그리드 데이터베이스에서 질의 처리 최적화 기법

서윤석 김천과학대학 2008 김천과학대학 논문집 Vol.34 No.-

In this pager, a method of query processing optimization in the grid database is proposed. To decrease network transmission cost, frequently used data is cached from meta database. Existing cache management method has a unbalanced resource problem, because it doesn't manage replicated data in each node. Also, it increases network cost by cache misses. In the case of data modification, if cache is not updated, queries can be transferred to wrong nodes and it can be occurred others nodes which have same cache. Therefore, it is necessary to solve the problems of existing method that are using unbalanced resource of replica and increasing network cost by cache misses. In this paper, cache management method for query forwarding optimization is proposed. The proposed method manages caches through cache manager. To optimize query forwarding, the cache manager makes caching data from lower loaded replicated node.

공간 데이터 웨어하우스에서 효율적인 색인 재구성 기법

서윤석 김천과학대학 2006 김천과학대학 논문집 Vol.32 No.-

A SDW(Spatial Data Warehouse) is a subject oriented, integrated, time variant, non volatile collection of geographical information data for decision support. A SDW server organizes index using stored data for short response time. Existing methods that organize index are bulk insertion and index transfer methods. Bulk insertion method has high clustering cost for organizing index and searching cost. Also, index transfer method reduces the index organization time. However, it is improper for the index reorganization method of a SDW where periodic source data are inserted. In this thesis, the efficient IRUPKIndex Reorganization Using Partial Index) method in a SDW is proposed. This method is an efficient reorganization method that transfers partial index and stores partial index with expecting physical location. This method clusters spatial data making it suitable to organized index and changes created clusters to partial index and transfers pages that store partial index. A SDW server reserves sequent physical space of disk and stores partial index in the reserved space. Through inserting partial index into organized index in a SDW server, searching, splitting, remodifying costs are reduced to the minimum.

u-GIS 환경에서 지오센서 데이터 스트림의 효과적인 처리를 위한 GS-DSMS의 설계

서윤석 김천과학대학 2011 김천과학대학 논문집 Vol.37 No.-

In u-GIS environments, GeoSensor generates sensor data streams including spatial information as well as various conventional sensor data from RFID, WSN, Web CAM, Digital Camera, CCTV, and Telematics units. This GeoSensor enables the revitalization of various ubiquitous USN technologies and services on geographic information. In order to service the u-GIS applications based on GeoSensors, it is indispensable to efficiently process sensor data streams from GeoSensors of a wide area. In this paper, we propose a GeoSensor data stream processing system for u-GIS computing over real-time stream data from GeoSensors with geographic information. The proposed system provides efficient gathering, storing, and continuous query processing of GeoSensor data stream, and also makes it possible to develop diverse u-GIS applications meet each user requirements effectively.

실시간 공간 데이터베이스 시스템을 위한 질의 최적화 기법

서윤석 김천과학대학 2003 김천과학대학 논문집 Vol.29 No.-

Real-time spatial database application systems require that spatial data have to be updated within valid time and user queries should be processed in real-time. In these systems, it needs to be considered a processing about real-time spatial data and real-time spatial query they are not considered in a conventional spatial database application systems. Real-time spatial queries in real-time spatial database application systems are hard to be processed successfully, because spatial operation takes major time in total processing and its real-time spatial data are easy to miss their validity. Therefore, it is required a query processing technique that makes queries meet time constraints of real-time spatial query with keeping validity of real-time spatial data. real-time operation post-placing technique is proposed in order to meet time constraints of real-time data and queries in real-time spatial query processing with managing real-tome spatial data and considering that manages real-time spatial data used in real-time spatial query.

공간 객체의 효율적 중복영역 제거기법

서윤석 김천과학대학 2002 김천과학대학 논문집 Vol.28 No.-

클라이언트-서버 환경의 공간 데이터베이스 시스템은 방대한 양의 공간 데이터 전송에 따르는 네트워크 부하가 발생한다. 이러한 환경에서 화면 이동, 확대 및 축소 등의 유사한 영역의 데이터를 요구하는 질의는 매우 빈번하게 발생하며, 이미 전송된 영역과 중복되는 영역의 데이터를 재전송하여 네트워크부하를 가중시킨다. 이러한 문제는 생성된 질의 결과 중 클라이언트에 이미 전송된 데이터들을 제거함으로써 해결이 가능하다. 본 논문에서는 클라이언트-서버 공간데이터베이스 시스템에서 영역 요구 질의의 효율적인 처리 방법을 제안하고, 이를 기반으로 동일한 데이터의 중복 전송을 제거하기 위한 공간객체관리자를 설계하고 구현한다. Spatial Database systems in client-server environment have network overhead caused by transmission of a large amount of spatial data. In this environment, queries such as screen movement, enlargement, or shrinkage which request data in similar region are frequently occurred, and they increase network overhead by re-transmission the same data in duplicate region with earlier transmitted region. Removing transmitted data from query results can solve this problem. In this paper, we propose an efficient method for region request query in client-server spatial database systems, and design and implement a spatial object manager for removing re-transmission of the same data.

캐싱기법을 이용한 SQT의 설계

서윤석 김천과학대학 2007 김천과학대학 논문집 Vol.33 No.-

S-XML(Spatial-eXtensible Markup Language) is expanded form of XML which supports heterogeneous spatial data for both integrating and circulating. The S-XML support system is based on Spatial Database Management System(SDBMS) which stores, manages and provides S-XML to the users. For this purpose, the transformation process between S-XML data and binary data is required. Especially the spatial data has complex data type and large data size than aspatial data, so it demands heavy transformation cost. Therefore, a method to reduce the transforming cost of spatial data in S-XML is required. In this thesis, the SQT(Spatial Query transformer) using caching method is proposed. The proposed method caches only spatial data in S-XML. This S-XML is the result of transforming binary data which is processed by SDBMS. At first, in the process of transformation, S-XML transformation component uses OID within the results data, which in turn is processed by SDBMS, When processing query, types are separated by analysis. In the case of range query, it uses the R-tree index and for the other cases, hash index is used. Depending on the search results, if the cached data is available, replacement to S-XML is possible without transformation costs. But, there is a consistency problem with cache data and database of SDBMS in cases of insert, delete and update query processing in SDBMS. For this purpose, consistency is maintained by using OID information which is sent through the update processed by SDBMS. The proposed SQT using caching method is reduced the transforming cost.

공간 데이터베이스를 지원하기 위한 무결성 시스템의 설계

서윤석 김천과학대학 1999 김천과학대학 논문집 Vol.25 No.-

This paper describes the design of an extended spatial query processing system which provides integrity maintenance in the spatial database system that manages the correlation between spatial data and aspatial data. Geographic information system that models geographic and spatial information in the real world is implemented based on spatial database system. Spatial database system should handle data not as simple object but as abstracted complex object and maintain integrity for various derived data in the database system efficiently and represent the conceptual characteristic of spatial data as the format of the constraint. In this paper, we propose the spatial constraint language that can represent integrity information for spatial data as the format of the predicate logic in order to extend spatial query processor. The proposed spatial constraint language is stored as the format of reduced query trec in the meta database and is used as the information to guarantee the consistency of spatial database for user's query.

공간 데이터에서 시공간 데이터의 효율적인 집계 연산

서윤석 김천과학대학 2005 김천과학대학 논문집 Vol.31 No.-

Recently, the data that is simply summarized from spatial data warehouses is required. Aggregate operation can be obtained from spatial data warehouses directly whenever it is needed, but it costs lots of time. Thus pre-aggregation is required for the efficient OLAP(On-Line Analytical Processing) operation. There are two kinds of spatiotemporal data: current and past data. Previous studies, which manage the pre-aggregate information from spatial data warehouses, provide only one type of pre-aggregate information, so previous studies can not provide real time responses when the current and past data are queried frequently. Also, the method that gives only past aggregate information can not supply the OLAP operation by time dimension because it can not maintain the historical information. In this thesis, method that can manage the pre-aggregate information of current and past spatiotemporal data efficiently is proposed. The proposed method maintains current aggregate information by aR-tree and keeps historically summarized aggregate information by HISAT(Historically Summarized Aggregate Table) at the same time.

공간 데이터베이스에서 확장된 R-Tree를 이용한 효율적인 공간 질의 처리

서윤석 김천과학대학 2004 김천과학대학 논문집 Vol.30 No.-

R-Tree provides indirect access path to spatial object by using MBR(Minimum Boundary Rectangle). Various kind of object approximation method has been proposed, but most of them has low filtering performance. MBR decomposition methods has been proposed to upgrade filtering performance but it has disadvantage when object is very complex because it require large size of additional space to store decomposed MBR. Another solution to solve low performance of filter step in R-Tree is object decomposition method. It has good filtering performance but when information on original object is needed, object has to be reconstructed. In this thesis, we proposed extended R-Tree which has grid for secondary filter. Girds consist of cells which store information on original object. Each cells represents whether any part of original object is existed in cell boundary or not. Proposed method requires relatively small amount of space, and it provides easy way of inspection.

그리드 데이터베이스에서 동적 계층 구조를 이용한 부하 분산 기법

서윤석 김천과학대학 2009 김천과학대학 논문집 Vol.35 No.-

This paper suggests a reallocation method based on dynamic hierarchical structure to solve the problems with reallocation in the proximity The method suggested in this study dynamically creates reallocation manager that collects information of each database at the time of reallocation in a logical way. In addition, by forming reallocation manager on the basis of hierarchical structure, it supports stepwise dispatch of information and reallocation algorithm, which, in turn, enables support of local reallocation algorithm and efficient balance adjustment. In this method, quicker processing is possible by distributing the burden of the uppermost reallocation manager to collect and analyze all the information to lower level reallocation manager using summarized information at the time of performing reallocation algorithm. This method is found to contribute to the increase in total processing volume after reallocation via performance evaluation.