Answers are grounded in this paper and cite their sources.
Industrial cyber–physical systems (CPSs) are increasingly required to operate in dynamic, large-scale, and heterogeneous environments while ensuring high availability, resilience, and continuous operation. However, most existing industrial CPS architectures remain strongly centralized, hierarchically dependent, and difficult to adapt to evolving operational demands, making them vulnerable to single points of failure and limited scalability. This paper proposes a distributed architectural approach for industrial CPS based on service-oriented and multi-agent paradigms, explicitly designed to reduce hierarchical dependency and eliminate the need for centralized control. The proposed architecture introduces software instances coupled to physical and virtual resources across multiple edge-computing layers, enabling uniform integration of devices, services, agents, and distributed ledgers. Focused on industrial applications, the architecture integrates a distributed IEC 61499–compliant system based on Function Block Networks. The architecture integrates a distributed management framework composed of three interconnected subsystems: service distribution, failure management, and reputation-based trust evaluation. These subsystems rely on decentralized consensus mechanisms to allocate and replicate services, detect and recover from failures, and, through a reputation mechanism specified at the architectural level to assess the trustworthiness of participating agents. The proposed approach is designed to enable distributed orchestration of industrial services, to enhance fault tolerance through service active replication, and to support resilient decision-making even in the presence of failures or potentially malicious agents. A proof-of-concept on an emulated 10-node heterogeneous cluster demonstrates constraint-aware service allocation and autonomous recovery from a single node crash, while trust dynamics, malicious-agent behavior, ledger performance and large-scale consensus remain to be evaluated. By combining edge computing, multi-agent systems, service-oriented design, and distributed ledger technologies within a standards-compliant framework, this work provides an architectural foundation, supported by initial proof-of-concept validation, toward next-generation autonomous industrial CPSs.
No takes yet. Share an insight, caveat, or question.
Alves et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: