Load-aware Channel Selection for 802.11 WLANs with Limited Measurement

TL;DR

This paper introduces a learning-based method using Gaussian Process Regression to efficiently identify the least loaded Wi-Fi channels with limited measurements, reducing measurement time by up to 46%.

Contribution

It presents a novel load-aware channel selection approach that minimizes measurement overhead using Gaussian Process Regression, improving efficiency in WLANs.

Findings

Reduced load measurement time by up to 46%.

Accurately tracks channel load with limited measurements.

Effective in experimental WLAN scenarios.

Abstract

It has been known that load unaware channel selection in 802.11 networks results in high level interference, and can significantly reduce the network throughput. In current implementation, the only way to determine the traffic load on a channel is to measure that channel for a certain duration of time. Therefore, in order to find the best channel with the minimum load all channels have to be measured, which is costly and can cause unacceptable communication interruptions between the AP and the stations. In this paper, we propose a learning based approach which aims to find the channel with the minimum load by measuring only limited number of channels. Our method uses Gaussian Process Regressing to accurately track the traffic load on each channel based on the previous measured load. We confirm the performance of our algorithm by using experimental data, and show that the time consumed for the load measurement can be reduced up to 46% compared to the case where all channels are monitored.