Research On Model Of Ids Based On RBF Neural Network

Intrusion detection is an active preventive measure of network security, it not only can be achieved by monitoring the network for internal attacks, external attacks and misuse in real-time protection, and make up a firewall effectively. But also can be combined with other network security products, to protect network all-round. With the initiative and real-time, it is an important and useful supplement for firewall. However, traditional intrusion detection systems whose detection speed is still slow, and false negative rate is high.This dissertation is described from the basic concepts of intrusion detection, firstly introduces the classification of intrusion detection, the existing common network attacks and security measures, and analyzes the existing deficiencies. In intrusion detection system, if only to use the normal rule base, only to determine the data is normal or abnormal, assuming that the normal rule base is not complete, it will lead to high rate of false alarm, but also can not detect specific intrusion type; but simply using neural network to identify the invasion type, all the test data must be handled through the neural network, which increases the time complexity. Therefore, from the point of accelerating the detection speed and improving the identification accuracy, we put forward a model of IDS based on RBF neural network of adaptive genetic algorithm.At the same time, this paper produces the artificial neural networks to apply in intrusion detection, focusing on BP neural network. Considering the problems of BP neural network in practice, we make an in-depth study of BP algorithm and summary its shortcomings. By comparison, we have finally chosen RBF neural network to apply the model for intrusion detection, and use genetic algorithm to optimize RBF network, analyze the design ideas of the model and the implementation process of each module in detail, of course RBF network module is the most important. In theory, the model combines the advantages of RBF neural network and self-adaptive genetic algorithms, from a certain extent, it can improve the accuracy of intrusion detection to reduce false alarm rate.Finally, we test the performance and efficiency of the model, by comparing the test results show that the RBF neural network and adaptive genetic algorithm are applied to the intrusion detection which is feasible. The model of IDS based on RBF neural network of adaptive genetic algorithm proposed in this paper has a certain research value.