Robust localization in sensor networkswith iterative majorization techniques

Self localization in sensor networks with measurements that include outliers is an important problem. E.g., distance measurements based on non-line-of-sight observations can be quite wrong. If not handled properly, such outliers can greatly influence the positioning accuracy. To achieve robustness we consider positioning with Huber estimators. The Huber cost function interpolates between the lscr₁ and the lscr₂ norms. The minimization of the Huber cost function can be efficiently obtained via iterative majorization techniques, with the advantageous property of guaranteed convergence to a local minimum.