Planning of Edge Computing for Factories

Information technology (IT) infrastructure is essential to a factory and significantly impacts production efficiency. The structure of such an IT infrastructure (cloud, fog or edge computing based) is crucial for the total investment, the resulting environmental impact, and the contribution to the value creation in the factory. This work presents a concept for planning edge computing in a factory environment. Decision-makers, who do not need expert knowledge of IT systems, are guided through a six-phase planning process that supports them in determining a suitable IT infrastructure and its architecture for a given factory. The resulting IT infrastructure includes the determined selection, placement, and connection information of IT resources and software applications. Computing paradigms significantly impact the IT infrastructure, mandating the specification of an appropriate edge computing model presented in three parts. The first part is the edge-cloud continuum combining the understanding of the Internet of Things (IoT), edge, fog, and cloud computing. The continuum specifies the placement of IT resources based on network boundaries. The second part is the edge level that combines the edge-cloud continuum with physical placement within a factory's horizontal and vertical dimensions. In the third part, the five edge criteria - connectivity, latency, ownership, autonomy, and volume - are proposed, enabling an automatic determination of the edge level and a recommendation for the use of edge, fog, and cloud computing at different levels of consideration. The six-phase planning framework, as the core of this work, allows the operationalization of the edge computing model. In the first three phases, the user models the value-added data flow within a factory, capturing factory goals, available data sources, and data processing applications at different stages of development. In the fourth phase, a scoring model is used to calculate a vertical placement recommendation for each application based on edge criteria allowing the automatic determination of an edge level representing the vertical placement options. In the fifth phase, all collected IT resource requirements and edge-level specifications are matched with the available IT resource database. An optimization algorithm calculates the best application and resource configurations based on cost, environmental impact, and topology configuration. In the sixth and final phase, the user evaluates the resulting topologies using domain-specific, customized visualization methods to make an informed decision about the resulting IT infrastructure. A software prototype was developed that facilitates the subsequent application of the developed planning framework in three case studies: a learning factory environment, an automotive factory, and an electronic component manufacturing factory. The three case studies differ in their area of origin, planning scope, information accessibility, and production volumes, reflecting the heterogeneity of the current factory landscape.