A Novel Framework for Processing Continuous Queries on Moving Objects
Traditional techniques for processing continuous queries on moving objects reduce query re-computing through single-threaded and shared execution between multiple queries, and dont make use of the parallel computing capabilities of the ubiquitous multi-core CPUs. Thus, to explore this kind of parallelism, a Multithreading based Framcwork for Continuous Queries (MFCQ) is proposed which adopts a strategy of re-computing all of the queries periodically. The framework divides the query process into three phases:the updating, optimization and ex ecution stages; multi-threading based methods are used in each phase. Moreover, the framework is deemed to be general, becausc it is compat ible with various index techniques and query algorithms. By using the framework, a query index based KNN algorithm and an object index based KNN algorithm are proposed respectively. Experimental results show that the multi-threading framework executed on the multi-core platform outperforms thc traditional YPK-CNN algorithm.
multi-threading framework continuous qucries KNN query grid index
Liang Zhao Ning Jing Luo Chen Zhinong Zhong
College of Electronic Scicnce and Engineering, National University of Defense Technology, Changsha 410073, China
国际会议
11th International Conference,WAIM 2010(第十一届网络时代管理国际会议)
九寨沟
英文
321-332
2010-07-14(万方平台首次上网日期,不代表论文的发表时间)