项目名称: 移动对象数据库中海量时空轨迹数据压缩方法研究
项目编号: No.61202064
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 计算机科学学科
项目作者: 刘奎恩
作者单位: 中国科学院软件研究所
项目金额: 25万元
中文摘要: 随着卫星定位技术、物联网定位技术和智能终端的普及,如保存与管理移动对象产生的海量时空轨迹数据,成为当前移动对象数据库研究中的核心关键问题之一。目前轨迹数据压缩方法研究集中在压缩单条轨迹数据,很少考虑移动对象运动模式的相似性带来的数据冗余,以及压缩后轨迹的索引与查询效率,难以适用于移动对象数据库环境。 本项目针对日益增长的轨迹数据海量性挑战,研究适用于移动对象数据库的轨迹压缩方法,首先提出一个多轨迹冗余度模型,并基于时空统计及计算几何方法快速聚类近似副本集合,为轨迹压缩提供冗余信息的表征与提取方法,然后进一步结合移动对象数据库环境,采用分层概略化方法,建立轨迹压缩存储及索引模型,支持压缩后轨迹的高效访问,接着面向轨迹压缩需求特征提炼度量指标,进行自适应调整,最后在理论方法研究的基础上,与移动对象数据库进行集成验证。
中文关键词: 移动对象数据库;时空轨迹压缩;轨迹相似性;误差受控;
英文摘要: Trajectory is everywhere.With the widely spreading of positioning techniques of satellites, wireless sensors and smart devices, trajectory data is increasing by blasting and challenges the serviceability of moving object database. Recent works of trajectory compression focus on single trajectory only, and they consider little on redundancy between multiple trajectories, which is the inherent attribute of trajectories because objects move similarly in nature. Furthermore, the efficiency of inquiries on compressed trajectories is also critical important in moving object databases. To tackle the growing massive volume of trajectory data, the aim of this project is to seek a novel method of trajectory compression applying for moving object database. Firstly, we propose a redundant model between multiple trajectories, employ statistical methods and computational geometry to cluster 'near-duplicates' from segments of trajectories, and abstract an item of 'near trajectory dictionary' from each cluster. Secondly, we introduce a hierarchical sketch model to represent an original trajectory, supporting with necessary data structures and indexing methods. Thirdly, to fulfill different requirements of trajectory compression, we adopt an adaptive framework on choosing and scheduling the compression algorithms. Finally, compr
英文关键词: Moving Object Databases;Trajectory Compression;Trajectory Similarity;Bounded Error;