摘要： 介绍流检测中常见的哈希算法，从信息熵的角度分析异或移位(XOR-SHIFT)算法。将异或运算扩展到字节，利用区域网络检测中数据包IP地址低字节比高字节变化频繁的特点，通过对称交叉异或运算，将五元组中的信息量尽可能更大化地表现在哈希值中，从而优化哈希算法的散列性能，使后续以流标识为约束进行的IP流处理更高效。

关键词: 哈希, 异或移位, 熵, 对称交叉异或

Abstract: This paper refers to several hash algorithms in IP flow detection, and analyzes XOR-SHIFT algorithm using the theory of entropy chiefly. XOR operation is extended to bytes, which can get much more random value from five-tuple because of difference between high byte and low byte of IP address in paceket. As a result, better performance comes into being in hash algorithm, which helps dealing with IP flows conveniently according to flow identifiers.

Key words: hash, XOR-SHIFT, entropy, Symmetry Cross XOR(SC-XOR)

中图分类号: 

• TP311