The SB+ -tree: an efficient index structure for joining spatial relations

Abstract

This paper presents a spatial access method based on the B+ -tree structure. This new structure is called SB+ -tree, and should allow commercial databases an access method for spatial objects without major changes, since most commercial databases already support B+ -tree as an access method for text data. SB+ -tree is a hybrid of the existing spatial access methods. For each axis of the space, a set of indexing points is generated, where an indexing point is created whenever a new minimum bounding rectilinear rectangle (or MBR) begins or ends. The indexing points are then used to create an SB+ -tree. The number of SB+ -trees generated is dependent upon the number of dimensions of the approximation of the object, and not on the number of relations. We have developed an algorithm which uses SB+ -tree for performing spatial join between two spatial relations. Through simulation, we show that the performance of spatial join using SB+ -tree is more efficient than that of R+ -tree.