A blockchain and A-DCNN integrated framework for privacy protection and intrusion detection of industrial IoT

In the industrial Internet of Things (IIoT), the interconnection between networks and hardware and between networks themselves produces a substantial volume of data. This leads to various security and privacy issues, greatly increasing the risk of data privacy breaches. Existing methods are not effective in ensuring secure storage of this data, and the effectiveness of intrusion detection is also not satisfactory. This paper proposes using blockchain and an attention-deep convolutional neural network (A-DCNN) based on deep learning to protect data security further and improve intrusion detection accuracy in IIoT. Firstly, blockchain is implemented using the hyperledger fabric framework in a consortium chain, combined with the interplanetary file system, to achieve distributed data storage. This prevents data poisoning attacks and ensures data immutability and traceability. Secondly, a deep variational autoencoder is employed for the purpose of re-encoding the original data into a novel format, thus deterring potential attackers from conducting inference attacks. In addition, in the data preprocessing stage, the synthetic minority over-sampling technique combined with edited nearest neighbors is applied to enhance the sample data, significantly improving the accuracy, precision, and other evaluation metrics for small samples. Finally, the A-DCNN model is used in the intrusion detection module to detect malicious intrusions in IIoT. We conducted experiments using the new generation IIoT 4.0 datasets TON_IoT and the IoT network traffic datasets BoT-IoT. In the BoT-IoT datasets, the five-class precision reached 97.03%, and the binary classification accuracy reached 98.50%. In the TON_IoT datasets, the ten-class precision reached 99.50%, and the binary classification accuracy reached 97.86%. Compared to several existing techniques, these experimental results demonstrate that this framework effectively protects IIoT data and detects malicious intrusions in IIoT.