Privacy-Preserved Cyberattack Detection in Industrial Edge of Things (IEoT): A Blockchain-Orchestrated Federated Learning Approach
Mohamed Abdel‐Basset, Nour Moustafa, Hossam Hawash
Abstract
The Industrial Internet ofThings (IIoT) plays an essential role in the digital renovation of conventional industries to Industry 4.0. With the connectivity of sensors, actuators, appliances, and other industrial objects, IIoT enables data availability, improved analytics, and automatic control. Thanks to the complex distributed nature, a wide range of stealthy and evolving cyberattacks become a major threat to the trustworthiness and security of IIoT systems. This makes the standard security procedures unable to assure the trustworthiness of IIoT that protect against cyberattacks. As a remedy, this article presents a blockchain-orchestrated edge intelligence (BoEI) framework that integrates an innovative decentralized federated learning (called Fed-Trust) for cyberattack detection in IIoT. In the Fed-Trust, a temporal convolutional generative network is introduced to enable semi-supervised learning from semi-labeled data. BoEI includes reputation-based blockchain to enable decentralized recording and verification of the transactions for guaranteeing the security and privacy of data and gradients. Fog computing is exploited to offload the block mining operation from the edge side thereby improving the overall computation and communication performance of Fed-Trust. Proof of concept simulations using two public datasets validate the robustness and efficiency of the Fed-Trust over the cutting-edge cyberattack detection approaches.