An Efficient Grid-based Privacy-preserving Framework Against User-Server Collusion in Spatial Crowdsourcing
Lening Yuan, Junyi Li, Zhimao Gong · 2022
With the popularity of spatial crowdsourcing services, the issue of privacy leakage has attracted widespread attention from institutions and researchers, especially the location privacy of participants. To address this issue, several location privacy-preserving frameworks have been proposed. However, these frameworks have limited performance in terms of task assignment efficiency and lack consideration of the threat of brute force attacks under user-server collusion. Facing these challenges, in this paper, we propose an efficient grid-based privacy-preserving framework to achieve a two-sided improvement in assignment efficiency and security level. For the first challenge, we utilize a recoding algorithm to construct a grid-index Quadtree to reduce the time cost of finding suitable workers, and finally achieve faster-than-linear task assignment complexity. For the other challenge, we introduce a two cloud server model, including a Re-Server and a SC-Server, and propose a Two-Step Re-Encryption scheme based on secure Knn computation. The task assignment request is first sent to the Re-Server and then forwarded to the SC-Server after re-encryption. SC server returns the assignment result and the worker is finally notified by the Re-Server. To break the way of user-server collusion, we also design two parameters, hash signature and valid timestamp, to achieve unlinkability between data and user identities. The extensive experiments based on real-world datasets show that our framework is more efficient in task assignment than state-of-the-art works and more suitable for big data scenarios.