Research On Network Security Situation Element Acquisition And Evaluation Technique

With the development and popularization of internet,the influence of it on society is getting bigger and bigger.The problem of network security is becoming more and more serious.Gradually,it also becomes the key problem of internet,network services and applications,which needes to be solved in further development.In order to cope with the current security threats to the network,network security situational awareness technology is put forward.It can assess the network security situation and predict the future development trends through integrate all the aspects of the security elements.This paper aims at the elements acquisition and evaluation in situation awareness system.The main content includes the following two aspects:First,in order to reduce the time complexity of situation element acquisition and cope with the low detection accuracy of small class samples caused by imbalanced class distribution of attack samples in large-scale networks,a situation element acquisition mechanism based on deep auto-encoders network is proposed.In this mechanism,the improved deep auto-encoders network is introduced as basic classifier to identify data types.On the one hand,when the auto-encoder is in the taning,the training rule based on cross entropy function and back propagation algorithm is adopted,which can overcome the problem of slow weights updating by the traditional variance cost function.On the other hand,in the stage of fine-tuning and classification of the deep network,an active online sampling algorithm is applied in the classifier to select the samples online for updating the network weights,which can eliminate redundancy of the total samples,balance the amounts of all sample types and improve the classification accuracy of small class sample.Through the simulation and analysis of the instance data,the scheme has a better accuracy of situation element acquisition and smaller communication overhead in the data transmission.Second,we research the network security situation evaluation.Aiming at the single structure of current situation assessment,incomplete elements and too much dependents on the experience of experts,we put forward a method of situation assessment based on analytic hierarchy process and bayesian network.Primarily,a hierarchical situation assessment model is built by analyzing the indicators of network security situation comprehensively,such as the status of network,aggressive behavior,and system vulnerabilities.Then,in order to solve the problem of indexes weight completely depending on subjective experience,the analytic hierarchy process algorithm optimized by improved glowworm swarm optimization is been used to calculate the weight of each index,and according to the specific weight of second-class indexes,it determines the structure of bayesian network.At last,the simulation results show that the evaluation model can reduce the impact of expert experience,and effectively improve the credibility of the assessment.