一种基于概率模型的运动趋势查询处理

运动趋势的准确预测是实现移动对象数据库中各种预测性时空查询处理的基础.提出了一种基于概率模型的运动趋势查询及处理方法.该方法将移动对象未来可能出现的位置定义为一种随机变量.运动趋势查询的处理就是检索随机变量的概率密度函数集合并进行概率计算的过程.为了获得较为准确的概率密度函数集合,提出一种通过对大量历史轨迹随机抽样来估计概率密度函数的方法.还设计了一种能够有效存储概率密度函数和提高运动趋势查询处理效率的索引结构.实验结果表明,提出的方法能够有效支持运动趋势查询的处理并提高对移动对象运动趋势预测的准确性.

A probabilistic approach for movement trend query processing

Accurate predicting the movement trend is the fundamental issue for processing many kinds of predictive queries in moving objects database.A probabilistic approach for processing queries about the movement trend of moving objects was proposed.The method treated the future position of a moving object as a random variable.The probability that an object moves into query range was computed according to a probably density function(PDF).In order to obtain the PDFs required for probability calculation,a trajectory analyzing algorithm which retrieved the estimated PDFs from historical trajectories was also presented.Finally,an index structure was designed to efficiently support the storing and accessing of the retrieved PDFs.The experiment evaluation shows that the proposed solution can effectively support the processing of probabilistic movement trend query and improve the correctness of predictions about the movement trends of moving objects.