【摘要】 提出一种基于离散小波变换和分形原理的DDoS攻击检测方法。该方法通过离散小波变换的多分辩率分析突现DDoS攻击特征,对小波变换系数进行盒分形维计算,将经实验确定的关键盒维数作为多维空间的向量序列,最后使用经过样本训练的K-nn(K最近邻)分类器进行攻击识别。实验结果表明分形与小波相结合取得了较好的检测效果,与离散小波检测方法相比,该方法提高了检测精确度。更多还原

【Abstract】 Based on the discrete wavelet transformation and fractal theory, this paper presents a DDoS detecting algorithm. This method has the character of DDoS attack appear clear by multiresolution analysis of wavelet transformation, and undertake box dimensions calculation for the transformation coefficient of discrete wavelet. the key box dimensions are confirmed by the experiment and are regarded as vector sequence of multidimensions space. Then the DDoS are recognized by applying the K-nearest neighbor classifier trained by the samples. The results of experiment show that the presented algorithm of combination of wavelet transformation and fractal theory can detect the DDoS attacks effectively. Comparing with the detecting method of discrete wavelet transformation, this method has improved the accuracy of detecting.更多还原