Pré-Publication, Document De Travail (Preprint/Prepublication) Année : 2026

Placement of Distributed Machine Learning Services for AI- and Smart Grid-Enabled IoT Platforms

Résumé

Artificial intelligence, fog computing, and smart grids allow for better energy management and personal data control. However, such heterogeneous platforms complexify service placement due to hardware and network heterogeneity. This paper proposes a distributed machine learning service for fog computing platforms model, and two new algorithms that aim to reduce time and energy consumption. The first algorithm groups services geographically, while the second places services vertically in the topology to reduce network usage. The Batsim simulator have been used to deploy distributed AI applications for measuring their duration and energy usage of the fog architecture. Results show that the first algorithm induces higher time and energy consumption than the second one, which outperforms multiple baseline algorithms for both metrics.

Fichier principal
Vignette du fichier
Placement of Distributed Machine Learning Services for AI.pdf (642.16 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)
licence

Dates et versions

hal-05279358 , version 1 (04-02-2026)

Licence

Identifiants

  • HAL Id : hal-05279358 , version 1

Citer

Jérémie Rodez, Patricia Stolf, Thierry Monteil. Placement of Distributed Machine Learning Services for AI- and Smart Grid-Enabled IoT Platforms. 2026. ⟨hal-05279358⟩
4564 Consultations
145 Téléchargements

Partager

  • More