Research On Network Security Situation Prediction Method Based On Hidden Markov Model

As human beings enter the information age,the network has undergone large-scale development,and the corresponding networks have been attacked in an endless stream,and the number of attacks on the network has increased year by year.Therefore,network security has become an urgent problem to be solved.More and more research experts have proposed research on network security situational awareness technology,providing solutions for the increasingly prominent network security problems.This paper studies the network security situation recognition method based on hidden markov model,aiming at more accurately realizing the identification of network security situation.In this way,the network security management personnel can identify the current network security posture,perform a more accurate and in-depth understanding of the current network state,and timely adjust the current network security deployment.At the same time,the network security management personnel can also deploy the defense measures in the current network environment in time according to the predicted network security situation in the future,so as to avoid more serious security problems in the network in the future.In this paper,the basic concepts of network security situational awareness and hidden markov model are studied respectively.It is found that the network security situation elements can correspond to the key elements in the hidden markov model.Therefore,based on the characteristics of network security situation,this paper establishes a hidden markov model for network security situation,and designs a process based on hidden markov model to identify and predict network security situation.Firstly,according to the internal attribute characteristics of the alarm information in the network,an original alarm information screening model is established,in order to filter the redundant information in the original alarm information.Secondly,the hidden markov model for the network security situation is established.The filtered alarm information is put into the observation layer of the model,and the network security posture is classified as the hidden layer of the model.Then,the problem of parameter selection of hidden markov model is studied,and amore scientific solution is proposed.The initial state transition probability distribution matrix is determined by considering the joint effect of attack behavior and protection measures on the network security situation.Furthermore,this paper improves the Forward algorithm of hidden markov model and more accurately depicts the network security situation at a certain moment.At the same time,the Baum-Welch algorithm of hidden markov model is improved,and the factors of considering the state of the previous moment are added.The improved Baum-Welch algorithm is used for parameter learning to provide more accurate calculation for the identification of network security situation basis.Finally,according to the Viterbi algorithm of the hidden markov model,a hidden security state sequence corresponding to a certain alarm information sequence can be identified,and the prediction of the network security situation at the next moment can be realized.It is proved by experiments that the improved hidden markov model algorithm is more accurate than the unimproved hidden markov model algorithm for network security situation recognition,which improves the accuracy of network security situation recognition and also achieves accurate prediction of the network security situation at the next moment.