Efficiently Indexing Moving Objects on Road Network

With the development of GPS and wireless techniques, more and more moving objects are required to maintain their position on road network. Because the maximum objects throughput that can be tracked with a given accuracy is determined by the efficiency of the updates, it is the key topic that how to index these huge numbers of moving objects efficiently in this situation. The recent GTR-tree (group update & time parameter R-tree) not only buffered insertion and deletion entries and deal with them in group, but also introduced the velocity and the timestamp of moving objects to reduce the updates of objects which move at constant speed. It has high efficiency in I/O. However, the GTR-Tree holds low efficiency in CPU time because of obsolete table. This paper proposes a modified GTR-tree (call RGTR-tree for short), which could maintain high efficiency in CPU time by discard obsolete table. The new technique is empirically evaluated and compared with two other approaches and with the GTR-Tree, a competitor that is based on the R*-tree. Finally, experiment results show that both algorithms outperforms TPR*-tree in update and query performance of disk I/O, especially as most objects move at a constant speed.