Research On Network Security Situation Assessment And Prediction Based On Bayesian Network And Support Vector Machine

With the development of economy,computers have entered into thousands of households,and the openness,sharing and interconnection of networks have become more and more significant,bringing great convenience to people's life and work.However,various network security incidents keep happening,and network security is facing great challenges.Network security situational awareness technology emerges in this environment,and the study of this technology is of great significance to the study of network security.Network security situation awareness refers to the acquisition of the security elements that will affect the security situation of large-scale networks,and the fusion,mining and analysis of data.Finally,visualization technology is used to predict the future trend.This thesis mainly uses bayesian network modeling and support vector machine method to evaluate and predict network security situation.The specific research work is as follows:1.A network security situational awareness model based on bayesian network is proposed.Random forest method is used to mine the factors affecting network security,and bayesian network security method is used to recombine and analyze the factors affecting network security,so as to obtain the most accurate and detailed network security situation information.The random forest method can be used to process multi-feature data without feature selection,and important features can be given after training.At the same time,the parallelization method can be used to train quickly,which can improve the speed of constructing bayesian network.For the uncertain security situation in the network,the bayesian method is used as the inference tool,and the network topology can be modified through the bayesian scoring function,and finally the network security situation value can be obtained through quantitative analysis.2.The support vector machine method is proposed to improve the accuracy of network security situation prediction.Support vector machine(SVM)is a mature andefficient algorithm in the current network security situation algorithm because of its versatility,simple calculation,high operational efficiency and perfect theory.Because the prediction accuracy of SVM method is greatly affected by the parameters of kernel function,genetic algorithm is used to optimize the parameters of the function.Firstly,the situation data is trained,and then the optimal prediction model parameters are obtained through the genetic algorithm.Finally,the network security situation is predicted through the optimized prediction model to obtain the optimal prediction value.3.Experimental verification and analysis.In this thesis,KDD CUP 99 dataset was used to verify the model.Simulation experiments showed that the proposed method could improve the speed and accuracy of network security situation assessment and prediction,which all showed the feasibility of the proposed method.