Abstract: This paper simulates Distributed Denial of Service(DDoS) attack with NS2, and analyzes different samples of traffic based on wavelet. The relationship between sampling methods and estimation of Hurst parameter is presented. The analysis and experiments show that when selecting the traffic flow samples only in attacking period, the Hurst parameter will decrease, and when the sampling period includes non-attacking period, the Hurst parameter will increase.

Key words: Distributed Denial of Service(DDoS), Hurst parameter, wavelet, sampling methods

摘要： 采用NS2模拟分布式拒绝服务攻击，利用小波方法对受攻击的不同业务流采样样本进行分析，通过理论分析和模拟实验，证明采样方法与Hurst指数的估计关系密切：如果只选择攻击时段内的业务流样本，H值减小；当样本既包含正常流时段又包含攻击时段时，H值增大。

关键词: 分布式拒绝服务, Hurst指数, 小波, 采样方法

CLC Number: 

• TP393