The growth of edge computing and Internet of Things platforms is reshaping distributed systems, introducing challenges related to device heterogeneity, limited resources, and dynamic environments. Existing orchestration approaches largely rely on centralized cloud infrastructures and static workflows. This paper presents LADbridge, a system that enables natural language–driven composition and invocation of distributed APIs in edge and IoT ecosystems. The architecture shifts LLM-based reasoning from the cloud to edge nodes, supporting dynamic service discovery, semantic matching, and language-guided orchestration of heterogeneous endpoints. Experimental results show that LADbridge accurately interprets user requests and generates executable service plans across diverse service environments while maintaining practical latency. Its lightweight design allows deployment on resource-constrained edge devices without requiring specialized hardware.

LADbridge: LAnguage-Driven composition of APIs on Distributed Edge Devices

Francesco Cosimo Mazzitelli;Zimeo Eugenio
2026-01-01

Abstract

The growth of edge computing and Internet of Things platforms is reshaping distributed systems, introducing challenges related to device heterogeneity, limited resources, and dynamic environments. Existing orchestration approaches largely rely on centralized cloud infrastructures and static workflows. This paper presents LADbridge, a system that enables natural language–driven composition and invocation of distributed APIs in edge and IoT ecosystems. The architecture shifts LLM-based reasoning from the cloud to edge nodes, supporting dynamic service discovery, semantic matching, and language-guided orchestration of heterogeneous endpoints. Experimental results show that LADbridge accurately interprets user requests and generates executable service plans across diverse service environments while maintaining practical latency. Its lightweight design allows deployment on resource-constrained edge devices without requiring specialized hardware.
2026
979-8-3195-4259-5
Modeling , Large language models , Planning , Cognition , Cognitive systems , Natural languages , Printing , Retrieval augmented generation , Internet of Things , Architecture
File in questo prodotto:
Non ci sono file associati a questo prodotto.

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12070/78126
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact