Weighted least squares (WLS) is an extension of ordinary least squares estimation in which each observation or measurement is assigned a weight reflecting its reliability or uncertainty, so that more trustworthy data points exert greater influence on the fitted solution. In indoor positioning and CSI-based sensing, WLS is used to combine heterogeneous signal measurements or fuse estimates from multiple sources with differing noise levels, improving localization accuracy over unweighted approaches. Key variants include iteratively reweighted least squares (IRLS), where weights are updated across successive iterations to achieve robustness against outliers, and recursive weighted least squares, which updates estimates sequentially as new measurements arrive, making it well suited to real-time tracking scenarios.
Dictionary term