基于 LRU-BF 策略的网络流量测量算法

摘要

Aiming at the naïve algorithm’s deficiency of high false negative probability, a novel scheme called LRU-BF(least recent used & Bloom filter) was presented. In order to achieve high accuracy, the algorithm adopted mechanisms of LRU eliminating and Bloom filter representation to separate the process of heavy-hitter fliteration from the heavy-hitter recognition. Based on statistical theory, analytical expressions about upper-bound error probability were deduced. Simulated results indicate that LRU-BF can achieve space saving and lower error probability compared with Naïve-LRU algorithm. Meanwhile, it can also support the 10Gbit/s line-speed processing.%　　针对传统大流检测算法漏检率高的缺陷,提出了一种基于 LRU-BF(least recent used & Bloom filter)策略的流量测量算法。该算法使用 LRU 淘汰机制、Bloom filter 快速表示方案,将“大流过滤”和“大流判断”分离,较大地提高了测量的准确性。基于“概率论”的相关知识,对算法进行了理论分析,建立了错误概率上界的解析表达式。仿真结果表明：与传统 Naïve-LRU 算法相比,LRU-BF 具有较低错误概率和空间复杂度的同时,也能满足单线路10Gbit/s 的线速报文处理能力。