User-AP Association Management in Software-Defined WLANs

Suzan Bayhan, Estefanía Coronado, Roberto Riggio, Anatolij Zubow

Research output: Contribution to journalArticleAcademicpeer-review

5 Citations (Scopus)
1 Downloads (Pure)

Abstract

Despite the planned operation of enterprise wireless local area networks (WLANs), they still experience unsatisfactory performance due to several inefficiencies. One of the major issues is the so-called sticky user problem, in which users remain connected to an access point (AP) until the signal quality becomes too weak. In this paper, we leverage software-defined networking (SDN) to propose a user association solution for WLANs aiming to mitigate such inefficiencies, thus improving resource utilization. As it is a computationally hard problem, we also design various low-complexity user-AP association schemes that consider not only signal quality but also AP loads and minimum quality requirements for user traffic. Moreover, to provide simultaneous content distribution in a sustainable mode, we propose exploiting link-layer multicasting to decide on user-AP associations. Our analysis via simulations and experimentation on an open-source testbed shows that considering user-AP association jointly with multicast delivery leads to a significant performance increase over the default client-driven approach: the median throughput is 11\times higher when all users request the same content and the achieved improvement decreases to 68% for 100 contents. Moreover, due to more efficient use of the airtime, unicast users achieve higher throughput if multicast delivery is exploited.

Original languageEnglish
Article number9115672
Pages (from-to)1838-1852
Number of pages15
JournalIEEE transactions on network and service management
Volume17
Issue number3
Early online date12 Jun 2020
DOIs
Publication statusPublished - Sep 2020

Keywords

  • 22/2 OA procedure

Fingerprint

Dive into the research topics of 'User-AP Association Management in Software-Defined WLANs'. Together they form a unique fingerprint.

Cite this