A Privacy Preservation Model for Dynamic Trajectory Datasets

Abstract Trajectory datasets are proposed to collect the visited location path of users and their visited times. They are generally used in location-awareness applications. Aside from these applications, they are often released to the data analyst with a suitable business reason. Although they are more beneficial, they have several issues that must be considered when they are utilized, e. g. , data privacy and data utility. To achieve data privacy and data utility in trajectory datasets, a well-known privacy preservation model, LKC-Privacy, was proposed. That is, before trajectory datasets are utilized, the at most L-size of unique subsequent paths are suppressed to be at least K indistinguishable tuples. Moreover, the protected sensitive values relate to each group of indistinguishable subsequent paths, they must have the confidence of data re-identification to be at most C. Thus, the trajectory datasets are satisfied by LKC-Privacy constraints, they can guarantee that all at most L-size of subsequent paths always have at least K - 1 duplicate tuples and every protected sensitive value has the confidence of successful re-identification to be at most C. However, LKC-Privacy is proposed to address privacy violation issues in trajectory datasets that are focused on performing one-time data releases. To the best of our knowledge about trajectory datasets, some trajectory datasets are dynamic, i. e. , they allow change when new data is available. Thus, LKC-Privacy could be insufficient to address privacy violation issues in these trajectory datasets. To rid this vulnerability of LKC-Privacy, a new LKC-Privacy is proposed in this work. Moreover, the proposed model is evaluated by extensive experiments. From the experimental results, they show that the proposed privacy preservation model is more effective than LKC-Privacy.