A User Application-Based Access Point Selection Algorithm for Dense WLANs
Munsuk Kim, Ye Na Kim, Sukyoung Lee, Nada T. Golmie, SeungSeob Lee
The current commercial access point (AP) selection schemes are mostly based on signal strength, but perform poorly in many situations because they ignore the actual load distributions among the APs. To address this shortcoming, a number of alternative schemes that consider the current utilizations and achievable throughputs of APs have been proposed. However, most of them introduce additional latency or battery consumption overheads while communicating with either a certain centralized server or the APs. In this paper, we propose a user-centric AP selection (UCAS) scheme, in which a mobile phone automatically measures and stores the achievable throughputs of the APs that the user utilizes in frequently visited places, and employs this learning information when the user revisits the places, to choose an appropriate AP. We effectively mitigate the learning overheads by utilizing the throughputs of active applications running in the users mobile phone, and improve the learning accuracy by considering the traffic characteristics of the applications classified based on the supervised machine-learning technique. Using a measurement study of APs in real dense wireless local area network (WLAN) environments, we show that our UCAS scheme chooses an AP with a better achievable throughput, over the previous AP selection approaches.
, , Y.
, Lee, S.
, Golmie, N.
and Lee, S.
A User Application-Based Access Point Selection Algorithm for Dense WLANs, PLoS One, [online], https://doi.org/10.1371/journal.pone.0210738
(Accessed June 25, 2021)