基于IM-GA-BPNN的网络入侵检测算法设计

Algorithm of Intrusion Detection Based on IM-GA-BPNN

为了提高网络入侵检测系统的检测率、实时性和误报率,实现对网络进行有效的入侵检测,设计了一种基于免疫遗传算法和BP神经网络的网络入侵检测方法,首先建立四层的网络模型,采用训练数据对BP神经网络进行训练;为了进一步优化参数,通过免疫遗传算法对神经网络的参数进行优化,通过个体的复制、选择、交叉和变异来提高解的多样性,实现最优参数的求解。将KDD99 CUP入侵检测数据库中的数据作为仿真数据实验,将所提的模型IM-GA-BPNN与其它方法如BPNN、PCA-NN和PCA-PSO-NN进行比较,结果表明所提模型具有最高的检测率,同时具有检测效率高的优点。

In order to improve the detection rate and real-time performance, a network intrusion detection algorithm based on immune genetic algorithm is designed. Firstly, a network model with four layers is established, where the training data is used to train the BP neural network. To optimize the parameters, the immune genetic algorithm is used to train the neural network again. In the immune genetic algorithm, the personality is copied, selected, crossed and varied to increase the multiple kinds for the solutions, so that the optimal parameter can be learned. The experiment is implemented by using the intrusion detection database of KDD99 CUP as the simulation data,. The result of the proposed model is compared with BPNN, PCA-N and PCA-PSO-NN and it shows the proposed model has the highest detection rate and detection efficiency.