Volume - 7 | Issue - 4 | december 2025
Published
20 December, 2025
The continued growth of the Industrial Internet of Things (IIoT) has produced sensitive data in abundance across multiple distributed industrial environments. Current trends highlight the need to ensure data privacy and the capability of collaborative intelligence across IIoT networks.This research proposed a Blockchain-Enabled Privacy-Preserving Federated Learning (BcPPFL) framework for securing collaborative model training through Industrial IoT (IIoT) networks. The proposed BcPPFL framework was studied experimentally with benchmark IIoT datasets (ToN-IoT and N-BaIoT) by varying the number of clients (10–100) within a non-IID data distribution. The federated learning model consistently determined successful iterations above 92% accuracy across the benchmarking while demonstrating convergence with respect to learning by reaching optimal efficiency variably (40, 50 rounds) in heterogeneous environments in IIoT. Blockchain-Enabled Privacy-Preserving Federated Learning (BcPPFL) did not significantly introduce latency as block confirmation times did not exceed an average of 3.8 seconds at 85 TPS throughput. Security-wise, it was validated that BcPPFL reduced individual data reconstruction and membership inference attacks by 64-70%, with respect to successful attacks, and improved common poisoning detection rates from 46% to over 89%, due to the secure aggregate in consideration of smart contract enforcement. These statistical results evidenced a scaled collation of data, resilience to cybersecurity attacks, and the proven use of a collaborative platform for real-time IIoT applications where privacy and integrity are imperative.
KeywordsFederated Learning Blockchain Industrial IoT Privacy Preservation Smart Contracts Secure Aggregation Differential Privacy Industry 4.0 Edge Computing Cybersecurity
