Predictive Join Processing between Regions and Moving Objects

  • Authors:

  • Antonio Corral;Manuel Torres;Michael Vassilakopoulos;Yannis Manolopoulos

  • Affiliations:

  • Dept. of Languages and Computing, University of Almeria, Almeria, Spain 04120;Dept. of Languages and Computing, University of Almeria, Almeria, Spain 04120;Dept. of Informatics with Applications in Biomedicine, University of Central Greece, Papasiopoulou 2-4, 35100, Lamia, Greece, and, Dept. of Informatics, Alexander TEI of Thessaloniki, Greece 57400;Dept. of Informatics, Aristotle University, Thessaloniki, Greece 54124

  • Venue:
  • ADBIS '08 Proceedings of the 12th East European conference on Advances in Databases and Information Systems
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

The family of R-trees is suitable for indexing various kinds of multidimensional objects. TPR*-trees are R-tree based structures that have been proposed for indexing a moving object database, e.g. a data-base of moving boats. Region Quadtrees are suitable for indexing 2-dimensional regional data and their linear variant (Linear Region Quadtrees) is used in many Geographical Information Systems (GIS) for this purpose, e.g. for the representation of stormy, or sunny regions. Although, both are tree structures, the organization of data space, the types of spatial data stored and the search algorithms applied on them are different in R-trees and Region Quadtrees. In this paper, we examine a spatio-temporal problem that appears in many practical applications: processing of predictive joins between moving objects and regions (e.g. discovering the boats that will enter a storm), using these two families of data structures as storage and indexing mechanisms, and taking into account their similarities and differences. With a thorough experimental study, we show that the use of a synchronous Depth-First traversal order has the best performance balance (on average), taking into account the I/O activity and response time as performance measurements.