Instagram User Footprint Identification Based On Multidimensional Composite Encrypted Traffic Features

The development of smart phones and social networks has brought great convenience to our life.User behavior and user footprints have become more complex.Due to the increasing requirements of user privacy,user data is protected by encryption protocol,but it also makes it difficult to monitor malicious behavior.The existing user behavior identification methods use the statistical features of encrypted traffic,but there are too many interference data.It make the features of encrypted traffic fluctuate greatly.At present,user behavior identification mainly focuses on application identification and application service identification,but now user behavior becomes more complex,and fine-grained identification of user behavior also needs to be studied.In addition,encrypted traffic in the transmission process,generally composed of multiple behavior of the footprint flow,and interference data will be more.Therefore,it is challenging to eliminate interference data from encrypted data,then effectively realize user behavior identification and user footprint identification.In view of the above problems,this thesis proposes a new method of user behavior identification and user footprint identification based on the features of encrypted traffic transmission for Instagram,which mainly includes the following research contents.(1)A method of user behavior identification based on multi-dimensional composite encrypted traffic features is proposed.This method obtains stable data from the initial encrypted traffic for identification.It adopts a method of extracting stable data based on EADU(encrypted application data unit)features,which reduces the interference of network fluctuations on features.Based on the principle of maximum entropy,the method divides the length features of the stable data into the distribution range for sparse processing.It maps the feature space to the vector space to construct a multi-dimensional space.The experimental results show that the proposed method based on multi-dimensional composite encrypted traffic features can accurately identify the user behavior in social networks.And it has a low false positive rate.(2)A method of user footprint identification is proposed.The method needs to obtain stable data and EADU related features.It divides user footsteps into two ways.One is the selection of the footprint segmentation point under the known number of user footsteps.The relatively high time interval points are obtained as the footstep segmentation points by using the time interval of the stable data EADU.Another is the selection of the footstep segmentation point under unknown number of user footsteps.By obtaining the mean and standard deviation of time interval points of EADU with stable data.It uses the characteristics of time feature distribution of EADU,the multiple of the mean and standard deviation is calculated to obtain the corresponding time interval value,which is used as the footprint segmentation point.Then,the experiment uses the user behavior identification method of multi-dimensional composite encrypted traffic features to identify non-repetitive user footprints and repetitive user footprints respectively.The experimental results show that the method can achieve a better effect.(3)Based on the above methods,this thesis designs and implements a prototype system of user behavior identification and user footprint identification based on multi-dimensional composite encrypted traffic features.The whole system framework is designed,including data preprocessing module,user behavior identification module and user footprint identification module.The system can set parameters and output humanized.The system will generate EADU related feature files,CSV files,user behavior identification results,user footprint identification results,and give users an intuitive interface output.