一种关于路网的连续最近邻查询算法.docVIP

一种基于路网的连续最近邻查询算法摘 要：在路网中，连续最近邻（Continuous Nearest Neighbor，CNN）查询在基于位置的服务中尤为关键。现有的查询处理方法大多依赖于路网中查询对象的分布密度，其他处理方法如UNICONS等改进了这些不足。然而在查询对象密集分布的路网中，存在无效计算最近邻（Nearest Neighbor，NN）的问题。针对这个问题，本文提出并证明了非交叉点子路径中的预计算方法，并基于该方法提出了CNN查询算法。该算法利用分治法以交叉点为划分依据，将查询路径划分成子路径，然后对子路径中的结点进行NN查询，从而降低了NN查询的计算代价。通过实验，验证了本文提出的查询处理方法在CNN查询中的正确性和有效性，性能优势尤为明显。关键词：连续最近邻查询; 基于位置的服务; UNICONS; 预计算方法; 分治算法Algorithm for CNN Query over Road NetworkWang Heng1,2 Ying-Yuan Xiao1,21Tianjin Key Laboratory of Intelligence Computing and Novel Software Technology, Tianjin University of Technology，300384, Tianjin2Key Laboratory of Computer Vision and System （Tianjin University of Technology）, Ministry of Education, 300384, TianjinAbstract In Road Network (RN), Continuous Nearest Neighbor (CNN) query frequently used in Location-Based Services (LBS). The majority of the existing works on CNN queries are largely affected by the density of objects of interest, others such as UNICONS overcome these problems, yet there may be over-calculating problem. In this paper, we propose and proof a pre-computation theory based on non-intersection path, and then presented an algorithm for CNN query, which uses divide and conquer method to query NN on sub path, where is divided the query path into sub path based on non-intersection points, and then to reduce the computational cost. Experimental results show that our processing approach in the CNN query is correct and effective, especially the result is well performance when the intersection points sparsely distributed.Keywords CNN; LBS; UNICONS; pre-computation; divide and conquer method1 引言随着移动计算、无线通讯、GIS（Geographic Information System）、以及GPS空间定位、空间网络数据库（Spatial Network Database，SNDB）等技术的迅速发展，基于位置的服务（Location-Based-Service, LBS）得到了广泛的应用。连续最近邻（Continuous Nearest Neighbor）查询[1]作为LBS中重要的查询类型，引起了国内外学者的关注。现有的查询处理方法大多采用预计算的方法，该方法首先对路网进行处理（如空间区域划分、裁剪等），然后预先计算路网中结点的NN并进行存储，最后处理查询路径上的CNN。与其他查询处理方法相比，该方法明显提高了查询的性能，并得到深入的研究。研究成果有：Feng等人[2]提出一种启发式的区域裁剪算法，通过该算法获得预计算的查询区域（r-region），然后计算查询路径