Indexing Moving Objects Using Query Oriented Parallel Grids

Abstract

To1 improve updating efficiency and querying accuracy for spatial moving object data, we propose a parallel data structure for indexing moving objects. The structure contains a main index and an auxiliary index, which are used for supporting range based and identity based spatial object query operations, respectively. It also utilizes a query index which hooks updating operations to querying operations that may be influenced. Thus, it avoids locking relevant spatial objects and indexing structures as existing approaches do when range query operations are processed. At the same time, it also supports time slice semantics for parallel operations. Experimental results show that, under high working load, the structure can not only guarantee querying accuracy, the throughput is also obviously higher than that of the existing methods. The index improves the degree of system parallelism, makes it possible for object updating and querying operations in same ranges be processed in parallel, and therefore improves the overall efficiency of the system.