Research On Fundamental Theory And Method For Non-Line-Of-Sight Target Positioning And Tracking

The location estimation of a mobile station (MS) in wireless communication systems has gained considerable attention over the past decade, especially since the Federal Communication Commission (FCC) passed a mandate requiring cellular providers to generate accurate location estimates for Enhanced-911 (E-911) services. One of the main problems facing accurate location in wireless communication systems is non-line-of-sight (NLOS) propagation, since line-of-sight (LOS) propagation may not exist. This has boosted the research in the field of wireless location in NLOS environments.According to difference channel environments, the statuses of MS and prior information, the location estimation methods in wireless communication systems, especially the methods for mitigating the influence of the NLOS propagation, are studies in this dissertation. The primary contributions are summarized as follows:1. In order to locate MS in LOS environments, we derive the Cramer-Rao Lower Bounds (CRLB) of four localization techniques, including the time of arrival (TOA), time difference of arrival (TDOA), time of arrival/received signal strength (TOA/RSS) and time difference of arrival/received signal strength (TDOA/RSS) based methods. Then two hybrid location methods (TOA/RSS and TDOA/RSS) based on two steps least square method are proposed. Simulation results show that the proposed hybrid algorithms outperform TOA and TDOA based methods.2. In order to locate MS in LOS/NLOS mix environments, we study NLOS identification algorithms. First, we analyze the performance of the TOA and TDOA based methods. The analytical results show that we should develop NLOS identification algorithms. Then a novel NLOS identification algorithm based on overlap Area is proposed to deal with TOA measurements. Comparison to other TOA NLOS identification algorithms, the proposed method can identify NLOS BS with more accuracy and less computation complexity, and fit for moveless MS. Finally, a novel NLOS identification algorithm based on Chi-square distribution is proposed to deal with TDOA measurements. Comparison to other TDOA NLOS identification algorithms, the proposed method can identify the number of NLOS BSs.3.A novel algorithm based on quadratic reflection assumption and multi-path information is proposed to obtain precise mobile location in where there exist multi-path signals and little scatters. The CRLB is also derived. Comparison to other scatters information based location methods, the proposed algorithm breaks through a single reflection assumption.4.A novel algorithm based on robust objective function is proposed to obtain precise mobile location in where the number of LOS BSs or multi-path signals is insufficient. Traditional location algorithms are based on mean-squares objective function and it loses optimality in NLOS environments. The proposed algorithm replace mean-squares objective function with robust objective function, it does not require the prior knowledge of the NLOS error distribution and can give a closed-form solution.5.We study the learning location methods with prior information. First, a novel location algorithm based on Kriging method is proposed to mitigate NLOS errors when outliers not exist in the training data set. Simulation results show that the proposed algorithm comes close to meeting CRLB. Then we extend least squares support vector machine (LS-SVM) algorithm to robust objective functions, and use this algorithm to solve mobile location problem. Simulation results show that the proposed algorithms can effectively mitigate outliers.6.We study the NLOS tracing algorithms. First, a NLOS tracing algorithm based on enhanced quadrant constraint is proposed. Then a robust LMS algorithm is proposed to filter measurements. Finally, an adaptive NLOS tracing algorithm is proposed. Although the proposed adaptive NLOS tracing algorithm will add the computation complexity, Simulation results show that it outperforms the other two.