Hong-Koo Kang

A grid-based node split algorithm for managing current location data of moving objects

Abstract There is rapidly increasing interest in Location Based Service (LBS) which utilizes location data of moving objects. To efficiently manage the huge amounts of location data in LBS, the GALIS (Gracefully Aging Location Information System) architecture, a cluster-based distributed computing architecture, is proposed. The GALIS using the non-uniform 2-level grid algorithm performs load balancing and indexing for nodes. However, the non-uniform 2-level grid algorithm has a problem creating unnecessary nodes when moving objects are crowded in a certain region. Therefore, a new node split algorithm, which is more efficient for various distribution of moving objects, is proposed in this paper. Because the algorithm proposed in this paper considers spatial distribution for the current location of moving objects, it can perform efficient load balancing without creating unnecessary nodes even when moving objects are congested in a certain region. Besides, the various data distribution configuration for moving objects has been experimented by implementing node split simulators and it’s been verified that the proposed algorithm can split nodes more efficiently than the existing algorithm.

STMPE: an efficient movement pattern extraction algorithm for spatio-temporal data mining

With the recent development of LBS(Location Based Service) and Telematics, the use of spatio-temporal data mining which extracts useful knowledge such as movement patterns of moving objects gets increasing. However, the existing movement pattern extraction methods including STPMine1 and STPMine2 create lots of candidate movement patterns when the minimum support is low. As a result of that, the performance of time and space is sharply increased as a weak point. Therefore, in this paper, we suggest the STMPE (Spatio-Temporal Movement Pattern Extraction) algorithm in order to efficiently extract movement patterns of moving objects from the large capacity of spatio-temporal data. The STMPE algorithm generalizes spatio-temporal data and minimizes the use of memory. Because it produces and maintains short-term movement patterns, the frequency of database scan can be minimized. Actually, the STMPE algorithm was improved twice to 10 times better than STPMine1 and STPMine2 from the result of performance evaluation.

UR-tree: an efficient index for uncertain data in ubiquitous sensor networks

With the rapid development of technologies related to Ubiquitous Sensor Network (USN), sensors are being utilized in various application areas. In general, a sensor has a low computing capacity and power and keeps sending data to the central server. In this environment, uncertain data can be stored in the central server due to delayed transmission or other reasons and make query processing produce wrong results. Thus, this paper examines how to process uncertain data in ubiquitous sensor networks and suggests an efficient index, called UR-tree, for uncertain data. The index reduces the cost of update by delaying update in uncertainty areas. In addition, it solves the problem of low accuracy in search resulting from update delay by delaying update only for specific update areas. Lastly, we analyze the performance of UR-tree and prove the superiority of its performance by comparing its performance with that of R-Tree and PTI using various datasets.

An efficient compression technique for a multi-dimensional index in main memory

Recently, in order to retrieve data objects efficiently according to spatial locations in the spatial main memory DBMS, various multi-dimensional index structures for the main memory have been proposed, which minimize failures in cache access by reducing the entry size. However, because the reduction of entry size requires compression based on the MBR (Minimum Bounding Rectangle) of the parent node or the removal of redundant MBR, the cost of MBR reconstruction increases and the efficiency of search is lowered in index update and search. Thus, to reduce the cost of MBR reconstruction, this paper proposed a RSMBR (Relative-Sized MBR) compression technique, which applies the base point of compression differently in case of broad distribution and narrow distribution. In case of broad distribution, compression is made based on the left-bottom point of the extended MBR of the parent node, and in case of narrow distribution, the whole MBR is divided into cells of the same size and compression is made based on the left-bottom point of each cell. In addition, MBR was compressed using a relative coordinate and the MBR size to reduce the cost of search in index search. Lastly, we evaluated the performance of the proposed RSMBR compression technique using real data, and proved its superiority.

TMOM: a moving object main memory-based DBMS for telematics services

Recently with the growth of the Internet and the activation of wireless communication, telematics services are emerging as promising next-generation business in the IT area. In order to provide telematics services, technologies in various areas are required but particularly DBMS technology for efficient data processing and management is an essential key technology for all types of telematics services. This paper designed and implemented TMOM (Telematics Moving Object Main Memory DBMS) to meet efficiently the requirements that telematics services should be provided in real-time. TMOM follows the spatial data model of OpenGIS Simple Features Specification for SQL, and provides spatial, spatio-temporal and temporal types and corresponding operators for processing the moving object data. In addition, it supports optimized spatial, spatio-temporal and trajectory indexes in the main memory for fast search of large moving object data and provides a recovery function that can minimize disk input-output to maximize system performance. Also, for high transmission efficiency in data import/export between telematics applications and a back-end DBMS, it supports a compression function and a data caching function suitable for the characteristics of spatial data.

An Extended R-Tree Indexing Method Using Selective Prefetching in Main Memory

Recently, researches have been performed on a general method that can improve the cache performance of the R-Tree in the main memory to reduce the size of an entry so that a node can store more entries. However, this method generally requires additional processes to reduce information of entries. In addition, the cache miss always occurs on moving between a parent node and a child node. To solve these problems, this paper proposes the SPR-Tree (Selective Prefetching R-Tree), which is an extended R-Tree indexing method using selective prefetching according to node size in the main memory. The SPR-Tree can produce wider nodes to optimize prefetching without additional modifications on the R-Tree. Moreover, the SPR-Tree can reduce the cache miss that can occur in the R-Tree. In our simulation, the search, insert, and delete performance of the SPR-Tree improved up to 40%, 10%, 30% respectively, compared with the R-Tree.

NUNS: A Nonuniform Network Split Method for Data-Centric Storage Sensor Networks

If data have the same value frequently in a data-centric storage sensor network, then the load is concentrated on a specific sensor node and the node consumes energy rapidly. In addition, if the sensor network is expanded, the routing distance to the target sensor node becomes longer in data storing and query processing, and this increases the communication cost of the sensor network. This paper proposes a nonuniform network split(NUNS) method that distributes the load among sensor nodes in data-centric storage sensor networks and efficiently reduces the communication cost of expanding sensor networks. NUNS splits a sensor network into partitions of nonuniform sizes in a way of minimizing the difference in the number of sensor nodes and in the size of partitions, and it stores data occurring in each partition in the sensor nodes of the partition. In addition, NUNS splits each partition into zones of nonuniform sizes as many as the number of sensor nodes in the partition in a way of minimizing the difference in the size of the split zones and assigns each zone to the processing area of each sensor node. Finally, we performed various performance evaluations and proved the superiority of NUNS to existing methods.

Development of an Efficient Conversion System for GML Documents

In this paper, we designed and implemented a Conversion System for GML documents, and evaluated the performance of the system. The conversion system that uses the BXML-based binary encoding specification can directly display geographic information from BXML documents and can convert GML documents to BXML documents without loss of information, and vice versa. BXML is generally more effective in scanning cost and space requirement than GML and coordinate values in the form of BXML can be used directly without conversion.