DLPriv: Deep Learning Based Dynamic Location Privacy Mechanism for LBS in Internet-of-Vehicles
Ziwen Wang, Baihe Ma, Zhihong Liu, Yong Zeng, Zhe Wang, KaiChao Shi · 2023
In the Internet of Vehicles (IoV), location-based service providers rely on collecting and analyzing user-generated trajectory data to deliver high-quality experiences anytime and anywhere. However, the utilization of data mining technologies raises concerns regarding the potential inference of personal privacy information, posing security risks to user privacy. Existing studies typically use differential privacy to obfuscate actual trajectories, albeit at the cost of reduced data utility. To address this challenge and strike a balance between privacy and data utility, we propose DLPriv, a dynamic location privacy mechanism. DLPriv leverages a Long Short-Term Memory (LSTM) network to predict a driver's next location, which in turn determines the level of data obfuscation. This mechanism considers two optimization objectives, trajectory privacy and utility, and enables fine-grained control over privacy levels. Experimental results demonstrate that our proposed mechanism improves data utility by 37% compared to existing work, while providing the same level of privacy.