Research Interests of Yokota Laboratory

NOTE:: Here is the Japanese version of this page. We are doing our best to let you know what our laboratory research is all about. You can find here valuable information about our main research projects. If you would like to know much more, or if you feel like you can contribute better in enriching this page, please feel free to let us know, any comment is welcome. We are here to answer to any inquiry!

Research Outlines

Research relating to advanced features of magnetic disks: Usually, in order to access any magnetic disks, physical operations are required. As a result, magnetic disks cannot afford to improve in speed as rapidly as for the processor or the memory, and become the stage of potential bottlenecks for the computer system. Besides, when the durability of the stored data is strongly desired, increasing the reliance is indispensable because physical operations increases considerably the failure probability. Furthermore, since the storage capacity is rapidly extended by the improvement of magnetic disks, reduction of the management cost of data is also required. That is why, recently, the direct secondary memory storage connection to networks such as NAS (Network Attached Storage) or SAN (Storage Area Network) has generated considerable interest and triggered massive research efforts. Continuing in the same research direction of increasing the reliance of disks by introducing advanced features, we proposed failure resistance mechanisms, recovery mechanisms, load distribution mechanisms, and so forth on the disk side. Actually, while conducting this research, our laboratory was granted, in the dependable information system and secure LSI chip section of the Takeda research encouragement prizes, the most valuable research prize.
Research relating to the integration of multimedia data: This research examines how to build a framework which would allow, either from inside the university or from outside, to access useful multimedia data (research materials, lecture contents and so forth) which is organically unified (refer to these Japanese pages: specification research "information study, @A02 Research on production and practical use of contents).
Research relating to the efficient management of data to take advantage of the storage capacity increase:: Over the last couple of years, it has been widely noticed that first, the storage capacity treated by computer is extraordinarily increasing, second, the storage medium unit price per capacity is noticeably decreasing, and third, the multimedia contents are also considerably increasing. However the access performance to the stored data is still not enough. Moreover, even efficient data management is required in order to offer sufficient performance and reliability level, the cost of the administrator for it is increasing steadily. this research examines efficient data management techniques in order to enhance the performance and reliability of data stored at a very large scale, and this considering system configuration aspect, software based aspect, artificial intelligence aspect, coding theory aspect, and other various aspects.
Research relating to building a framework for workflow management: With the recent IT technology increasing development, the tendency of conducting business processes within and beyond the enterprise boundaries (Intranet/ Internet) is also carried out actively. However, considering the traditional Workflow management systems, in regard of the failure avoidance and recovery processing mechanisms, still a lot need to be done. So far, the concept of nested transactions between independent active database systems has been introduced in the workflow management system; it has been widely accepted that it can constitute a strong mechanism against failure occurrence and it can contribute notably in failure recovery.
Research relating to parallel architecture for data engineering: Our current research interests include the general area of computer architecture and data engineering. In fact, in order to make the present computing environment easier to use, considering the configuration of the computer systems themselves is essential. Specifically, it is necessary to put an emphasis, not only on the numerical computation capabilities, but on improving the throughput/expansion of data or knowledge which are stored in the nonvolatile storage. Here, we have to think about parallel distributed processing that includes, beside the processor, the secondary memory storage; methods, such as load balancing, data arrangement, optimization, exclusive control, recovery, and failure resistance are to be examined.
Research relating to the distributed directory structure:: In a parallel distribution environment, in order to manage efficiently the data, defining the structure of the distributed directory, which actually shows what kind of data is stored in which node, is important. In particular, when it is prerequisite to updating a database and the likes, it is necessary to offer a high speed access path while driving down the update cost. Furthermore, performing the load distribution of data with a distributed directory is more and more desired. Moreover, it is critical to have the capability of protecting data from unit failure in a large-scale storage. Our research is oriented towards defining a distributed directory structure which supplies similar capabilities..

Publications

Publications