Research On The Wireless Localization Algorithms Under NLOS/LOS Hybrid Environment

In real-world environment, since all kinds of obstacles block the direct way between mobile stations and base stations, thus, the signal sent by mobile station has to be reflected or diffracted until received by the base stations, leading to a much longer path than the direct way. In additional, there is also a large bias between the measured angle of arrival and the true one. If the measured distances and AOAs are used by the traditional localization algorithms to localize the MS, the performance can be very poor. Thus, this thesis pays attention on the methods to mitigate the NLOS error by studying the properties of the NLOS transmission and its influence on the localization algorithms.Firstly, two kinds of NLOS identification algorithms are proposed based on the fact that some parameters under the LOS environment are found to be different from those under the NLOS environments. The threshold can be deduced by studying the property of the residual in the LOS environments. Then, the LOS BSs can be identified from the NLOS BSs by comparing the residual with the threshold, and the position of MS is determined by the LOS BSs. Simulation results show that the proposed algorithms work well in identifying the NLOS BS, and the localization accuracy is also high.Since there is trend that NLOS error may increases with the direct path between the MS and BSs according to the existing references and reports. Two kinds of TOA rational fraction transformation based NLOS mitigation algorithms have been proposed. By rational fraction transformation of the measured TO As, the NLOS error will be reduced significantly. Simulation results show that the proposed algorithms outperform the traditional algorithms in distance dependent models.Four localization algorithms based on optimization theory are proposed refer to the existing algorithms. The object function is set to minimize all kinds of residuals, and the constraints are deduced from the relationship between the MS and BSs, and the position will be estimated by searching the feasible region where the true MS located at. Simulation results show that the proposed localization algorithms can mitigate the NLOS error significantly, especially with the existence of LOS BSs.Finally, two kinds of residual weighting algorithms are proposed by studying the property of the residuals under both NLOS and LOS environments. The intermediate estimations gain different weights according to the values of their residuals. Since the accurate estimations gain much larger weights due to their small residuals, while the inaccurate estimations obtain smaller weights in opposite, thus, the final estimation can be very accurate when the LOS BSs are enough. And simulation results show that the proposed weighting algorithms outperm the traditional algorithms.