An Access Method for Prediction of Moving Objects' Location

Currently, proposed access methods for prediction of moving objects' location pay more attention to the performance of window query but neglect cost of update operations. In this paper we perform an analysis to explain a defect which is not considered by many existing approaches and thus affects the performance of numerous update operations. Motivated by this, we propose two new index structures, called the ETPR-tree and BiR-tree, which both extend the TPR-tree to support object identifier query by different way. Using experimental comparison, we illustrate that the BiR-tree which uses B~+-tree for object ids as auxiliary index structure has the best window query and update performance under all conditions.