Differential privacy-guaranteed trajectory community identification over vehicle ad-hoc networks

In vehicle ad-hoc networks (VANETs), a large number of location-based services can be provided for users according to their movement features. Meanwhile, privacy may be leaked when users publish true trajectory data to VANETs servers. Traditional privacy-preserving methods have some limitations, for example, k-anonymous have to predefine the adversaries' background knowledge. Therefore, this paper proposes a new algorithm to partition and cluster trajectories into a proper community with respect to the particularity of VANETs based on differential privacy. The experimental results show that our method outperforms the suppression algorithm and partition algorithm in perspective from information loss and operation efficiency.