摘要： Deltoid算法是检测数据流变化的有效算法，但以现有方法为其设定检测阈值须额外增加工作空间和在线操作量。该文提出基于假设流量的阈值设定算法DTSUHT，无需增加工作空间，只需进行离线操作。基于真实网络数据的实验结果表明，用DTSUHT设定阈值，Deltoid算法可得到与现有阈值设定方法相似的性能。

关键词: 数据流, 假设流量, 显著变化

Abstract: Deltoid is an efficient algorithms for finding significant changes among data streams, but its threshold setting methods increase memory usage and online operations. This paper proposes a new threshold setting method namely Deltoid Threshold Setting Using Hypothetic Traffic (DTSUHT) to solve such problems. Experiments on real network traffic show that Deltoid algorithm with DTSUHT achieves almost the same performance of existing threshold setting methods.

Key words: data stream, hypothetic traffic, significant change

中图分类号: 

• TP301.6