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

Intrusion detection is a technology which can protect our information. It can monitor our systems or networks, and find the intrusions. However, the intrusion detection system has many deficiencies for its short history, e.g., higher false rate, the weakness of active detecting, etc.This paper has further deep research on intrusion detection based on fuzzy neural network technology after analysizing these existing issues using intrusion detection technology based on fuzzy neural network.Firstly, according the limitation of the traditional intusion detection method based on BP fuzzy neural network, an intrusion detection method based on Takagi-Sugeno fuzzy neural network is proposed, which adopts the neural network structure based on Takagi-Sugeno model. Meanwhile, a feature select algorithm (Selected-F) is designed to reduce the dimension of network data properties, a method of category detection to reduce computational cost,to complete intrusion detection.Secondly, an improved genetic algorithm is designed to mediate the various parameters of neural network. Fuzzy thinking to analyze network data are adopted to solve the fuzzy problems of intrusion detection, which improves the robustness of genetic algorithm.Finally, we have a simulation experiment, using KDD CUP 99 standard data set as an experimental data, with Matlab in the Windows XP operating system environment. The experiments show that with the method, attacks could be detected effectively, precisely and real-timely.