Distributed Controller-Switch Assignment in 5G Networks

Software defined networking (SDN) is a promising technology in fifth generation wireless networks (5G) where due to the adoption of a centralized SDN-controller, resources such as processing and storage, can be utilized in an optimal manner. Although SDN was first considered with a logically central...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE eTransactions on network and service management 2021-06, Vol.18 (2), p.2193-2207
Hauptverfasser: Tohidi, Ehsan, Parsaeefard, Saeedeh, Hemmati, Ali Akbar, Maddah-Ali, Mohammad Ali, Khalaj, Babak Hossein, Leon-Garcia, Alberto
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext bestellen
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Software defined networking (SDN) is a promising technology in fifth generation wireless networks (5G) where due to the adoption of a centralized SDN-controller, resources such as processing and storage, can be utilized in an optimal manner. Although SDN was first considered with a logically centralized controller, due to delay, reliability, and scalability challenges, moving towards multiple distributed controllers is inevitable. In distributed control schemes, an assignment that associates a controller with each switch leads to three challenges of (1) Computational complexity, since the assignment is an NP-hard problem, (2) Resource and energy efficiency, to obtain an assignment with the lowest number of controllers in order to reduce resource and energy consumption, and (3) Dynamicity, where a dynamic approach of assignment is required to adapt to the network's traffic changes. In this paper, we investigate the controller-switch assignment problem given the aforementioned challenges, and propose efficient algorithms for static and dynamic scenarios, that even achieve quantitative optimality guarantees in special cases. As shown through simulations, the proposed lower complexity algorithms not only outperform earlier works but also approach the performance of exhaustive search schemes, in some scenarios.
ISSN:1932-4537
1932-4537
DOI:10.1109/TNSM.2021.3068979