MDI-BV: A Blockchain-Verified Multimodal Deep Learning-Based Intrusion Detection System for Industrial Internet of Things

As the Industrial Internet of Things (IIoT) is deeply applied in smart manufacturing and remote control, its open architecture and heterogeneous device connections intensify cybersecurity threats such as data injection and DoS. Traditional IDS relying on static rules and single-feature analysis fail to handle complex attacks. This paper proposes MDI-BV, a blockchain-verified multimodal deep learning intrusion detection system. Based on the CIC-IDS-2017 dataset, it extracts four features, utilizes a multi-branch neural network with concatenation fusion and regularization to enhance robustness and generalization. A Hyperledger Fabric-based consortium blockchain ensures end-to-end verifiability and tamper-proof traceability of training/inference. Experiments show it reaches ≥ 99.98% in accuracy, precision, recall and F1-score, with stable performance, strong data integrity and IIoT practicality.