Accuracy enhancement of Wi-Fi positioning by ranging in dense UWB networks

Abstract

This study proposed a fusion-based approach of Wi-Fi fingerprint positioning with relatively accurate UWB ranging in dense wireless networks (DWNs) to optimize indoor localization accuracy. Such an optimization was achieved through a proposed objective or “loss” function that minimizes the total positioning errors of the nodes composing a virtual geometric structure with the internode distances preserved during optimization in the DWN. Theoretical analysis and simulation results of DWNs up to 100 nodes revealed that, with at least 6 extra measured distances among as few as 4 participating nodes including the device-under-targeting (DUT) itself, the resulting localization accuracy improvement was proportional to the square root of the number of participating nodes towards the limit of the ranging error. Per experiment outcomes of a 10-node UWB DWN, our Wi-Fi/UWB ensemble reduced the baseline Wi-Fi K-nearest neighbors (KNN) fingerprint localization error by up to 69.8 %, further validating our theoretical and simulation results.