Experiment-driven platform for link quality estimation in IEEE 802.11 WLANs

Experimental link quality assessment of recent Wi-Fi networks remains a challenge due to the rapid development of the Wi-Fi technology, the lack of availability of public datasets, and the difficulty to build such datasets. This paper addresses all three issues by first providing a publicly-available dataset using a custom-made Wi-Fi 5 experimental testbed. We then present an open-source framework for estimating the Frame Delivery Ratio (FDR) of a Wi-Fi link using the experimental data. The proposed solution relies on a small number of input features to build an estimation model of high accuracy, with an R2 coefficient of 0.89 and a mean absolute error of 0.06.