Least Median of Squares for non-line-of-sight error mitigation in GSM localization

Nowadays, almost everyone has a mobile telephone. This requires GSM localization systems to have greater applicability all of the time. The ldquokiller issuerdquo in this field is the well-known non-line-of-sight (NLOS) error, which hinders localization robustness and accuracy in real scenarios. Many techniques have been developed to deal with this problem, but they usually require prior statistical information or error modeling, or are computationally expensive. In this paper, we present a simple method based on the least median of squares technique to mitigate the NLOS effect. The method we propose is widely used in artificial vision applications and manages to overcome NLOS error effects, yielding higher location accuracy and robustness than other techniques. It can successfully deal with more than half of the available corrupted measurements- no matter how severely-without any previous statistical knowledge-corrupted measurements are not identifiable-and with reduced computation load.