Dimensionality Reduction Of Higher Dimensional Data In Spatio-Temporal Databases

Moving objects databases belong to spatio-temporal databases category, and moving objects are typically seen as moving in 3-dimensional Euclidean (X, Y, T) space. Actually, most of movement objects are restricted in 2-dimensional fixed network space, such as cars move in roads and trains in rail roads. With the development of the proliferation of the mobile position technology and the wireless communication technology, more and more application areas call for that database management system can have the ability to storage, query and provide any time position of a moving object, which move in 2-dimensional fixed network. Thus, building the spatio-temporal data model and index of 2-demensioal fixed network becomes import.After analyzing and introducing the typical spatio-temporal data model and index, in this paper, we aim at exploiting that movement occur in 2-dimensional fixed network, focus on how to reduce the dimensionality of the data using the space constraints, how to build a low dimension and high efficiency spatio-temporal index.Then individually giving a moving object database system based on transportation network, a 2-dimentional spatio-temporal data model 2DSTMOFN, and a 2-dimensional spatio-temporal index 2DSTFI.Firstly, we give the relative knowledge about spatio-temporal databases,it main introduces the classified spatio-temporal data model and data index. At the same time, we also introduce the trajectory of moving object in 2-dimensional fixed network.Secondly, we give out the design proposal of moving object database system, which based on transportation network. It introduces the architecture of the system, and the traffic network model of the system. The system provides the foundation for researching the data model and data index of moving object in 2- dimensional fixed network.Thirdly, we develop a new spatio-temporal data model, named 2-Dimensional Spatio-Temporal data model for Moving Objects in Fixed Network(2DSTMOFN). In the model, to reduce the dimensionality, trajectories of (x, y, t) are transformed to (loc, t), briefly the idea is using the linear referencing and the updating operations of motion vectors.Finally, on the bases of 2DSTMOFN, we propose a new spatio-temporal index, named 2-Dimentional Spatio-Temporal Index for Moving Objects in Fixed Network (2DSTFI). 2DSTFI is easy to implement and maintain, it consists of a 2-dimensional 2DR-tree for managing the fixed networks, a Hash structure for the newest location of moving objects, and a grid file for managing the past trajectory of moving objects. Extensive experiments are conducted to evaluate the performance of the proposed indexing mechanism and show that 2DSTFI performs considerably better than MON-tree.