Research On Network Security Situation Based On Multi Source Data Fusion

With the increasingly complex network environment,increasing the scale of network,various network attacks emerge in endlessly,the traditional network security system has been unable to meet the people demand for network security,the model of network security platform building needs new technical support.About network security situation perception theory research,timely develop people awareness of network security problems of vision,breaks through the limitation of traditional security defense system,guide people to consider the problem of network security from the point of view of the structure of the whole network,for the timely in view of the potential of the network attack hit prediction provides a feasible solution.Network security situational awareness is based on data fusion,the data of the network security platform is about to be integrated to judge the running state of the network,so as to predict the next phase of the network.From the point of view of data fusion,this paper designs a network security situational awareness model based on multi-source data fusion.The core technologies used include data fusion,data mining,the theory of unbalanced data,etc.In the process of building a network security situational awareness model,this paper does the following work on how to construct an efficient and accurate data fusion model and improve the efficiency of the model:(1)through the analysis of the relationship between perceived levels of model and data fusion of network security situation,put forward in feature extraction stage of processing data fusion algorithm to improve,is based on noise from the encoder(sdae)and the random forest algorithm(RF)algorithm SDAE-RF.And in the model pre test stage design elements quantization algorithm model of early warning based on rough set.Experiments show that the SDAE-RF algorithm has certain advantages compared with the traditional algorithm,which improves the accuracy of classification,and verifies the validity of the early warning model.(2)in processing data fusion model in imbalanced data problems,and puts forward two kinds of improvement,one is the algorithm design of the redundant attribute detection algorithm based on rough set,by calculating the weights of the attributes of the elements of the removal of redundant attributes;the second is in the random forest algorithm based for unbalanced data processing were optimized,design the fast random forest algorithm FRF(fast Radom forest)in the network situation element extraction stage.The experimental results show that the improved network situation factor extraction algorithm has obvious effect on the processing of unbalanced data.