, we would like to establish a “reconfigurable wireless communicator,” whose frequency band can be changed according to communication conditions and/or a country’s regulations using reconfigurable RF and baseband processors and downloadable software. This is a kind of cognitive radios based on SDR (Software-Defined Radios). As the second step, in 2011, we have carried out design of low- voltage ( <1.8V) and low-power ( <15mW) circuits because a present quadrature modulator needs a higher-supply voltage over 2 V.
Decreasing power-supply voltage, it is becoming difficult to work three- transistorstacked mixers properly at low power-supply voltage. Therefore, we proposed two types of HP-CQMODs as the approach to lower power-supply voltage and power consumption; one is a folded-cascode HP- CQMOD using current mirrors, and the other is a passive HP-CQMOD with a newly-proposed calibration circuit. Features of these proposals are as follows:
Simulation parameters are based on the Rohm 0.18-um CMOS technology provided through the VDEC. The proposed quadrature modulators can provide the sideband rejection ratio (SRR) of more than 50dB with about 13 mW, thereby being applicable to QAM systems.
Principal Investigator: Victor I. Ryzhii
Title: Detector of modulated terahertz radiation based on double-carbon nanotube and double-graphene nanoribbons
Amount: Y=720,000
Principal Investigator: Vitaly Klyuev
Title: Sentence Alignment from Comparable Corpora
Amount: Y=470,000
Our goal was to create the database of semantically similar sentence pairs in English and Chinese from Wikipedia as multilingual corpora. This database may be a key component in the technologies of machine translation and cross-lingual retrieval. It may help make these technologies and their applications more intelligent.
To align the sentences, we applied the idea of stf-idtf. The key feature of this technique is to utilize the total number of word pairs divided by the length of the longer sentence. We set a threshold to decide if two sentences in different languages can be candidate sentences for alignment. A dynamic programming approach was utilized to find the best candidates for alignment (candidates with the maximum score). To evaluate the quality of selection, we proposed a similarity measure. The main results of our research are as follows: The high quality of Chinese-English sentence pairs were obtained from Wikipedia; undergraduate and graduate students were involved in the project; two papers were presented at the international scientific conferences, one of them was sponsored by IEEE.
Principal Investigator: Wenxi Chen
Title: Real Measurement and Theoretical Simulation for Deep Body Temperature Using a Human Abdominal Phantom
Amount: Y=940,000
This study is to build a physical phantom based upon the physiological and anatomical understanding, and to investigate the estimation performance in terms of accuracy and computational complexity by different deployment of cutaneous sensor array. Our final goal is to optimize sensor array deployment surrounding the abdominal trunk in order to achieve the highest performance in estimating the deep body temperature.
Principal Investigator: William Rozycki
Title: Determining the most effective style of presentation of research findings at international conferences Composition
Amount: Y=430,000
This research is part of a larger study comparing European and Asian styles of research presentation in the field of engineering. The research questions are 1) Are there differences in style between European research conferences in engineering and Asian research conferences in engineering? and 2) If these differences exist, which style affords the greatest comprehension for non-native speakers of English (NNSEs), and are there any preferences by NNSEs in terms of interest?
Findings at this point in the project show that average body orientation of presenters in two East Asian conferences was 131.1 degrees and average body orientation of one conference in Europe was 132.7 degrees. Research will continue with data gathered from an additional European research conference, but preliminary data shows that the body orientation does not significantly differ between the two venues. The ratio of slide text to spoken words will be compared as the next stage of the research, followed by simulated presentations before test subjects.
Principal Investigator: John Brine
Title: Grammatical highlighter for extensive reading on eReaders
Amount: Y=950,000
According to the literature on language acquisition, reading for meaning contributes to learning vocabulary and is likely to improve academic success. Research on second language acquisition, however, has shown that deficiencies in lexical knowledge where English language is the medium of instruction can cause difficulties for ESL students. Considering that EFL students may have even more difficulty, this project sought to support EFL students’ lexical and conceptual knowledge through creation of an eReader tool that facilitates student word recognition ability and fluent reading. Our system includes Wikification technologies to extend exploration of lexical and conceptual knowledge resident in context. We explored integration with other tools with an application engine. This system is intended to support extensive reading by providing glosses with intelligent hints and definitions that improve text readability. The eReader helps readers to remain focused on content rather than solely vocabulary meaning.
Principal Investigator: Junji Kitamichi
Title: Project on task scheduling of Data Center and multi-core processor for Power Consumption Reduction
Amount: Y=980,000
The 2011 Japan Crisis causes the lack of electric power. Then we researched and developed task scheduling algorithm of Data Center and multi-core processor for power consumption reduction. We developed and evaluated the proposed task scheduling algorithm. We discussed several heuristic algorithms for combinatorial optimization problem because the task scheduling problem is one of ombinatorial optimization problems and NP-Hard problems. We selected a heuristic algorithm which is suited for the task scheduling problem and refined it. And then, we evaluated the refined algorithm using some benchmark sets.
Principal Investigator: Anh T. Pham
Title: Design and Performance Evaluation of the Next-Generation Ultra-Wideband Optical-Wireless Home/Home-Access Network
Amount: Y=830,000
his research devotes to design and performance evaluation of the next-generation ultra-wideband home/home-access networks. The optical-wireless (OW) technology is proposed considering that it is a compelling solution (1) to overcome the RF spectrum scarcity and (2) to provide ultra-wideband wireless connectivity for home/home-access networks. The objective in this research focuses on novel modulation, multiplexing, coding and signal detection techniques to improve the performance of both home (indoor) and home-access (outdoor) OW links in various atmospheric conditions so that higher speed connectivity (up to Gb/s) for multiple users can be achieved.
Principal Investigator: Song Guo
Title: On the User Cooperation and Network Cooperation in Wireless Networks
Amount: Y=1,110,000
Cooperative communication offers an efficient and low-cost way to achieve spatial diversity by forming a virtual antenna array among single-antenna nodes that cooperatively share their antennas. It has been well recognized that the selection of relay nodes plays a critical role in the performance of cooperative communication. Most existing relay selection strategies focus on optimizing the outage probability or energy consumption.
To fill in the vacancy of research on throughput improvement via cooperative communication, we study the relay selection problem with the objective of optimizing the throughput for the following two cases. (1) For unicast, it is a P problem and an optimal relay selection algorithm is provided with a correctness proof. (2) For broadcast, we show the challenge of relay selection by proving it NP-hard. A greedy heuristic algorithm is proposed to effectively choose a set of relay nodes that maximize the broadcast throughput.
Principal Investigator: Xin Zhu
Title: Continuous systolic blood pressure monitoring during sleep
Amount: Y=690,000
Sleep occupies a large part of an individual’s life and is essential for human beings to recover from exhaustion and maintain health. However, many cardiovascular diseases, such as hypertension, atherosclerosis, stroke, and coronary artery disease, may influence the quality of sleep (QOS) and lead to other severe symptoms and even sudden death during sleep. Recently, it is discovered that sleep apnea has a strong relation with high blood pressure (http://www.interq.or.jp/kyuushu/sas/hypertension.htm). Therefore, monitoring blood pressure variations during sleep is an important for the diagnosis and prediction the risk of cardiovascular diseases. Last year, we have developed a sleep monitoring system to measure the physiological data during sleep. This year, we expect to develop the algorithm and analysis method for the estimation of systolic blood pressure during sleep.
Bachelor's thesis: Yusuke Konishi, Monitor of Systolic Blood Pressure using Pulse Arrival Time during Rest, Bachelor's thesis, 2012
Conference papers: Xin Zhu, Wenxi Chen, Kei-ichiro Kitamura, and Tetsu Nemoto, Comparison of pulse rate variability indices estimated from pressure signal and photoplethysmogram, IEEE-EMBS International Conference on Biomedical and Health Informatics, Hong Kong, Shenzhen, China, January, 2012 Xin Zhu, Wenxi Chen, Testu Nemoto, and Kei-ichiro Kitamura, Adaptive pulse template method for accurate detection of heart beat from pressure signals measured during sleep, IEEE iCAST 2011, Dalian, China, September, 2011
Principal Investigator: Yuichi Okuyama
Title: Development of Parallel Algorithm for Two-Dimentional Continuous Dynamic Programming
Amount: Y=830,000
Two-dimensional continuous dynamic programming (2DCDP) is known as a full pixel matching algorithm based on the DP matching. The 2DCDP can acquire full pixel correspondence between two images in very high accuracy compared with existing algorithms. Free view point TV (FTV) is one of the applications of 2DCDP. The FTV can construct a time-series 3D movie from only three similar view point movies. This 3D movie can be seen from various viewpoints. However, this 3D reconstruction process is calculated offline because the 2DCDP has high dependency and huge calculation costs. Current performance of 2DCDP is inadequate for real-time FTV processing although the calculation costs of 2DCDP should have been reduced to 1/4 1/10 in this research.
The accumulation process of 2DCDPcontains 4-dimentional loop. The loops for i and j are completely independent with other values. On the other hand, the loops for m and n are depended on anterior values of m and n. We apply the i-j loop for a computation inside of GPU. The m-n loop are deformed to the diagonal coordinate system and divided into two parts for implementation of dual GPU. We implemented the 2DCDP algorithm using CUDA environment provided by Nvidia. The i-j loops are replaced to kernel (function for GPU) call with 16x16 blocks. This implementation runs about 2.8 times faster than multi-cored implementation. We also developed dual GPU implementation, and this runs about 1.5 times faster than the single one. In addition, this implementation can deal with images which require double size of memory of single GPU implementation. Totally, this implementation obtain 4.1 times faster processing speed of than multi-cored CPU one.
Principal Investigator: Yuichi Yaguchi
Title: Incremental Spatial Classification for Growing Huge Database
Amount: Y=860,000
The objective of this project is to visualize time-changed large-scale networkstructured, for instance, transition of general or individual community topic on Twitter. Visualization or visual inspections are important analysis techniques for data mining.
General network mining method is not able to visualize and too difficult to largescale data, thus spatial classification technique is given the solution for visualizing and Mining large-scale network structured data.
Conventional technique of spatial classifications are able to apply only one time shot of data structure, thus, this project tries to applicable to time-changed leargescale network-structured data. The first result is verified to able to apply objetive. This research project is now continue by Kakenhi H-24-8.
Principal Investigator: Yukihide Kohira
Title: Implementation of General-Synchronous Circuits into FPGA
Amount: Y=740,000
In the recent LSI design, implementation into FPGA is widely used instead of implementation into ASIC because time to market becomes short and verification becomes easy. However, the circuit implemented into FPGA tends to be slower than that into ASIC. We investigated not complete-synchronous framework, in which a clock signal is inputted to each register simultaneously, but general-synchronous framework, in which it is not assumed, because the general-synchronous circuits are expected to work faster. In this project, we established the design flow for the implementation of general-synchronous circuits into FPGA. Experiments showed that the clock period obtained by our proposed design flow was improved by 11% on average.
Principal Investigator: Yutaka Watanobe
Title: Filmification Modeling Environment for Parallel Programming
Amount: Y=750,000
The purpose of the project is to develop a programming/modeling language *AIDA and the corresponding environment (Filmification modeling environment). AIDA is a language of explanation of application models and algorithms involved in corresponding computations. *AIDA compiler generates executable codes based on template programs for different types of platforms. In this project, language constructs, the run-time platform for the environment, and template programs have been enhanced and designed for parallel programming.
Principal Investigator: Emiko Kaneko
Title: The effects of working memory training on oral proficiency - Comparison between experimental and control groups
Amount: Y=390,000
In the field of second language acquisition, mechanical practices such as shadowing and repeating are somewhat outdated. In the context of English as a foreign language (EFL), however, many learners suffer from strong anxiety to speak English naturalistically, and such mechanical training might provide scaffolding to them. This kind of training is also expected to make the processing of working memory more efficient, leading to better L2 oral proficiency. In this ongoing research, non-communicative, or pseudo-speaking, practices were conducted with Japanese university students majoring in computer science, and its effects on their spontaneous speech were investigated in terms of complexity, accuracy and fluency. As a result of 11 short training held once a week, significant interactions were found between group (control vs. experimental) and time (pre- and post-tests) on fluency and complexity, suggesting that the pseudo-speaking practices have positive effects on spontaneous speaking.
Principal Investigator: Gennadiy Nikishkov
Title: Generation of finite element meshes from CT scans of composites
Amount: Y=840,000
Performing finite element analysis for unidirectional composites requires generation of finite element meshes with edges oriented along waved composite fibers. Main difficulty of mesh creation is related to weak visibility of fiber shades in the CT scan image. While in some areas of the image the fiber lines are completely invisible, the finite element mesh should cover the whole cross-section of the composite coupon.
In this research we developed a computational procedure for automated generation of 2D mesh of quadrilateral finite elements based on CT scan image of unidirectional composite material. Mesh of quadrilateral elements with the specified density is generated using image analysis with edge detection algorithm and integration of an ordinary differential equation for the fiber line slope. Image analysis and mesh generation software have been created for unidirectional composite coupons with fiber waviness.
Principal Investigator: Hiroyuki Sagawa
Title: Structure and Reactions of Unstable Nuclei
Amount: Y=780,000
We studied structure and reactions of unstable nuclei which were found in recent radioactive beam experiments. There are some new data of reaction cross sections which show a large odd-even staggering effect in Ne and Mg isotopes. We analyze these data by a Glauber model accommodating the large extension of the radial wave functions. Especially we focus on the role of pairing interactions on radii of unstable nuclei. We pointed out the importance of the pairing correlations to understand the observed odd-even staggering effect in the observed reaction cross sections.
Principal Investigator: XIrina I. Khmyrova
Title: Study of resonant sensor with electrical readout
Amount: Y=920,000
We developed analytical models of a resonant micro-electromechanical system (MEMS) with a high-electron mobility transistor (HEMT)-like structure and singleand an array of resonant cantilevers over its two-dimensional electron gas (2DEG) channel. The developed models allow to evaluate the impact of fringing electric field on output source{drain current for any number of cantilevers in the array at different spacing between them. It is demonstrated that the increase in the number of cantilevers results in increasing amplitude of the current peak as it was predicted. The models developed can be used for the optimization of the resonant MEMS in question.
Principal Investigator: Shinya Oku
Title: Optimization of Radiation Monitoring Algorithm for External Dose
Amount: Y=1,340,000
Establish a method to optimize the combination of three methods, which are scintillation counter (Geiger-Muller counter), pocket type radiometers and film badges, to estimate dose of radiation. Create an i-phone application to use public radiation concentration data and measured data of individual for the precise estimation of yearly exposure dose.
Principal Investigator: Tsutomu Hashida
Title: What makes “ autonomous learners ”?: An interdisciplinary approach to the investigation of motivation for lifelong learning and global career development
Amount: Y=560,000
This interdisciplinary project focused on one of the key concepts to learner autonomy: motivation to learn further. More specifically, we took up an indispensable skill to survive in an international environment, English, and investigated how motivation to learn English changed after study- abroad experience in order to pursue implications to cross-cultural training in our university. In the project, we assessed the impact of studying abroad on motivation for learning English and analyzed the extent to which their autonomy could be strengthened. We also investigated how their motivation related to the improvement in their English oral proficiency.
Principal Investigator: Ryuichi Oka
Title: 3D Motion Image Reconstruction for Clinical Inspection of Arbitrary Parts of Heart from 2D MRI Motion Images
Amount: Y=1,000,000
Human heart includes the motion of right ventricle which related with motion of tricuspid. The objective of the research is to establish an image processing technology to improve diagnosis of heart ventricle using 2D MRI motion image.
Right ventricle is a part of heart and supply the blood. Anomaly of right ventricle is popular disease (tricuspid atresia etc) and it is related with the motion of tricuspid. The relationship of them makes a difficulty for extracting right ventricle itself. Diagnosis of heart is realized by extracting it through the extraction of triscupid using only some slices of 2D MRI motion image. We succeeded to extract the motion of right ventricle as well as the motion of tricuspid.
Principal Investigator: Jie Huang
Title: Analysis and enhancement of HRTFs for the improvement of a 3-D sound system constructed by horizontally arranged loudspeakers the Aizu Region
Amount: Y=700,000
We have developed a 3-D sound system with wide sweet spot by 5 loudspeakers arranged on the horizontal plane at the level of listener’s ears as same as the existing 5.1 channel home theater systems. The system can create 3-D sound images with spatial properties including azimuth, elevation and distance for listeners seated near the center position of loudspeakers. Since the new 3-D sound system does not require any new or additional equipment for many audio users, it is expected to be easily commercialized. However, the elevation localization accuracy of the new 3-D sound system is still low due to the normalized HRTF (Head Related Transfer Function) data base. This research is to analyze how the HRTFs are related to elevation localization and to emphasize the characteristics related to elevation localization, so that the listeners can obtain stronger elevation cues than normal HRTFs.
Principal Investigator: Jungpil Shin
Title: A Study on Pen Computing
Amount: Y=770,000
The purpose of our project, "A Study on Pen Computing," is the development of Pen Computing algorithm based on the Handwriting with high performance. We have developed the algorithms ad systems for the following goal.
Principal Investigator: Keitaro Naruse
Title: Development of Swarm Robot System for Agriculture
Amount: Y=1,260,000
The goal of this research project is to develop a robotic weeding system for rice fields. We have done the followings in this year. (1) For the locomotion mechanism, we have developed two prototypes for each of the two locomotion mechanisms. They have been verified the requirements of the water-proof and the mud-proof and have shown good enough performance. (2) For the localization system, we have developed a swarm localization method with radio sensor nodes. We have developed sensor node prototypes and have made basic tests of them in an office environment.
Principal Investigator: Michael Cohen
Title: Cybervehicle
Amount: Y=1,280,000
This research is at the intersection of visual music, ubiquitous computing, and vehicular informatics. In this project, we focus on the display capabilities of cars, via windshield wipers and spatial sound. Inspired by the ubicomp (ubiquitous computing) vision in which all aspects of an environment have potential I/O capability, we explore the ability of windshield wipers deployed as output devices, synchronized to music, and the potential of spatial sound in cars for both passengers and drivers, especially for navigation and situation awareness.
Principal Investigator: Qiangfu Zhao
Title: Card user authentication based on linear sub-space representation
Amount: Y=1,020,000
Last year, in 2011, we proposed several methods for card user authentication, including methods based on linear fusion, morphing based fusion, and generalized morphing based steganography. The last method is a combination of image morphing and linear fusion, and it is more robust to “blind source separation ”attacks. Based on these results, the proposed authentication system can be adopted very easily without changing existing cards/terminals significantly.We also improved the proposed system in 2 directions. First, we proposed a method for generating more natural facial images based on image morphing and interactive genetic algorithm (IGA). Second, we proposed a new image morphing technique to morph and demorph image more quickly. For the former, we wrote a MatLab program for IGA, and showed that IGA can fine-tune the feature points and produce more natural facial images in a few generations, with a small population size. For the latter, we wrote a MatLab program, and proved through experiments that the new method is at least ten times faster than the morphing program we used so far. We have reduced the computing time (using the same PC) from about 50 seconds to about 3 seconds to obtain one morphed image.
Results related to this research
Principal Investigator: Saji N. Hameed
Title: Aizu Weather - a high quality environmental information system for Fukushima:
Amount: Y=1,340,000
Environmental information systems provide accurate and timely estimates of the state of our surrounding environment. Such timely information is useful not only for decision makers at various levels, but also for the common public in carrying out their day to day activities. In particular the Fukushima area would benefit from accurate and more detailed weather and climate forecasts than are currently provided by operational agencies. Further the dynamics of micro-climate in the region has received little attention in main stream research. The research work undertaken in this project provides the first steps towards realizing such benefits to our local society, while advancing basic micro-climate research.
In 2011, the project sought to improve the quality of local environmental information for Fukushima through detailed weather and environmental predictions. In particular detailed weather and radiation plume transport were successfully developed and deployed in real-time to demonstrate the potential of the technology. The system computed weather forecasts as well as forecasts of potential radiation plume transports twice a day and provided forecasts upto 2 days in advance. As part of the project, we also developed technologies that can be used to build autonomous low power and low cost weather sensor networks. A web-based system was developed that will allow for the collection of data from the sensors and visualize the information in real-time.
Principal Investigator: Toshiaki Miyazaki
Title: Die-hard sensor network II: toward fully WiFi-based system
Amount: Y=1,600,000
In the case of disaster, quick and detail surveillance of the damaged area is very important to make a rescue plan to save human lives. However, the detail surveillance is usually done by hands in actual rescue scenes because there are no effective aids. Our goal is to develop a wireless sensor network (WSN) system, so called ‘die-hard sensor network’, which can monitor the disaster-hit region automatically by only scattering many sensor nodes in the region. The WSN has the following advantages; (a) no individual setup action is required in the field, (b) whole WSN system is never down and continuously monitors the disaster-hit region even if some sensor nodes involved in the WSN die, and (c) it has so scalable that we can expand the monitoring area and improve the sensing quality by only deploying extra sensor nodes as many as needed. In addition to these features, this system has a good cost performance compared to other surveillance systems such as a remote-controlled unmanned helicopter. Thus, it can be applied to many critical disaster-hit regions, and we believe that our system must contribute not only to the rescue processes, but also to protecting the rescuers from secondary disasters.
In our previous project, a prototype of the die-hard sensor network system was developed using a wireless infrastructure based IEEE 802.15.4. However, through some experiences and interviews with rescuers, we recognized that it is much more useful to mash up information obtained from our system with that of other existing rescue systems to monitor the disaster-hit region effectively. To realize an easy interoperability with other systems, IEEE 802.15.4-based protocols should be changed to WiFi-based ones. In this project, besides keeping the abovementioned concept, a fully WiFi-based sensor network will be developed.
Principal Investigator: Subhash Bhalla
Title: Making Large Scale Database Systems Usable
Amount: Y=780,000
any researchers are making efforts to improve the capability of a database in terms of improvements in functionality. On the other hand, the usability of a database is crucial for users of application systems in Healthcare and GISs. Our study aims to improve the query language systems for databases. We consider the specialized domains of Electronic Health Records and the Geological maps for qquery systems. These domains provide an understanding of the query level requirements for the end-users. We investigate new proposals in this area, and aim to evaluate their usability for the end-users.
Principal Investigator: Yoshiko Ogawa
Title: Top science based on data analysis of Kaguya and Hayabusa missions
Amount: Y=820,000
The purpose of this project was to develop the effective tools for data processing based on IT-based technology, targeting the Japanese planetary missions. We developed the top science based on the analyses of data from Kaguya and Hayabusa. The key software to handle the mission data was ENVI/IDL. It was essential for us to develop and improve the algorithm of data processing using the universal software. By using ENVI/IDL, we succeeded in conducting a wide variety of scientific analyses and make new scientific findings. Our achievement includes the following subjects: (a) Geological classification of the lunar surface using multi-band images; (b) Extraction of geographical features from the high-resolution images and digital terrain models of the planets; and (c) Mineralogical and petrological analyses of the lunar and asteroidal reflectance spectra. As for (c) especially, we published a paper in Geophysical Research Letters (Ogawa et al., [2011]).
Principal Investigator: Naohito Nakasato
Title: Construction of a many-core computing system
Amount: Y=1,860,000
The inter-processor I/O is important research area related to parallel computers that was our focus in this project. There are a few practical network topologies used in existing parallel computers. For instance, many supercomputers and clusters adopt the fat-tree network. Another network topology frequently used is the torus topology. Since the torus topology network require local interconnect only, it is scalable in nature and power efficient in principle. To further advance the development related to the local interconnect architecture in our university, we have constructed a special computer cluster AIZU-CUBE with the local interconnect. This cluster consists of a node with GPU accelerator. We have tried to construct the cluster with a newly developed high-performance network cards and commercially available multi-port network cards.
Principal Investigator: Shigaku Tei
Title: New Algorithms/Protocols of Internet of Things and Their Applications to Smart Home/Town for Elderly People
Amount: Y=2,600,000
In the last year, three laboratories* joined together and challenge a new structure of IoT.
*Three laboratories project including Computer Networks Laboratory, Computer Communications Laboratory, and Performance Evaluation Laboratory in the University of Aizu.
Principal Investigator: Takafumi Hayashi
Title: Cloud Service Infrastructure
Amount: Y=2,350,000
In the project the following services are studied:
Sensor data mashup for environmental data (Temperature, humidity, atmospheric pressure, CO2 density, radiation dose.
The following our prototype system are constructed and evaluated.
The following key technologies are also studied in the project:
Principal Investigator: Hirohide Demura
Title: Initiative for Fukushima Data Center and Space Science Hub, UoA
Amount: Y=1,530,000
This Multi-lab Research Project was carried out by three laboratories; Multimedia Systems Laboratory (Profs. Asada, Demura, and Hirata), Database Systems Laboratory (Prof. Terazono), and CAIST/ARC-Space (Profs. Ogawa, Honda, and Kitazato). This is composed of two parts; Part 1: Fukushima Invitation of the Data Center for Deep Space Missions and Part 2: Developments for Hayabusa-2, FY2014 Launch, etc. Achievements of Part 1 are #1-1 Revision of Japanese Master Plan of Large Research Projects with Science Council of Japan, #1-2 Planning for renewal of JAXA infrastructures for data archive as remote-back-up system, #1-3 Arrangements for collaborations with NICT in Science Cloud Concept, #4 Preparations/Researches of Space Data Archives. That of Part 2 are #2-1 Development of Liquid Crystal Tunable Filters, #2-2 Development of UV-LEDs with Spectrometer provided by IANCU, Taiwan (FY2010 MOU of UoA-IANCU (MINERVA-2/Hayabusa-2)) as organics finder, and #2-3 Software License of AN- SYS as Thermal Analysis Tool for optical instruments.
2012 THE UNIVERSITY OF AIZU ALL RIGHTS RESERVED