摘要
网路入侵过程中入侵特征种类繁多,形成耦合性,很难形成较为规则的分布,传统的入侵检测方法都是假设网络入侵特征呈现独立高斯分布的,但是,一旦入侵特征耦合性较差,造成非高斯入侵数据拟合能力差,导致检测精度不理想。为了避免上述缺陷,提出了一种基于变异特征估计算法的非均匀分布入侵检测模型。在海量的网络操作数据中,提取出变异特征,根据提取的特征能够进行网络入侵检测。利用变异特征估计算法,能够建立合理的非均匀分布入侵检测模型,从而检测出网络入侵行为。实验结果表明,在非均匀分布的环境下,利用该算法对网络攻击行为进行检测,使非高斯数据具有更强的拟合能力,极大地降低了网络入侵检测的误报率和漏报率,提高了入侵检测的检测率。
Put forward a kind of of variation feature estimation algorithm based on non-uniform distributed intrusion de- tection model. In vast amounts of network operation data, to extract the variation characteristic, according to the features of extraction ability of network intrusion detection. Algorithm is estimated based on the variation characteristics, to estab- lish reasonable non-uniform distributed intrusion detection model, so as to detect the network intrusion behavior. Experi- mental results show that the nonuniform distribution of the environment, the algorithm presented in this paper to test the network attack behavior, make the non-gaussian data has better fitting ability, greatly reduces the network intrusion de- tection rate of false positives and non-response rates, increased rates of detection of intrusion detection.
出处
《科技通报》
北大核心
2013年第8期169-171,共3页
Bulletin of Science and Technology
基金
广东省教育厅
佛山市
中央电大
省电大科研项目立项
广东省电大远程教育开放基金项目(YJ1110)
关键词
入侵检测
非均匀分布
变异特征
高斯分布
intrusion detection
non-uniform distribution
variation characteristics
gaussian distribution
