基于混沌神经元的延时滥用入侵检测模型

在研究混沌神经元延时特性的基础上,构建了MLP/CNN混合前馈型神经网络.提出基于混沌神经元的滥用入侵检测模型,它既具备MLP的分类功能,又具有混沌神经元的延时、收集和思维判断功能,具有灵活的延时分类特性,因而能够有效地识别分布式入侵.使用从网络数据流中获取的样本,以FTP口令穷举法入侵为例,对该模型进行仿真和整体测试,结果表明可以依据实际情况设置入侵判据,本文对FTP入侵检测的精确率在98%以上,误报率和漏报率均小于2%.该模型可以推广到检测分布式DOS等具有延时特性的攻击行为和具有延时分类要求的其它系统中.

A Time-Delayed Misuse Intrusion Detection Model Based on Chaotic Neuron

A hybrid MLP/CNN neural network is constructed based on research on time-delayed characteristic of chaotic neuron A misuse detection Model based on chaotic neuron is proposed,which has both the capability of classification which MLP has and the functionality of time-delay,collection and judgment which chaotic neuron has.Because this intrusion detection system has flexible time-delay characteristic,it can identify distributed intrusion effectively.The simulation and test is conducted using samples captured from data traffic.The detection rate of FTP (File Transfer Protocol) brute-force attack is up to 98%.The false alarm and false negative rates are both less than 2%.The model proposed in this paper may be generalized to time-delayed intrusion detection systems such as distributed DOS etc.and other time-delayed classification systems.