BC-FedX: A Blockchain-Enhanced Cross-Layer Federated Learning Framework for Adaptive Security and Privacy Preservation in Cyber-Physical Systems

Cyber-Physical Systems (CPS) deployed in smart grids, industrial IoT, and autonomous systems face critical security and privacy challenges due to decentralization and heterogeneous devices. Existing blockchain-enabled federated learning approaches improve trust but suffer from scalability limitations and vulnerability to poisoning attacks. This paper proposes BC-FedX, a blockchain-enhanced cross-layer federated learning framework integrating hierarchical federated learning, lightweight proof-of-trust consensus, adversarially robust aggregation, and zero-knowledge privacy verification. Experimental results demonstrate 96.7% detection accuracy, 41% communication reduction, and 52% improved resilience against poisoning attacks. The framework is suitable for secure industrial IoT and smart grid deployments.