K-Order Neighbor: The Efficient Implementation Strategy for Restricting Cascaded Update in Realm

A realm is a planar graph over a finite resolution grid that has been proposed as a means of overcoming problems of numerical robustness and topological correctness in spatial database. One of the main problems of realm is cascaded update. Furthermore, cascaded update causes heavy storage and complex management of transaction. Virtual realm partially resolves the problem of space overhead by computing the portion of realm dynamically. K-order neighbor is a concept commonly used in Delaunary triangulation network. We use K-order neighbor in the Voronoi diagram of realm objects to restrict cascaded update. Two main update operations - point insertion and segment insertion are discussed. In point insertion, the distortion caused by cascaded update is restricted to 1-order neighbor of the point. In segment insertion, two end points of the segment are treated specially. This strategy can be used in both stored realm and virtual realm.