Kernel Based Network Intrusion Detection System

Network intrusion detection is one of the core technologies of network information security.The essence of intrusion detection is a problem of pattern recognition.SVM is a hotspot in pattern recognition,it has been proved that modeling with SVM can not only improve the performance of generalization,but also has a higher accuracy and lower false alarm rate.First,feature extraction algorithm is performed to filter the noise data and extract the key feature that indicate the intrusion,thus the dimension of intrusion data,training and testing time are all decreased;and then an improved intrusion detection model based on kernel method is constructed,it bases on the CIDF and combines feature extraction method,such as KPCA and KICA,with LSSVM. Experiment results show that the proposed system has a remarkable performance in detecting both existed intrusions and mutated ones.