摘要
提出了一种顽健的K近邻查询处理算法ROC-KNN,根据网络拓扑动态地将查询区域划分成若干子区域。每个子区域中选择一个簇头节点收集其他节点的感知数据,并将其发送至下一个子区域的簇头节点,直至遍历所有子区域。给出了2种分布式的启发式算法,用于设置子区域大小和选择簇头节点,以减少能量消耗。设计了一种利用子区域中非簇头节点恢复查询处理过程的算法,降低了查询处理因簇头节点失效而中断的概率。实验结果表明,ROC-KNN在能量消耗、查询成功率方面均优于现有的算法。
A robust K nearest neighbor query processing algorithm called ROC-KNN was proposed.It divides the query region into several sub-regions according the network topology.Each sub-region has a cluster node which collects the sensory data in it,sends it to the cluster node in the next sub-region until traversing all the sub-regions.Two distributed heuristic sub-region size setting and cluster node election algorithms were proposed to reduce the energy consumption.A query processing recovery algorithm using the non-cluster nodes in each sub-region was designed to reduce the outage probability caused by cluster node failures.The experimental results show that the ROC-KNN outperforms the existing algorithms in terms of energy consumption and query success rate.
出处
《通信学报》
EI
CSCD
北大核心
2010年第11期171-179,共9页
Journal on Communications
基金
国家自然科学基金资助项目(60673127)
国家高技术研究发展计划("863"计划)基金资助项目(2007AA01Z404)
江苏省支撑计划基金资助项目(BE2008135)
工信部电子信息产业发展基金资助项目
南京航空航天大学基本科研业务费专项科研基金资助项目(NS2010101)
江苏省普通高校研究生科研创新计划基金资助项目(CX10B_112Z)
南京航空航天大学博士学位论文创新与创优基金资助项目(BCXJ10-07)~~