Privacy-Preserving Subset Aggregation with Local Differential Privacy in Fog-Based IoT

As a typical novel IoT(Internet of Things) architecture, fog-based IoT is promising to decrease the overhead of processing and movement of large-scale data by deploying storage and computing resources to network edges. However, since the edge-deployed fog nodes cannot be fully trustable, fog-based IoT suffers some security and privacy challenges. This paper proposes a novel privacy-preserving scheme that can implement data aggregation from a subset of devices in fog-based IoT. Firstly, our scheme identifies the subset to be aggregated by computing the Jaccard similarity of attribute vectors of the query users and IoT devices, where the local differential privacy is employed to protect the attribute vectors. In addition, we use local differential privacy truth discovery to protect the data of IoT devices and improve the accuracy of the aggregation result. Finally, experiments show that our scheme is efficient and highly available by comparing it with state-of-the-art works. Theoretical analyses demonstrate that our proposed scheme has excellent performance on both computational costs and communication costs.