Abstract: K-anonymity for location privacy is an important solution to protect the user’s trajectory in LBS．However, K-anonymity for location privacy can not always protect user’s privacy if the attackers have any background knowledge. We propose a dynamic anonymity algorithm based on historical data trajectory prediction. This algorithm uses the trajectory prediction ability to define the background knowledge of attackers’, when the user’s trajectory has a predicted risk, changs the value of anonymity set. A series of experiments on synthetic datasets are made, the results show this algorithm is feasible and effective.


Key words: history data, dynamic anonymity algorithm, trajectory prediction