A Novel Time-Obfuscated Algorithm for Trajectory Privacy

• Main Authors: Chung, Haowei, 鐘浩維
• Other Authors: Hwang, Renhung
• Format: Others
• Language: en_US
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/60868005984944334519

碩士 === 國立中正大學 === 通訊工程研究所 === 100 === Location-based services (LBS) which bring so much convenience to our daily life have been intensively studied in the past years. Generally, a LBS query processing can be categorized into snapshot and continuous queries which search on user location information and reply searching results to the users. A LBS has full control to these location information, causing a user privacy concern. If a LBS has a malicious intention to infer the user privacy by tracking the user routes to their destinations, it incurs a serious problem. Most existing techniques addressed privacy protection mainly for snapshot queries. However, providing privacy protection for continuous queries is more important and challenging since a malicious LBS can easily obtain a complete user privacy information by observing a sequence of successive query requests. In this thesis, we propose a comprehensive trajectory privacy technique and combines ambient conditions to cloak location information based on the user privacy profile to avoids a malicious LBS reconstructing a user trajectory. We first propose a r-anonymity concept which preprocesses a set of similar trajectories R to blur the actual trajectory of a user. We then combine k-anonymity with s road segments to protect the user privacy. We introduce a novel time-obfuscated technique which breaks the sequence of the query issuing time for a user to confuse the LBS from knowing the user trajectory by sending a query randomly from a set of locations residing at the different trajectories R. Despite the randomness incurring from the obfuscation process for providing a strong trajectory privacy protection, the experimental results show that our trajectory privacy technique maintains the correctness of the query results at a competitive computational cost.