Research Into Intrusion Detection Approach To System Implementation Based On Neural Network

With the Internet's rapid development and wide application, computers and computer networks have become the target of crime, security of computers and networks has become the focus of studies. Intrusion detection, an active security protection technology, is a necessary supplement to the traditional static defense technology and has been thoroughly studied and widely used.The intelligence is the hotspots in studies of intrusion detect and the theory of machine learning is widely used in this field. The thesis applied neural network method to IDS in order to improve the efficiency of detecting and increase the ability of identifying the unknown attacks. The main content is as follows.(1)Analyze intrusion detection and neural network technology, and study the shortcoming of the standard BP. In order to improve the convergence velocity of BP and solve the problem that the training process easily falls into local minimum, the optimized algorithm which combines adaptive learning rate adjustment with the additional momentum is proposed and is applied to the intrusion detection network.(2)Analyze intrusion detection data set of KDDCUP 99 and use PCA to reduce the dimensions of the data. Design and analyze the experiment result of optimization-based BP neural network by the above data, which verify that the improved BP network can make up the shortcomings of the standard BP network and improve the efficiency of detecting.(3)Analyze the detection ability of improved BP network on unknown pattern attack and classification attack. The experiment result shows that it has great ability of detecting the unknown pattern attack and better ability than other types on known type of DoS attack and Probing attack.(4)Using VC6.0, and by TCP/IP protocol analysis, a prototype system based on an optimization-based BP neural network intrusion detection is designed. In the LAN environment, the test results confirm the good performance of the system.Finally, the summary of the thesis and the next step of research ideas are given.