An Incremental Approach to Closest Pair Queries in Spatial Networks Using Best-First Search

• Main Authors: CHEN, Chunan, SUN, Weiwei, ZHENG, Baihua, Mao, Dingding, LIU, Weimo
• Format: text
• Language: English
• Published: Institutional Knowledge at Singapore Management University 2011
• Subjects: Closest Pair; Spatial networks; Location-based services; Buffer management; Databases and Information Systems; Digital Communications and Networking
• Online Access: https://ink.library.smu.edu.sg/sis_research/1411; http://dx.doi.org10.1007/978-3-642-23091-2_13
• Institution: Singapore Management University

This paper addresses the problem of k Closest Pairs (kCP) query in spatial network databases. A Best-First search approach namely BFCP (Best-First Closest Pair) is proposed. Given two data sets of objects in a spatial network, BFCP first finds the 1st CP by computing the 1st NN (nearest neighbor) of each object in the set with smaller cardinality. Then BFCP retrieves the 2nd , 3rd , ... , kth CP in an incremental way by searching the next NN of the currently found CP's source point. Furthermore, a novel buffer replacement policy called MDU (Minimum Distance Unit) is proposed to reduce I/O cost of BFCP. Unlike LRU, which records only the last reference time, the MDU policy considers both temporal locality and spatial locality when selecting a buffer page as the victim. A comprehensive experimental study is conducted to demonstrate the advantage of BFCP and MDU.