Because the Edge Computing (EC) paradigm allows processing of vast amounts of data in proximity to the respective source, latency and quantity constraints are no longer a limiting factor. That enables the development of novel data-driven applications and the extension of the solutions space for valueadded services in production. The complexity and diversity of factories, combined with the continuing discovery of new datadriven solutions, poses a challenge for practitioners to thoroughly determine where, which, and how data should be processed. This, however, is crucial for deciding how and whether to invest in EC. This paper proposes a multiphase concept for the systematic assessment of whether and where EC is most beneficial in a given production environment. It is comprised of human and machine interpretable functions. Combining multiple functions leads to a data-driven solution, which forms links between the data sources (assets) of a production environment and the desired outcome (goals). Four main criteria for EC are derived to enable the exposure of areas with increased EC potential, forming the baseline for a scoring system. The concept is designed so that its application is feasible within an industrial context. First analyses show the prospect of the approach and suggest potential benefits for providing practical implementation guidance.
Introducing distributed computing paradigms to the manufacturing domain increases the difficulty of designing and planning an appropriate IT infrastructure. This paper proposes a model and solution approach addressing the conjoint application and IT resource placement problem in a factory context. Instead of aiming to create an exact model, resource requirements and capabilities are simplified, focusing on usability in the planning and design phase for industrial use cases. Three objective functions are implemented: minimizing overall cost, environmental impact, and the number of devices. The implications of edge and fog computing are considered in a multi-layer model by introducing five resource placement levels ranging from on-device, within the production system, within the production section, within the factory (on-premise), to the cloud (off-premise). The model is implemented using the open-source modeling language Pyomo. The solver SCIP is used to solve the NP-hard integer programming problem. For the evaluation of the optimization implementation a benchmark is created using a sample set of scenarios varying the number of possible placement locations, applications, and the distribution of assigned edge recommendations. The resulting execution times demonstrate the viability of the proposed approach for small (100 applications; 100 locations) and large (1000 applications, 1000 scenarios) instances. A case study for a section of a factory producing electronic components demonstrates the practical application of the proposed approach.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.