Historical index structure for reducing insertion and search cost in LBS

Abstract

A major issue in LBS (Location Based Service) is the handling of numerous historical moving object data, affecting query performance and service quality in application systems. In order to store and search lots of data rapidly, an effective index structure is required for improving not only the insertion method, but also the search performance. In order to improve the performance of both applications, we propose the GIP + (Group Insertion tree with Projection Plus) for historical data management such as the trajectory of a vehicle. This index structure, based on the GIP, employs the separated buffer node method for reducing overlaps. The GIP + also uses projection storage for improving search performance by grouping the intersected child node in a node. Additionally, the link between the buffer nodes is designed to directly connect to the next buffer node. To effectively combine these methods and improve the performance, different node levels in the GIP + are also arranged for applying the separated buffer node, the projection storage, and the link. The designed historical index structure is useful for inserting and searching data which is arranged on a time axis.