The Research Of Network Intrusion Detection Technologybased On RBFNN-HMM

With the rapid development of Internet, network security problemshave become increasingly severe。The current security defense techniquesare mainly static security techniques,for example,firewall,router filter。Static security techniques work on preventing invalid access of system。Static security techniques can not totally assure the security of networkand resist the attack of hacker when genuine network attacks occur,especially novel network attacks. Therefore, as an initiative anddynamic network security defense technique, the intrusion detectiontechniques become a new direction of the development of networksecurity.This dissertation is described from the Status of network security ,firstly introduces the traditional network security technology , theexisting common network attacks and the basic concepts of intrusiondetection,and research Status of intrusion detection, and analyzes theexisting deficiencies.At the same time, the deficiency and virtue of the hidden Markovmode (HMM) and neural network (NN) are analyzed in the article. Themandarin digit recognition model based on the RBFNN & HMM isproposed. And uses the RBF artificial neural network (RBFNN) fortraining, then proposes a intrusion detection method which is based onRBF artificial neural network hidden Markov model (RBF-HMM)combined the good resistance of neural network and the well-modelingadvantages of hidden Markov model to improve the recognitionperformance.Experiments show that the IDS is effective in reducing false positiverates,false negative rates better and it can be improved in detection ratesto detect unknown attack. All above improve the system performance.