Quadtree and R-tree indexes in oracle spatial

Abstract

Spatial indexing has been one of the active focus areas in recent database research. Several variants of Quadtree and R-tree indexes have been proposed in database literature. In this paper, we first describe briefly our implementation of Quadtree and R-tree index structures and related optimizations in Oracle Spatial. We then examine the relative merits of t h e two structures as implemented in Oracle Spatial and compare their performance for different types of queries and other operations. Finally, we summarize our experiences with these different structures in indexing large GIS datasets in Oracle Spatial.

Efficient continuous skyline computation

“…In contrast, the split operation of the quadtree is much simpler, and merely divides the node in each dimension in half. For point data, such as that managed in skyline queries, the superior performance of the quadtree on inserts and updates has been noted in a study of a commercial DBMS [15]. This study of Oracle Spatial shows that quadtrees are significantly faster for index creation and updates of point data.…”

“…However, it has been shown to be an effective disk-based index [8,11] and some commercial objectrelational systems already support quadtrees [15]. The claim that we make and support is that if the speed of skyline computation is critical, a quadtree is far more preferable than an R * -tree.…”

“…The algorithm begins by checking if the top heap element is a point (line 5). If the point is dominated by the growing mini skyline, it is ignored; else, it is added to the mini skyline (lines [8][9][10][11][12][13][14][15]. If the upper right corner of any internal node is not dominated by P old , then it may be discarded (line 18).…”

“…The R * -tree is chosen because indexed skyline query algorithms discussed previously have focused exclusively on the R-tree family of indices. Quadtrees have been shown to manage point data more effectively than the R-tree family in several notable experimental studies [15,13]. Since skyline queries deal exclusively with point data, it is for this reason we have chosen the quadtree as the best alternative.…”

“…Since skyline queries deal exclusively with point data, it is for this reason we have chosen the quadtree as the best alternative. For a broader comparison beyond the scope of skyline queries for indices in the R-tree family and the quadtree, the interested reader may consult [15,13].…”