Privacy-Preserving Data Aggregation Scheme Based on Federated Learning for IIoT

The extensive application of the Internet Things in industrial field has formed (IIoT). By analyzing and training data from Things, intelligent manufacturing can be realized. Due to privacy concerns, various institutions cannot shared, which forms islands. To address this challenge, we propose a privacy-preserving aggregation federated learning (PPDAFL) scheme for IIoT. In learning, is adopted protect model changes provide security devices. utilizing practical Byzantine fault tolerance (PBFT) algorithm, each round selects an IIoT device area as initialization node, uses single user while resisting reverse analysis attacks management center. Paillier cryptosystem secret sharing are combined realize security, tolerance, sharing. A performance evaluation show that reduces computation communication overheads guaranteeing privacy, message authenticity, integrity.