Blockchain-Enabled Secure and Trusted Federated Data Sharing in IIoT
Zhou Zhou, Youliang Tian, Jinbo Xiong, Jianfeng Ma, Changgen Peng
Abstract
Federated learning breaks down data silos and promotes the intelligence of the Industrial Internet of Things (IIoT). However, the principal–agent architecture commonly used in federated learning not only increases the cost but also fails to take into account the privacy protection and trustworthiness of flexible on-demand data sharing. To tackle the above challenges, we propose a secure and trusted federated data sharing (STFS) based on blockchain. Initially, we construct an autonomous and reliable federated extreme gradient boosting learning algorithm to crack the data isolation problem, providing privacy protection and verifiability. Furthermore, we design a secure and trusted data sharing and trading mechanism to ensure secure on-demand controlled data sharing and fair trading. Finally, the security of STFS is proved based on the universal composable theory. The results of ample experimental simulations demonstrate the good effectiveness and performance of STFS for IIoT applications.