Study On Mobile Trajectory De-anonymization Method Based On User Behavior Pattern

Nowadays,with the application and development of mobile sensing,GPS and other positioning technologies and location services,the acquisition of spatio-temporal data becomes very convenient.Location service providers can easily collect the location traces information of billions of users.For commercial interests and scientific research purposes,data of mobile trajectory can be released or shared.However,the direct release or sharing of user trajectory data without processing can bring serious privacy threat.In order to protect the privacy of users,the privacy protection mechanism is usually used to protect the user's mobile trajectory data set before releasing the trajectory data set.Anonymization is one of the important technologies in privacy protection mechanism,and its privacy protection effect depends on the exploration of de-anonymization method.Different anonymization technologies have different privacy protection effects.For the mobile trajectory data processed by different anonymous technologies,the corresponding method of anonymization of mobile trajectory is studied.Therefore,the anonymization technology is required to meet the requirements of such attacks,so that it can play a better protection effect on the user's mobile trajectory data.This paper considers the influence of time and position semantic information on the results of de-anonymization.Facing the data of mobile trajectory processed by different anonymous technologies,the user behavior patterns contained in the mobile trajectory are extracted to realize de-anonymization attack.Aiming at the goal of increasing the success rate of de-anonymization,the research on the method of de-anonymization based on the mobile trajectory of user behavior pattern is carried out.The main contents of this paper are as follows:(1)A semantic trajectory pattern extraction algorithm is designed to extract frequent semantic trajectory pattern set from the mobile trajectory to represent the user behavior pattern and use it as the basis for de-anonymity.The stay region is identified by the algorithm of stay region extraction,and a set of semantic stay region sequences are obtained according to the different importance of each location in different regions.The extracted semantic stay region sequence is used as the input of the algorithm of semantic trajectory pattern extraction.Combined with the transfer time between activity behaviors,frequent semantic trajectory pattern set reflecting user behavior patterns are mined.(2)According to the differences between different user behavior patterns,a new method of de-anonymization based on semantic trajectory pattern is proposed for anonymous mobile trajectory data processed by pseudonym technology.The frequent semantic trajectory pattern set representing user behavior pattern is extracted by using semantic trajectory pattern extraction algorithm.The maximum trajectory pattern set is extracted from it.The user mobility profile is constructed for anonymous users and real users respectively.The corresponding similarity measurement algorithm is designed to measure the similarity between user mobility profile,so as to identify the real users corresponding to anonymous traces.(3)According to the current situation that anonymous technology has more location protection and less disturbance to time,a new method of de-anonymization based on Hidden Markov model is proposed for anonymous mobile trajectory data processed by k-anonymity technology.The number of hidden states and time partition are generated directly from the data,and a more accurate user model is established through the definition of hidden state which is more consistent with the characteristics of user behavior.According to the established hidden Markov model to represent the behavior patterns of different users,the user mobility profiles are constructed,and the corresponding similarity measurement method is designed for the user mobility profiles to realize the mobile trajectory de-anonymization.