Abstract:

Connected vehicles generate vast amounts of data essential for enhancing transportation efficiency, safety, and user experience. However, the sharing of this data across multiple parties raises significant concerns about data security and individual privacy. Existing data-sharing mechanisms often rely on centralized architectures or basic anonymization techniques, which are vulnerable to data breaches and inference attacks. These approaches fail to provide strong privacy guarantees in multi-stakeholder environments, such as among manufacturers, service providers, and regulators. To address these limitations, we propose a novel framework titled MP-DPC (Multi-Party Differential Privacy Computation), which integrates secure multi-party computation (SMPC) with differential privacy (DP). This framework enables collaborative data analysis across multiple connected vehicle stakeholders without exposing raw data. Techniques such as distributed noise addition, cryptographic protocols, and privacy budget optimization are employed to ensure that individual data contributions remain confidential while preserving overall utility. The MP-DPC framework is used to support real-time analytics for traffic prediction, anomaly detection, and usage-based insurance, all while maintaining strong privacy guarantees. Each stakeholder processes encrypted or obfuscated data locally, and only aggregated insights are shared through privacy-preserving computations. Experimental evaluation of MP-DPC demonstrates its effectiveness in balancing data utility and privacy. The results show a significant reduction in privacy leakage risks compared to traditional methods, with minimal impact on data accuracy. This framework lays a foundation for trustworthy and scalable data-sharing systems in the evolving landscape of intelligent transportation.