Section: New Results

Passive Wi-Fi Link Capacity Estimation on Commodity Access Points

Participants: Diego Neves da Hora, Karel Van Doorselaer (Technicolor), Koen Van Oost (Technicolor), Renata Teixeira, Christophe Diot (Safran)

We propose an algorithm to estimate the link capacity based on passive metrics from APs, which is ready to be deployed at scale. We show that it is possible to estimate the link capacity per PHY rate based on a limited set of parameters related to the particular AP instance. Then, we extend the initial model to estimate the link capacity when the PHY rate varies. We measured the link capacity in different link quality conditions and found that more than $90\%$ of the estimations present error below $15\%$ without prior parameter tuning, and more than $95\%$ present estimation error below $5\%$ with appropriate parameter tuning using fixed PHY rate tests.