Temporal-Spatial Phase Point Moving Object Data Indexing: PM-Tree

With the development of mobile location technology and wireless communication technology, the application of moving objects has exhibited a broad application prospect. As moving objects' position varies as time goes on, the spatial data and temporal data generated continuously by moving objects has the characteristics of multi-dimension, complex data structure, massive data scale and complex data relationship. Usually the trajectory of moving objects was confined to a specific road network, so the index of moving objects based on the road network has become an important branch of the research of temporal spatial data index. At present, for the query optimization of the historical data of moving objects, the research focus on how to improve the efficiency of the window query. Usually such kind of indexes partitioned the trajectory data of moving objects by route, so the trajectory data of moving objects on a specific route was stored together. So this kind of indexes was a two-layer R-tree index structure, the upper layer was a 2D R-tree for indexing the routes in a region, and the lower one was also a 2D R-tree for indexing the moving objects in the ranges of routes in a certain period of time. In the view of these papers, the dimension of time in trajectory information was the same as the dimension of space. So dealing with the dimension of time, this kind of temporal spatial moving object index just extended the temporal dimension to R tree. However, because the algorithm of R tree cannot effectively reduce space overlapping of Minimal Bounding Rectangle (MBR), and it is more serious when the data volume is large and the dimension increases. In order to improve the efficiency of spatial-temporal trajectory information storage and query of moving objects in road network at some interval, this paper integrated temporal data and spatial data, and proposed a temporal-spatial phase point moving object data index (PM-Tree index). Firstly, this paper modeled the spatial trajectory of the moving object at some interval as a set of two-dimensional rectangles, and mapped it into a set of single-dimensional temporal and spatial phase points with parameters. Secondly, the paper discussed the partial order relationship among the temporal and spatial phase points on the phase plane. By partitioning the phase points with the partial order, a Phase-Point Order Branching was constructed. Then, based on Mon-tree index, the paper improved its lower layer of index structure by using the Phase-Point Order Branching structure, and proposed the spatial-temporal phase point moving object data index. This Index can realize the query optimization by the integration of spatial information and temporal information as spatial phase points, also it can avoid the low efficiency caused by MBR overlap in R tree and effectively reduce the search space. Finally, the paper realized the incremental dynamic update management of index. By comparing the performance of PM-tree index with that of Mon-tree index, the experimental results show that the PM-tree index can not only effectively improve the utilization of storage space, but also improve the query performance. © 2021, Science Press. All right reserved.