WiCo: Robust Indoor Localization via Spectrum Confidence Estimation

The past decade has witnessed emerging applications of achieving indoor localization using WiFi by estimating the Angle of Arrival (AoA). While various algorithms have been proposed, their performance in practical indoor environment is still limited. An important limitation lies in the fact that the confidences of AoA estimations from different devices are actually unequal, which has not been considered nor addressed. In this paper, we propose WiCo, a confidence-aware localization framework by analyzing the confidence of the spatial spectrum. Specifically, we first evaluate the unequal confidence caused by different beamwidth and multipath on different APs. Then, we propose a normalized distribution confidence and a full reference confidence to quantify the reliability of spatial spectrum on different devices. Finally, we resolve the unequal confidence factor caused by geometric deployment through re-weighting and perform localization. Extensive real-world experiments demonstrate that WiCo could reduce the median localization error by 43.8%.