A query optimizer requires selectivity estimation of a query to choose the most efficient access plan. However, an effective method of selectivity estimation for the future locations of moving objects has not yet been proposed. In spatial databases, existing methods for spatial selectivity estimation do not accurately estimate the selectivity of a query to moving objects, because they do not consider the future locations of moving objects, which change continuously as time passes.
In this thesis, we propose an effective method for spatio-temporal selectivity estimation to solve this problem. We present analytical formulas which accurately calculate the selectivity of a spatio-temporal query as a function of spatio-temporal information. Our analytical formulas can be basically used by various cost models for moving objects.
Recently, the TPR-tree has been proposed to support spatio-temporal queries for moving objects. And, various methods using the TPR-tree have been intensively studied. However, although the TPR-tree is one of the most popular access methods in spatio-temporal databases, any cost model for the TPR-tree has not yet been proposed. As a result, a query optimizer may choose an inefficient plan for a spatio-temporal query. Existing cost models for the spatial index such as the R-tree do not accurately estimate the number of disk(or page) accesses for spatio-temporal queries using the TPR-tree, because they only handle the spatial locations of objects at the current time. In this thesis, we propose a cost model of the TPR-tree for moving objects for the first time.
To conduct experiments in a realistic environment, we generate synthetic moving objects by using real-life spatial data with a reasonable skew distribution. In the experiments, the proposed methods provide accurate estimation results over various queries with different spatial area sizes and time interval lengths.
To our knowledge, the proposed spatio-temporal selectivity estimation ...