ROAD: A New Spatial Object Search Framework for Road Networks

• Published in: IEEE transactions on knowledge and data engineering 2012-03, Vol.24 (3), p.547-560
• Main Authors: Lee, K. C. K., Wang-Chien Lee, Baihua Zheng, Yuan Tian
• Format: Article
• Language: English
• Subjects: and search algorithms; Artificial neural networks; Indexes; indexing techniques; Location-dependent spatial query; Search problems; Spatial databases; spatial road network
• ISSN: 1041-4347; 1558-2191
• DOI: 10.1109/TKDE.2010.243

In this paper, we present a new system framework called ROAD for spatial object search on road networks. ROAD is extensible to diverse object types and efficient for processing various location-dependent spatial queries (LDSQs), as it maintains objects separately from a given network and adopts an effective search space pruning technique. Based on our analysis on the two essential operations for LDSQ processing, namely, network traversal and object lookup, ROAD organizes a large road network as a hierarchy of interconnected regional subnetworks (called Rnets). Each Rnet is augmented with 1) shortcuts and 2) object abstracts to accelerate network traversals and provide quick object lookups, respectively. To manage those shortcuts and object abstracts, two cooperating indices, namely, Route Overlay and Association Directory are devised. In detail, we present 1) the Rnet hierarchy and several properties useful in constructing and maintaining the Rnet hierarchy, 2) the design and implementation of the ROAD framework, and 3) a suite of efficient search algorithms for single-source LDSQs and multisource LDSQs. We conduct a theoretical performance analysis and carry out a comprehensive empirical study to evaluate ROAD. The analysis and experiment results show the superiority of ROAD over the state-of-the-art approaches.