Distributed Learning (DL) is a promising suite of techniques enabling progressive Internet of Things (IoT) systems with built-in intelligence. A diverse range of DL methods with specific performance, characteristics, and requirements are available for IoT users. However, implementing multiple DL techniques to serve various user groups and applications in traditional IoT settings can be difficult due to their specific needs and resource constraints. In light of emerging network paradigms with highly distributed computing and communication resources, such as joint terrestrial/non-terrestrial networks, it is essential to leverage available resources while simultaneously implementing various DL methods. Network Slicing (NS) can be highly beneficial in this regard, as it enables the implementation of heterogeneous services over a shared network infrastructure in a virtualized environment. In this study, we introduce NS, then propose a framework that enables DL-as-a-Service (DLaaS) over a distributed IoT platform using a multi-layer hierarchical structure using satellites at the highest level. This framework allows for the proactive deployment of distributed IoT systems and facilitates the management of heterogeneous services with varying requirements. Then, we explore the advantages and challenges of the proposed hierarchical structure, along with some solutions to address those issues, such as split learning (SL) and differential privacy (DP).

David Naseh, S.S.S. (2023). Multi-Layer Network Sliced Distributed Learning for Satellite IoT. IEEE Future Networks.

Multi-Layer Network Sliced Distributed Learning for Satellite IoT

David Naseh
;
Swapnil Sadashiv Shinde;Daniele Tarchi
2023

Abstract

Distributed Learning (DL) is a promising suite of techniques enabling progressive Internet of Things (IoT) systems with built-in intelligence. A diverse range of DL methods with specific performance, characteristics, and requirements are available for IoT users. However, implementing multiple DL techniques to serve various user groups and applications in traditional IoT settings can be difficult due to their specific needs and resource constraints. In light of emerging network paradigms with highly distributed computing and communication resources, such as joint terrestrial/non-terrestrial networks, it is essential to leverage available resources while simultaneously implementing various DL methods. Network Slicing (NS) can be highly beneficial in this regard, as it enables the implementation of heterogeneous services over a shared network infrastructure in a virtualized environment. In this study, we introduce NS, then propose a framework that enables DL-as-a-Service (DLaaS) over a distributed IoT platform using a multi-layer hierarchical structure using satellites at the highest level. This framework allows for the proactive deployment of distributed IoT systems and facilitates the management of heterogeneous services with varying requirements. Then, we explore the advantages and challenges of the proposed hierarchical structure, along with some solutions to address those issues, such as split learning (SL) and differential privacy (DP).
2023
David Naseh, S.S.S. (2023). Multi-Layer Network Sliced Distributed Learning for Satellite IoT. IEEE Future Networks.
David Naseh, Swapnil Sadashiv Shinde, Daniele Tarchi
File in questo prodotto:
Eventuali allegati, non sono esposti

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/11585/963833
 Attenzione

Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo

Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact