Enhancing IIoT Security with Deep Reinforcement Learning for Intrusion Detection
Selamnia Aymene, Lyes Khoukhi, Ayadi Mondher, Zakaria Abou El Houda · 2025
The integration of Industrial Internet of Things (IIoT) systems enhances productivity and data management but also introduces significant security challenges due to rising cyber threats. Traditional intrusion detection methods often struggle to keep pace with the evolving complexity of these threats. In this paper, we propose a novel approach leveraging Deep Reinforcement Learning (DRL) for intrusion detection, combining supervised learning within a DRL framework to address these challenges. Despite the difficulties in designing an effective reward system, our method, tested on the IoT23 dataset, demonstrates superior performance. Through careful parameter optimization and model adjustments, our DRL-based technique outperforms conventional machine learning methods in terms of accuracy, efficiency, and real-time applicability. The results highlight the potential of DRL as a robust alternative for securing IIoT infrastructures against sophisticated cyberattacks.