An LLM-based approach to maximize the :effectiveness of Industrial Internet of :Things implementations through the :selection of architecture
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
The selection of the most suitable architectural pattern for Industrial Internet of Things (IIoT) applications within the Industry 4.0 paradigm presents a critical challenge, as organizations grapple with a variety of contexts, sophisticated architecture, complex deployment, stringent performance requirements, and evolving security concerns. This thesis develops and validates a transparent, LLM-based approach and a decision framework for pattern selection. The approach formalizes decision rules that are used to select appropriate IIoT architecture for a provided use-case. Evidence was assembled through a qualitative synthesis of approximately 5,000 publications that were programmatically collected, deduplicated, ranked for relevance, and mined for key metrics describing pattern attributes and reported contexts. Domain experts in IoT were asked in a survey to subsequently specify use-case descriptions and decide how much weight to give each evaluation criterion, which were predefined. This expert mapping constitutes the ground truth that instantiates the theoretical model within the decision framework. A large language model (LLM) was deployed as a tool for context specialization, transforming generic architectural knowledge into case-specific recommendation. Conditioned on the expert-validated pattern–use-case mapping, the model in-gests previously unseen architecture/use-case descriptions and returns a rank-ordered set of suitable architectural patterns with accompanying rationales. The LLM operationalizes and generalizes expert knowledge to novel contexts. The findings indicate that pattern suitability is highly context specific, and that a structured, data-driven approach significantly enhances decision confidence. By formalizing pattern–usecase mappings and incorporating expert judgments, the research delivers a twofold contribution. On the one hand, it establishes a approach for making sense of IIoT architectural choices. On the other hand, it provides a practical, data-driven approach that system architects can apply immediately. The primary users of this approach are system architects, solution engineers, and industrial IT decision makers in manufacturing and industry related sectors. The secondary users include researchers and lecturers who can employ the model as a teaching and benchmarking instrument. Future studies might look at systems that adjust in real-time and examine their pros and cons across different industrial settings.
Fields of Study & Subjects
Digital Transformation in Industry
IoT and Edge/Fog Computing
Internet of Things and AI
Publication details
- Journal / source
- FH JOANNEUM ePUB
- Publication date
- 1 January 2026
- Language
- EN
- Document type
- Thesis