Industrial cyber-physical systems (iCPS) are central to smart manufacturing and Industry 4.0. By bringing physical processes together with computation, communication, and control, they make it possible for factories and machines to behave more intelligently and in real time. And as the pressure grows for systems that are more scalable, more capable, and faster, iCPS architectures are increasingly shifting toward distributed cloud–edge–device paradigms. In this paper, we provide a systematic survey of that architectural transition. We begin by introducing what iCPS are, along with their key components and a general conceptual model. Then we review and compare traditional hierarchical architectures, focusing on where they tend to break down—openness, interoperability, latency, and system integration. After that, we analyze and compare emerging cloud–edge–device architectures across multiple dimensions, including architectural structure, scalability, latency, and security. Finally, we discuss the main challenges and open research issues, aiming to align with current industrial needs and to offer useful guidance for designing next generation iCPS architectures.



