基于HCBF的大流检测机制*

针对计数性布鲁姆过滤器存储数据时计数器溢出的缺陷，提出了一种基于分层计数型布鲁姆过滤器（hierarchy counting Bloom filter，HCBF)的大流检测机制。该方法结合溢出概率函数的特性，将计数型布鲁姆过滤器从一层扩展到多层，并能自适应地配置各层计数型布鲁姆过滤器的参数，能够对大流进行较好的识别。基于互联网数据进行了仿真实验，结果显示：与计数型布鲁姆过滤器相比，在同样溢出概率条件下，提高大流检测精度的同时节省了大量的内存资源。

Mechanism based on HCBF for large flow inspect

For the limited counter overflow probability in counting Bloom filter, this paper proposed a novel mechanism based on hierarchy counting Bloom filter(HCBF) for large flow inspect. By extending the standard structure of counting Bloom filter(CBF) to multi-layer with the overflow probability void feature, the mechanism could not only adjust the paramer configed, but also controled the false probability of large flow inspect. Experiments are conducted based on the data either randomly produced by computer or sampled from the real network trace. Results demonstrate that the proposed mechanism can achieve finer space saving and better accuracy with same overflow probability.