Privacy Preserving For Network Traffic Data Based On Perturbation

With the rapid development of Internet technology,large-capacity storage technology and the gradual expansion of data sharing,the networking and transparency of data has become an irresistible trend.Data information generated by users in the network is frequently used for data mining,resulting in endless network privacy security issues.Therefore,privacy preserving for network traffic in the big data environment has become an important research topic,striving to achieve a balance between data utility and privacy security.Based on the research of network traffic data mining technology and privacy preserving model,this thesis proposes a privacy preserving method based on perturbation for network traffic data,and measures the data utility and security.The contribution of the thesis is listed as follows:(1)Aiming at the influence of redundant attributes contained in network traffic data on classification mining,a feature selection algorithm for network traffic attributes based on information entropy is proposed.A subset of features that are highly correlated with the application categories and have low redundancy between attributes are obtained.The experiment results demonstrate that the feature subset can maintain the characteristics of the original data and can be effectively used for classification.(2)In order to prevent the leakage of sensitive data in network traffic attributes,a perturbation-based network traffic data privacy preserving algorithm is designed,combining with the probability distribution of attributes.On the basis of generating data independently and identically distributed with the original attributes,the algorithm further uses order mapping to restore the corresponding relationship between attributes,so as to improve the utility of data as much as possible while protecting data security.(3)Experiments are employed to validate the effectiveness of privacy preserving algorithm for perturbation-based network traffic data from the aspects of data utility and security.The experiment results show that the perturbed data can still maintain high classification accuracy compared with the original data,that is,the data has better utility.At the same time,the similarity between the data is poor,which can hide the original data well,in other words,the security of the data is guaranteedIn summary,this thesis analyzes and network traffic classification methods and designs the features subset selection algorithm based on information entropy.The thesis focuses on the privacy preserving algorithm for network traffic data and simulation verification was performed on the real network traffic data set.The experiment results show that the network traffic data published by the algorithm can effectively ensure data security while maintaining data utility,alleviating the contradiction between them in existing algorithms.The perturbation-based privacy preserving algorithm proposed in this thesis can effectively solve the privacy leakage problems of network traffic in the process of data mining.