A Markov Prediction-Based Privacy Protection Scheme for Continuous Query

Abstract

The query probability of a location which the user utilizes to request location-based service (LBS) can be used as background knowledge to infer the real location, and then the adversary may invade the privacy of this user. In order to cope with this type of attack, several algorithms had provided query probability anonymity for location privacy protection. However, these algorithms are all efficient just for snapshot query, and simply applying them in the continuous query may bring hazards. Especially that, continuous anonymous locations which provide query probability anonymity in continuous anonymity are incapable of being linked into anonymous trajectories, and then the adversary can identify the real trajectory as well as the real location of each query. In this paper, the query probability anonymity and anonymous locations linkable are considered simultaneously, then based on the Markov prediction, we provide an anonymous location prediction scheme. This scheme can cope with the shortage of the existing algorithms of query probability anonymity in continuous anonymity locations difficult to be linked, and provide query probability anonymity service for the whole process of continuous query, so this scheme can be used to resist the attack of both of statistical attack as well as the infer attack of the linkable. At last, in order to demonstrate the capability of privacy protection in continuous query and the efficiency of algorithm execution, this paper utilizes the security analysis and experimental evaluation to further confirm the performance, and then the process of mathematical proof as well as experimental results are shown.