MOM-BASED NLOS IDENTIFICATION AND MOBILE POSITIONING IN WIRELESS CELLULAR SYSTEMS

摘要

The accuracy of mobile positioning, which is usually carried out by trilateration or/and triangulation, is dominantly limited by NLOS (non-line-of-sight) propagation errors. Therefore, the identification of NLOS measurements and the reduction of NLOS error effect is important. In this paper, we address this issue by first assuming the NLOS errors to be exponentially distributed. Then, by estimating the relevant parameters in this distribution through MOM (method of moments), we are able to distinguish the NLOS measurements from the LOS (line-of-sight) ones. In the meantime, the estimates of those parameters also indicate how large the NLOS error is (in expectation). By taking advantage of this information, much-improved positioning results can be obtained. The proposed algorithm for NLOS identification is tested and compared to existing methods, and the results are evaluated by a newly proposed performance index. This performance index takes into account the entire ROC (receiver operating characteristics) curve, instead of just isolated points on it. We also compare the accuracy of our mobile positioning scheme to other existing methods. Experimental results show that our scheme performs better in most cases, both in NLOS identification and in positioning accuracy.