BP神经网络智能入侵检测系统设计

Design for Intelligent Intrusion Detection System Based on Back Propagation Neural Network

为了解决传统静态安全技术缺乏对入侵进行主动检测的机制,而且在使用过程中需要人工实施和维护,难以满足当前网络安全要求的问题;一种针对误差信号函数和学习规则进行改进的BP算法在分析标准BP算法存在的问题和其原因的基础上被提出;采用该改进算法构建了一种结合误用检测和异常检测技术的基于BP神经网络的智能入侵检测系统模型;仿真实验结果表明与标准BP算法相比,该改进算法具有学习过程快的优点,并且该系统具有较高的检测正确率并能检测出新的未知的攻击模式。

Aiming at solving the problems existing in the conventional network security techniques,which lack the active intrusion detection mechanism and need manual work to implement and maintain and which has been unable to meet the higher safety requirements of the present network,on the basis of analysis of the standard back propagation algorithm,a BP improved algorithm which adjusts the error signal function and method of training is proposed.By using of the improved algorithm,a detailed design scheme of intrusion detection model based on BP neural network which combines misuse detection technique with anomaly detection technique and which is made up of six modules is put forward.Experiment results showed that comparing with the standard BP algorithm,this improved algorithm has the advantages of faster learning progress and this intrusion detection system can recognize the unknown attacks and has good recognizing ability to the known attacks.