文档介绍：Page 1 第1页
Memory-Efficient Regular Expression Search Using 内存高效使用正则表达式搜索
State Merging 国家合并
Michela hi Michela hi
Department puter Science and Engineering 计算机科学与工程
Washington University, St Louis, MO 华盛顿大学,圣路易斯,密苏里州
******@ ******@
Srihari Cadambi Srihari Cadambi
NEC Laboratories America 美国NEC公司实验室
Princeton NJ 普林斯顿新泽西州
******@nec- ******@nec-
Abstract — Pattern matching is a crucial task in several critical 摘要-模式匹配是在几个关键的关键任务
network services such as intrusion detection and policy man- 网络服务,如入侵检测和政策的人
agement. agement。 As plexity of rule-sets increases, traditional 由于复杂的规则集增加,传统
string matching engines are being replaced by more sophisticated 正在取代由更复杂的字符串匹配引擎
regular expression engines. 正则表达式引擎。 To keep up with line rates, deal 为了保持线路速率,处理
with denial of service attacks and provide predictable resource 拒绝服务攻击,并提供可预测的资源
provisioning, the design of such engines must allow examining 配置,这种发动机的设计必须允许研究
payload traffic at several gigabits per second and provide worst 有效载荷流量在每秒数千兆,并提供最糟糕的
case speed guarantees. 的情况下速度保证。 While regular expression matching using 虽然正则表达式匹配使用
deterministic finite automata (DFA) is a well studied problem 确定性有限自动机(DFA)是一个很好的研究的问题
in theory, its implementation either in software or specialized 在理论上,它的实施,无论是在软件或专门
hardware plicated by prohibitive memory requirements. 望而却步内存的要求是复杂的硬件。
This is especially true for DFAs plex regular 这尤其是对于较复杂的正则的DFA
expressions present in practical rule-sets. 表达式目前在实际的规则集。
In this paper, we introduce a novel method to drastically 在本文中,我们介绍一种新的方法大幅
reduce the DFA memory requirement and still provide worst-case 减少DFA的内存要求,并仍然提供最坏的情况
speed guarantees. 速度保证。 Specifically, we merge several “non-equivalent”具体来说,我们合并了几个“不等价”
states in a DFA by introducing labels on their input and output 在DFA通过引入标签上的输入和输出
transitions. 转换。 We then propose a data structure to represent the 然后,我们提出了一个数据结构来表示
merged states and the transition labels. 合并后的国家和过渡标签。 We show that, with very 我们表