Efficient historical R-trees

Abstract

The historical R-tree (HR-tree) is a spatio-temporal access method aimed at the retrieval of window queries in the past. The concept behind the method is to keep an R-tree for each timestamp in history, but to allow consecutive trees to share branches when the underlying objects do not change. New branches are only created to accommodate updates from the previous timestamp. Although existing implementations of HR-trees process timestamp (window) queries very efficiently, they are hardly applicable in practice due to excessive space requirements and poor interval query performance. This paper addresses these problems by proposing the HR+-tree, which occupies a small fraction of the space required for the corresponding HR-tree (for typical conditions about 20%), while improving interval query performance several times. Our claims are supported by extensive experimental evaluation.