Neural Network-based Intrusion Detection Systems And Realization

With the development of network, a growing number of departments and individuals start working on network or as a part of life, the issue of network security is increasingly important and gradually become a hot this today, the traditional protection technology of network security mainly rely on the firewall as a security protective measures. Along with the larger-scale network and the gradually diverse and complex invasion or attacks. Mainly relaying on firewall technology the defense of network security is difficult to deal with the new situation in the network. Intrusion Detection technology will make up the insufficient of firewall to enhance network security. Intrusion Detection as a new network technology has many development direction. Intelligent Intrusion Detection is one important branch.This paper introduced the background of network security and Intrusion Detection status, also analyzing the typical method of attack and invasion from vulnerability scanning, crack passwords, scripting attacks, denial of service attacks and other aspects. Intrusion Detection System is the basis by these methods to detect. With current Intrusion Detection products, Snort is an excellent open-source Intrusion Detection system, and even many commercial Intrusion Detection products also adopted some design ideas form snort. This paper analyzes the programming structure of the snort, the detection process of the packet and the detection engine in snort, and analyzing the deficiencies of snort according to the rules chain of the snort. With improvement on the Momentum Backprogation algorithm, the simulation test shows the increasing Momentum Backprogation than the standard BP. Acording to the improved algorithm and the plug-in mechanism of snort, design of the neural network plug-in which picking samples and the neural network plug-in which detecting in snort. Finally analyze the experiment, this has the exploration and study with practical significance for intelligent Intrusion Detection System.