Research On The Intrusion Detection Method Based On Takagi-Sugeno Fuzzy Neural Network

The intrusion detection technology, as one of the most crucial dynamic security technologies, can monitor running state of network, hosts and applications comprehensively and real-timely, provide real-time and active protection to interior, exterior attacks and mal-operation, and recognize and respond to the intrusion into network or hosts. As an important active security mechanism, Intrusion Detection will greatly enforce the traditional system security mechanism. By constructing dynamic security circle, the safety of systems can be improved in the highest degree, and hazards brought to systems by security threats can be remarkably reduced. So, research of the intrusion detection technology is widely concerned on paid attention to in the computer network security field.In this dissertation, difficulties that traditional intrusion detection is confronted with in new network environment are analyzed. In order to overcome these difficulties, a new detection algorithm—T-S FNN based network intrusion detection algorithm is presented. This algorithm uses T-S FNN to classify subjects, divide eigen-space of subjects and recognize normal behaviors and intrusions. A lot of research and experiments are done on the Takagi-Sugeno (T-S) model based fuzzy neural network to analyze it in detail. Referring to the principle of the BP algorithm and the structure of t-s model, an adapted BP algorithm is presented to adapt the membership parameters. In order to make the FNN more rapid and more effective, Genetic Algorithm is applied to train the weights. With the network intrusion detection taken as the starting point, T-S model base FNN is applied to network intrusion detection. According to the features of network intrusion detection, the corresponding detection rules and the data analyzing method of intrusion detection is explained.Simulation experiments about the new method of T-S FNN based network intrusion detection are performed on the Matlab platform. It is shown by the results that the new method is feasible, effective and extensible. With the false detection rate reduced, the rate of correct detection is raised by a certain extent too.