PLTA: Private Location Task Allocation using multidimensional approximate agreement
Abderrafi Abdeddine, Amine Boussetta, Youssef Iraqi, Loubna Mekouar · 2024
Mobile Crowdsensing leverages the widespread use of smartphones to gather valuable data for various applications. It primarily involves users who either request or perform sensing tasks alongside a platform that manages and coordinates the process. For effective operation, this platform must collect accurate user information, often including sensitive data such as location information. Most existing solutions face a trade-off between privacy and utility or suffer from high time complexity. Our solution relaxes this trade-off while maintaining low time complexity, using a unique combination of matching algorithms and approximate agreement protocols. Theoretically, we establish an upper bound on the maximum noise that can be added to users’ location data while preserving the performance of the matching algorithm. Empirically, we demonstrate that our algorithm achieves superior privacy-preserving matching performance and the lowest execution time compared to related work.