摘要： 分析RFID中间件查询数据的特点，提出一种对查询数据聚合转换的方法，减少查询索引的存储空间和数据插入时间。分析和比较已有多维查询索引的各方面性能，将多维索引KDB-tree应用到RFID中间件中。实验结果表明，KDB树索引在存储空间成本、数据插入成本和查询时间成本3个方面的综合性能最佳，在点查询上，KDB-tree只须单路径遍历索引树，数据查询时间少于其他方法。

关键词: RFID中间件, 聚合转换, KDB树, 事件过滤

Abstract: The characteristics of query data for RFID middleware are analyzed. Based on the characteristics, a method named aggregate transformation is proposed to save storage cost of index and reduce insertion time. The performance of several existing multidimensional indexes is analyzed and compared, KDB-tree index is supplied into RFID middleware. Theoretic analysis and experimental results demonstrate that KDB-tree index outperforms others in synthesized consideration of storage cost, insertion time cost and query time cost. In particular the query time cost of KDB-tree is distinctly lower than others because it provides single-path traverse in the query processing.

Key words: RFID middleware, aggregation transformation, KDB-tree, event filtering

中图分类号: 

• TP311