Privacy-Preserving and Fault-Tolerant Aggregation of Time-Series Data With a Semi-Trusted Authority

Time-series data aggregation in Internet of Things applications is a useful operation, where the time-series data is sensed by a group of users, and gathered by the aggregator for real-time analysis. However, some security and privacy challenges still affect the collection and aggregation process. Although existing privacy-preserving solutions achieve strong privacy guarantees, they introduce a fully trusted TA that is difficult to realize in the real world. Besides, they cannot be directly applied in time-series data aggregation scenarios due to unacceptable efficiency. In this article, we propose a privacy-preserving time-series data aggregation scheme with a semi-trusted authority. Moreover, our scheme also supports arbitrary aggregate functions and fault tolerance to enhance the reliability and scalability of data aggregation. Security analysis demonstrates that our proposed scheme achieves $(n-k)$ -source anonymity even if $k(k\leq (n-2))$ data providers collude with the cloud server. We also conduct thorough experiments based on a simulated data aggregation scenario to show the high computation and communication efficiency of our scheme.