The Challenges and Prerequisites of Data Stream Processing in Fog Environment for Digital Twin in Smart Industry

Abstract

Smart industry systems are based on integrating historical and current data from sensors with physical and digital systems to control product states. For example, Digital Twin (DT) system predicts the future state of physical assets using live simulation and controls the current state through real-time feedback. These systems rely on the ability to process big data stream to provide real-time responses. For, example it is estimated that one autonomous vehicle (AV) could produce 30 terabytes of data per day. AV will not be on the road before using an effective way to managing its big data and solve latency challenges. Cloud computing failed in the latency challenge, while Fog computing addresses it by moving parts of the computations from the Cloud to the edge of the network near the asset to reduce the latency. This work studies the challenges in data stream processing for DT in a fog environment. The challenges include fog architecture, the necessity of loosely-coupling design, the used virtual machine versus container, the stateful versus stateless operations, the stream processing tools, and live migration between fog nodes. The work also proposes a fog computing architecture and provides a vision of the prerequisites to meet the challenges.