Search

Trajectory classification (TC), i.e., predicting the class labels of moving objects based on their trajectories and other features, has many important real-world applications. Private trajectory data publication is to anonymize trajectory data, which can be released to the public or third parties. In this paper, we study private trajectory publication for trajectory classification (PTPTC), which not only preserves the trajectory privacy, but also guarantees high TC accuracy. We propose a private trajectory data publishing framework for TC, which constructs an anonymous trajectory set for publication and use in data services to classify the anonymous trajectories. In order to build a “good” anonymous trajectory set (i.e., to guarantee a high TC accuracy), we propose two algorithms for constructing anonymous trajectory set, namely Anonymize-POI and Anonymize-FSP. Next, we employ Support Vector Machine (SVM) classifier to classify the anonymous trajectories. Finally, the experimental results show that our proposed algorithms not only preserve the trajectory privacy, but also guarantee a high TC accuracy.

With the rapid development of cloud computing paradigm, data owners have the opportunity to outsource their databases and management tasks to the cloud. Due to the privacy concerns, it is required for them to encrypt the databases before outsourcing. However, there is no existing techniques handling range queries in a fully secure way. Therefore, in this paper we focus exactly on secure processing of range queries over outsourced encrypted databases. To efficiently process secure range queries, the extraordinarily challenging task is how to perform fully secure range queries over encrypted data without the cloud ever decrypting the data. To address the challenge, we first propose a basic secure range queries algorithm which is not absolutely secure (i.e., leaking the privacy of access patterns and path patterns). To meet a better security, we present a fully secure algorithm that preserves the privacy of the data, query, result, access patterns and path patterns. At last, we empirically analyze and conduct a comprehensive performance evaluation using real dataset to validate our ideas and the proposed secure algorithms.