摘要
研究点对点网络入侵检测优化问题。点对点网络是一种多跳的、无中心的、自组织无线网络,其主机经常根据需要移动,主机的移动会使网络拓扑结构不断发生变化,而且变化的方式和速度都是不可预测的,这给网络入侵检测带来了困难。传统的检测方法针对网络拓扑结构稳定的网络效果很好,对于自组织的不可预测的点对点网络人侵检测准确性不高。为了提高检测能力和准确度,提出了改进ART2的入侵检测方法(SART)。当人工神经网络中所存储的模式量较大时,可对学习所得模式进行有效组织进而提高检测效率,通过调节幅度与相位的判断条件线性组合来缩小聚类之间的大小差异。仿真结果表明,相比其它检测算法,改进后的算法聚类的检测率较高,误检率较低,可满足误用检测及异常检测的需求。
The optimization of point to point network intrusion detection was researched. For improving the detec- tion ability and accuracy, an improved intrusion detection method of ART2 was proposed based on SART. When the pattern storage of artificial neural network was large, the mode after leaning were organized effectively, and the detec- tion efficiency was improved. The judging condition linear combination of amplitude and phase was adjusted for de- creasing the difference between the clusters. Simulation result shows that the improved algorithm has higher clustering detection performance than the traditional algorithm, and the false detection rate is low. It can meet the demands of misuse detection and anomaly detection.
出处
《计算机仿真》
CSCD
北大核心
2014年第1期311-314,共4页
Computer Simulation
关键词
点对点网络
入侵检测
自适应谐振算法
人工神经网络
Point to point network
Intrusion detection
ART2 algorithm
Artificial neural network
