Existing semantic models offer limited support for representing distributed AI workflows that span heterogeneous edge, fog and cloud environments, causing AI processes and resources to be described with incompatible representations and hindering interoperability, orchestration and reuse.
This paper introduces a SAREF‑compliant ontology that unifies the representation of distributed AI workflows across the edge‑fog‑cloud continuum. Building on the SAREF4SYST ontology, we add concepts for AI pipelines, executable AI jobs, computational resources, deployment constraints and communication relationships, thereby providing a single semantic model for both AI workflows and heterogeneous computing infrastructures.
The ontology enables semantic interoperability, automated reasoning and resource‑aware orchestration while staying fully aligned with the ETSI SAREF ecosystem. It is evaluated through proof‑of‑concept smart‑grid energy service orchestration scenarios and validated with competency questions that test workflow deployment, execution reasoning and workload adaptation. All competency questions are successfully answered using SPARQL queries and semantic reasoning. Experiments show deployment success rates of 90‑100% and average orchestration decision times below 80 ms across heterogeneous edge‑fog‑cloud environments, demonstrating the ontology’s effectiveness for distributed AI orchestration.
Blogger's Review: By preserving compatibility with the SAREF ecosystem and delivering a comprehensive semantic description of AI workflows, this ontology markedly improves cross‑layer resource orchestrability and reasoning efficiency, offering solid semantic support for real‑world industrial AI deployments.