Digital-Twin-Enabled Industrial IoT: Vision, Framework, and Future Directions

Building digital twins (DTs) in industrial Internetof- Things (IIoT) is challenging, especially considering complex and large-scale network architectures, real-time data requirements, and computational demands. Traditional modeling-based approaches, relying on either small datasets with physics-based models or large datasets processed through artificial intelligence (AI) techniques, face limitations in adaptability and accuracy. In this article, we propose a hybrid DT framework to address these challenges, which integrates both physics-based models and AI techniques. The physics-based models, grounded in communication, computing, and caching (3C) resources, ensure alignment with known system behaviors and predetermined assumptions, while the AI components dynamically adapt to real-time network environments, allowing the DT to learn from the evolving environments. The proposed DT framework incorporates layers that support real-time data acquisition, data processing, and decision-making through continuous feedback, enhancing system performance and enabling proactive maintenance, quality control, and optimization. A hybrid model-based case study demonstrates that the proposed framework can reduce packet queuing size and improve network performance under varying network load and outage conditions. Finally, open research issues for DT in IIoT are discussed.