基于SART的入侵检测仿真研究
发布时间：2018-11-28 17:13
【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。
[Abstract]:The optimization problem of point-to-point network intrusion detection is studied. A point-to-point network is a multi-hop, centreless, self-organized wireless network. Its hosts often move according to their needs. The mobility of the host makes the topology of the network constantly change, and the mode and speed of the changes are unpredictable. This brings difficulties to network intrusion detection. The traditional detection method is very effective for the network with stable network topology, and it is not accurate for the self-organized and unpredictable point-to-point network intrusion detection. In order to improve detection ability and accuracy, an improved ART2 intrusion detection method (SART).) is proposed. When the number of patterns stored in artificial neural networks is large, the learning patterns can be organized effectively and the detection efficiency can be improved, and the difference between clustering can be reduced by adjusting the linear combination of the judging conditions of amplitude and phase. Simulation results show that compared with other detection algorithms, the improved clustering algorithm has higher detection rate and lower false detection rate, which can meet the needs of misuse detection and anomaly detection.
【作者单位】： 河北大学附属医院计算机中心;
【分类号】：TP393.08
[Abstract]:The optimization problem of point-to-point network intrusion detection is studied. A point-to-point network is a multi-hop, centreless, self-organized wireless network. Its hosts often move according to their needs. The mobility of the host makes the topology of the network constantly change, and the mode and speed of the changes are unpredictable. This brings difficulties to network intrusion detection. The traditional detection method is very effective for the network with stable network topology, and it is not accurate for the self-organized and unpredictable point-to-point network intrusion detection. In order to improve detection ability and accuracy, an improved ART2 intrusion detection method (SART).) is proposed. When the number of patterns stored in artificial neural networks is large, the learning patterns can be organized effectively and the detection efficiency can be improved, and the difference between clustering can be reduced by adjusting the linear combination of the judging conditions of amplitude and phase. Simulation results show that compared with other detection algorithms, the improved clustering algorithm has higher detection rate and lower false detection rate, which can meet the needs of misuse detection and anomaly detection.
【作者单位】： 河北大学附属医院计算机中心;
【分类号】：TP393.08
【参考文献】
相关期刊论文 前2条
1 于尧;李U，
本文编号：2363653
本文链接：https://www.wllwen.com/guanlilunwen/ydhl/2363653.html
