SAREF-based Ontology for Distributed AI Workflows across the Edge-Fog-Cloud Continuum
Proposes a SAREF-compliant ontology extending SAREF4SYST to model distributed AI workflows across heterogeneous edge, fog, and cloud environments Introduces concepts for AI pipelines, executable AI jobs, computational resources, deployment constraints, and communication relationships into a unified semantic model Enables semantic interoperability, automated reasoning, and resource-aware orchestration of distributed AI applications within the ETSI SAREF ecosystem Evaluated on smart grid energy se
Analysis
TL;DR
- Proposes a SAREF-compliant ontology extending SAREF4SYST to model distributed AI workflows across heterogeneous edge, fog, and cloud environments
- Introduces concepts for AI pipelines, executable AI jobs, computational resources, deployment constraints, and communication relationships into a unified semantic model
- Enables semantic interoperability, automated reasoning, and resource-aware orchestration of distributed AI applications within the ETSI SAREF ecosystem
- Evaluated on smart grid energy services orchestration scenarios with 90-100% deployment success rates and average orchestration decision times under 80 ms
- All competency questions validated successfully using SPARQL querying and semantic reasoning, demonstrating effective workload adaptation across heterogeneous environments
Why It Matters
This work addresses a critical gap in distributed AI systems: the lack of standardized semantic models for representing and orchestrating AI workflows across heterogeneous computing infrastructures. For AI practitioners building edge-cloud systems, this ontology provides a reusable, interoperable framework that can significantly reduce integration complexity and enable automated resource-aware orchestration.
Technical Details
- Extends the ETSI SAREF4SYST ontology with new concepts: AI pipelines, executable AI jobs, computational resources, deployment constraints, and communication relationships, creating a unified semantic model bridging AI workflows and heterogeneous computing infrastructures
- Fully aligned with the ETSI SAREF ecosystem, ensuring compatibility with existing smart system standards and enabling semantic interoperability across domains
- Validated through competency questions using SPARQL querying and semantic reasoning, covering AI workflow deployment, execution reasoning, and workload adaptation capabilities
- Tested on proof-of-concept smart grid energy services orchestration scenarios, achieving deployment success rates of 90-100% with average orchestration decision times below 80 ms across edge-fog-cloud environments
Industry Insight
- Organizations building distributed AI systems across edge-cloud infrastructures should consider adopting SAREF-based ontological frameworks to standardize workflow representation and reduce interoperability costs
- The sub-80ms orchestration decision times demonstrate that semantic reasoning can be performed efficiently enough for real-time distributed AI deployment, making this approach viable for production edge computing scenarios
- As edge AI adoption grows, semantic interoperability standards like this will become essential infrastructure—early adopters who build on SAREF-aligned ontologies will have a significant advantage in system integration and ecosystem compatibility
Disclaimer: The above content is generated by AI and is for reference only.