Research On Network Security Situatuion Element Acquisition And Evaluation

With the rapid development of the Internet of Things,cloud computing and other technologies,the vast amounts of communications equipment are filled in the life and industrial production.Therefore,the network security issues diverse and complicated.In order to effectively manage the network security status,dynamically understand the specific situation of large-scale network security,comprehensively analyze the past and current network conditions,and achieve as accurately as possible the prediction of the future network conditions.As a result,network security situation awareness has attracted more and more attention.This thesis mainly aims at the acquisition of network security situation elements and situation assessment,and the specific work is as follows:1.In view of the serious sample imbalance and low classification accuracy of element acquisition problems in the current massive situational element acquisition data set,the factor acquisition model based on sample imbalance is proposed by deep learning.Firstly,the convolutional neural network is utilized to extract the deep features of the network traffic data,realize the pre-classification of the samples,and obtain a base classifier.Secondly,the generation adversarial network generates the very realistic sample characteristics,which can expand small samples and effectively solve balanced the distribution of samples.Finally,combining with transfer learning on the expanded balanced data set,the classification accuracy and convergence speed of small samples are improved effectively.and the training speed of the model is also improved.The simulation results demonstrate that the expanded data set of generation adversarial network and the application of transfer learning,can effectively improve the model training convergence speed and the classification accuracy of network security situation elements.2.To solve the problem that traditional network security situation assessment technology relies too much on expert experience,and is highly subjective,the security situation assessment method based on IFOA-RBF neural network is proposed.This method firstly improves the fruit fly algorithm through the crossover and mutation mechanisms.The improved fruit fly algorithm was used to conduct global optimization for the weight,center of radial basis function and radial basis width of RBF neural network.Finally,the situation assessment was carried out using the optimized RBF neural network.The simulation results show that the evaluation results of optimized RBF neural network model are more precise when evaluates the network security situation.To a certain extent,it can reduce the influence of subjective factors in the evaluation process,and ensure the objective and true of evaluation results.