SlidingWindow based Multi-Join Algorithms over Distributed Data Streams

Abstract

This paper focuses on multi-way sliding window join (SWJoin) processing over distributed data streams. A novel Join algorithm is proposed based on two distributed data stream transfer models. To reduce the communication cost and lighten the workload on the central processor node, the algorithm filters out tuples that can’t contribute to multiway SWJoin results by transforming the join conditions of SWJoin into filtering conditions during data stream transfer. Furthermore, the algorithm guarantees that all necessary data for generating exact multi-way SWJoin results can be transmitted to the central processor node.