Efficient access method for multi-dimensional complex objects in spatial databases: BR tree

Abstract

In this paper we propose a new dynamic access method for spatial databases, called BR(Binary divided Region) tree. Search, insertion, and deletion algorithms for the access method are discussed in detail. Then, the proposed access method is compared with the representative spatial access methods like the R-tree and R +-tree through experimental results. For search, when the size of a user-specified rectangular region, called query window, is small, the proposed access method produces the best performance with respect to the number of disk accesses since it generates the smallest number of nodes and avoids overlapping rectangles. However, when the size of a query window is large, the search performance of the proposed access method becomes poor since it increases empty space, and R-tree requires the smallest number of disk accesses. For insertion, the proposed access method achieves up to 40 percent saving in the number of disk accesses and up to 20 percent saving in the amount of storage over R-tree and R +-tree.