Securing Critical Infrastructures: Deep-Learning-Based Threat Detection in IIoT

The Industrial Internet of Things (IIoT) is a physical information system developed based on traditional industrial control networks. As one the most critical infrastructure systems, IIoT also preferred target for adversaries engaged in advanced persistent threats (APTs). To address this issue, we explore deep-learning-based proactive APT detection scheme IIoT. In scheme, considering characteristics long attack sequences and long-term continuous attacks, our solution adopts well-known deep learning model, bidirectional encoder representations from transformers (BERT), to detect sequences. sequence optimized ensure model's judgment effectiveness. experimental results not only show that proposed method has feasibility effectiveness detection, but certify BERT model better accuracy lower false alarm rate when detecting than other time series models.