摘要： 针对数据源节点通信资源十分有限的缺陷，提出一种基于直方图的多数据流滑动窗口连接查询的降载策略。该算法综合中心处理节点和数据源节点的负载情况，给出降载比例计算公式，通过使用聚类技术构建中心直方图和数据源直方图，给出降载策略。结果证明该算法能够产生精确连接结果的极大子集，对多数据流窗口连接降载是高效的。

关键词: 数据流, 降载, 窗口, 连接算法

Abstract:

Considering communications resource of remote nodes is limited, this paper presents a histogram-based load shedding for window multi-joins over data streams. The formula of load shedding rate is given based on the load of central processing and data source nodes. A central histogram and source histograms are designed by using clustering technique, and load shedding is proposed based on these histograms. The results show that the algorithm can produce maximum subset of exact join result and it is efficient for shedding the load of streaming window multi-joins.

Key words: data stream, load shedding, window, join algorithm

中图分类号: 

• TN919.2