NLOS Measurement Identification for Mobile Positioning in Wireless Cellular Systems

Mobile positioning is the very first step toward location-dependent services in wireless cellular systems. When mobile positioning is carried out by trilateration or triangulation, its accuracy is dominantly limited by NLOS (non-line-of-sight) propagation errors. Therefore, the identification of NLOS measurements is important. In this paper, we address this issue by first assuming a statistic model for the NLOS error. Then, by estimating the relevant parameters in this model and comparing them to certain thresholds, we are able to distinguish the NLOS measurements from the LOS (line-of-sight) ones. By simulation, the proposed algorithm is tested and compared to an existing method found in the literature. In addition to the conventional performance index for evaluating the accuracy of identification in terms of detection probabilities and false alarm probabilities, a new performance index, which takes into account the entire ROC (receiver operating characteristics) curve, is proposed in this paper. The evaluation based on the newly proposed index shows the same trends in performance regarding the various identification methods as the conventional index.