The De-anonymization Method Based on User Spatio-Temporal Mobility Trace

Nowadays user mobility traces are more and more likely to de-anonymize users in addition to other types of data. Among them, model-based approaches usually provide more accurate de-anonymize than that of feature-locations-based approaches. More recently, Hidden Markov Model (HMM) based approaches are proposed to find out the mobility pattern of user mobility traces. However, in the key step of the hidden state definition, existing models rely on the fixed classification of time, space or number, which can hardly suit for all various users. In this paper, we propose a user de-anonymization method based on HMM. Different from current approaches, the method utilizes a novel density-based HMM, which uses the density-based clustering to obtain hidden states of HMM from three-dimensional (time, latitude and longitude) spatio-temporal data, and provide much better performance. Furthermore, we also propose a frequent spatio-temporal cube filter (FSTC-Filter) which significantly reduces the number of candidate models and thus improves the efficiency.

[1]  Hanan Samet,et al.  Aircraft Trajectory Prediction Made Easy with Predictive Analytics , 2016, KDD.

[2]  Hui Zang,et al.  Anonymization of location data does not work: a large-scale measurement study , 2011, MobiCom.

[3]  Xing Xie,et al.  Mining interesting locations and travel sequences from GPS trajectories , 2009, WWW '09.

[4]  Vinayak A. Rao,et al.  Markov-modulated Marked Poisson Processes for Check-in Data , 2016, ICML.

[5]  Vitaly Shmatikov,et al.  De-anonymizing Social Networks , 2009, 2009 30th IEEE Symposium on Security and Privacy.

[6]  Nicholas Jing Yuan,et al.  Regularity and Conformity: Location Prediction Using Heterogeneous Mobility Data , 2015, KDD.

[7]  Kevin Chen-Chuan Chang,et al.  Mobile user verification/identification using statistical mobility profile , 2015, 2015 International Conference on Big Data and Smart Computing (BIGCOMP).

[8]  Rong Wang,et al.  A De-anonymization Attack on Geo-Located Data Considering Spatio-temporal Influences , 2015, ICICS.

[9]  Xing Xie,et al.  Destination prediction by sub-trajectory synthesis and privacy protection against such prediction , 2013, 2013 IEEE 29th International Conference on Data Engineering (ICDE).

[10]  Vitaly Shmatikov,et al.  Robust De-anonymization of Large Sparse Datasets , 2008, 2008 IEEE Symposium on Security and Privacy (sp 2008).

[11]  Shashi Shekhar,et al.  Discovering personally meaningful places: An interactive clustering approach , 2007, TOIS.

[12]  César A. Hidalgo,et al.  Unique in the Crowd: The privacy bounds of human mobility , 2013, Scientific Reports.

[13]  Reza Shokri,et al.  Evaluating the Privacy Risk of Location-Based Services , 2011, Financial Cryptography.

[14]  Kevin Chen-Chuan Chang,et al.  Mobility Profiling for User Verification with Anonymized Location Data , 2015, IJCAI.

[15]  Luming Zhang,et al.  GMove: Group-Level Mobility Modeling Using Geo-Tagged Social Media , 2016, KDD.

[16]  Nadia Magnenat-Thalmann,et al.  Time-aware point-of-interest recommendation , 2013, SIGIR.

[17]  Martin Vetterli,et al.  Where You Are Is Who You Are: User Identification by Matching Statistics , 2015, IEEE Transactions on Information Forensics and Security.

[18]  Sébastien Gambs,et al.  De-anonymization attack on geolocated data , 2014, J. Comput. Syst. Sci..

[19]  Albert-László Barabási,et al.  Understanding individual human mobility patterns , 2008, Nature.

[20]  Elaine Shi,et al.  Link prediction by de-anonymization: How We Won the Kaggle Social Network Challenge , 2011, The 2011 International Joint Conference on Neural Networks.