The Research Of Intelligent Intrusion Detection Based On Back Propagation Neural Network

With the rapid development of Internet technology, Internet has been an important part of people's lives and brings us great convenience with its rich information resources. However, because of the opening of network, the bug of OS and the diversity of intrusion technology, security problems have become increasingly severe. Conventional network security techniques such as firewall and encryption technique which lack the active intrusion detection mechanism and need manual work to implement and maintenance have limited defense effects and has been unable to meet the higher safety requirements of the present network.As an active security defence technique, intrusion detection which based on neural network offers realtime protection against interior or exterior attack and mistaken operation. This technique which use of the advantages of self-organization, self-study and ability of generalization of the neural network has good recognizing ability to the known attacks as well as to the unknown ones and thus become the focus of network security research at present.This paper opens with the research on intrusion detection. Based on the analyzing of classical intrusion detection system models and techniques, the trend of intrusion detection is described. Then the back propagation neural network has been studied, including the principle, basic flow, problems and corresponding reasons. On the base of analysis of the standard back propagation algorithm, a BP improving algorithm which adjusts the error signal function and method of training is proposed. This improving algorithm has the advantages of slightly modify and easy implement. Using of the improving algorithm, a detailed design scheme of intrusion detection model based on BP neural network which combines misuse detection technique with anomaly detection technique and is made up of six modules is put forward and great emphasis is put in each key module. This intrusion detection model can recognize the unknown attacks and has good recognizing ability to the known attacks. The experimental results show that this system has great effect. Lastly, this thesis gives an analysis of the insufficiency of this model and the method to improve on this model.