基于模糊小波神经网络的主机入侵预测

Host Intrusion Prediction Based on Fuzzy Wavelet Neural Network

综合利用模糊技术、神经网络与小波技术,提出一种主机入侵预测模型FWNN-IP。将系统调用按危险度进行分类,并为高危险度的系统调用赋予较高的值,利用模糊化后的系统调用短序列分析程序(进程)的踪迹,达到入侵预测的目的。实验结果表明,FWNN-IP模型能够及时预测程序(进程)中的异常,采取更加积极主动的预防措施抵制入侵行为。

This paper proposes a host intrusion prediction model named Fuzzy Wavelet Neural Network Intrusion Prediction（FWNN-IP） by using fuzzy methodology, neural network and wavelet technology. System calls are classified according to their dangerous degrees and higher dangerous system calls are assigned greater number. Programs（processes） traces are analyzed by applying fuzzed short sequences of system calls, and the aim of intrusion prediction can be achieved. Experimental results show that FWNN-IP can predict abnormal behaviors of programs（processes） more quickly, and takes more active action to protect host.