The recent evolution of the traditional cloud computing vision towards the so-called edge computing has stimulated the emergence of new network paradigms, such as drone-assisted mobile networks. Here, drones acting as mobile edge servers provide storage and processing services so as to offload data and support ubiquitous connectivity. In such a scenario, inter-user interference represents a critical feature, as it can heavily hinder the network performance, due to the massive user densification of modern 5G-and-beyond networks. This work deals with the interference problem and the resulting network dynamics under a theoretical perspective by presenting a fully-distributed game-theoretical framework. More in detail, the interactions between users and servers are modeled as a marketplace scenario, where servers (i.e., drones), behaving as sellers, offer network functions, while users buy the offered services and, accordingly, interact with servers. The mathematical characterization of the marketplace interactions is achieved by employing a two-stage Stackelberg game in which a unique equilibrium and convergence point can be identified. Moreover, we show that the use of a reinforcement learning algorithm allows to achieve this equilibrium in a practical setting. Our numerical evaluation investigates the dynamics of multiple 5G use cases (namely eMBB, URLLC and a hybrid tradeoff case), each characterized by heterogeneous requirements and diverse KPIs which lead to very dissimilar behavior in terms of requested bandwidth and user distribution at the equilibrium. The results show how our game theoretical framework achieves a reduction of up to 60% in the network interference in a fully-distributed fashion, where each actor in the network can act in an autonomous and independent way to reach the convergence point.

A marketplace model for drone-assisted edge computing in 5G scenarios

Fabio Busacca;Laura Galluccio;Sergio Palazzo
2022-01-01

Abstract

The recent evolution of the traditional cloud computing vision towards the so-called edge computing has stimulated the emergence of new network paradigms, such as drone-assisted mobile networks. Here, drones acting as mobile edge servers provide storage and processing services so as to offload data and support ubiquitous connectivity. In such a scenario, inter-user interference represents a critical feature, as it can heavily hinder the network performance, due to the massive user densification of modern 5G-and-beyond networks. This work deals with the interference problem and the resulting network dynamics under a theoretical perspective by presenting a fully-distributed game-theoretical framework. More in detail, the interactions between users and servers are modeled as a marketplace scenario, where servers (i.e., drones), behaving as sellers, offer network functions, while users buy the offered services and, accordingly, interact with servers. The mathematical characterization of the marketplace interactions is achieved by employing a two-stage Stackelberg game in which a unique equilibrium and convergence point can be identified. Moreover, we show that the use of a reinforcement learning algorithm allows to achieve this equilibrium in a practical setting. Our numerical evaluation investigates the dynamics of multiple 5G use cases (namely eMBB, URLLC and a hybrid tradeoff case), each characterized by heterogeneous requirements and diverse KPIs which lead to very dissimilar behavior in terms of requested bandwidth and user distribution at the equilibrium. The results show how our game theoretical framework achieves a reduction of up to 60% in the network interference in a fully-distributed fashion, where each actor in the network can act in an autonomous and independent way to reach the convergence point.
2022
Drones, Edge computing, Marketplace, Game theory
File in questo prodotto:
File Dimensione Formato  
Marketplace-Model.pdf

solo gestori archivio

Tipologia: Versione Editoriale (PDF)
Licenza: NON PUBBLICO - Accesso privato/ristretto
Dimensione 2.08 MB
Formato Adobe PDF
2.08 MB Adobe PDF   Visualizza/Apri

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/20.500.11769/550481
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact