Research On Wireless Localization In LOS/NLOS Environments

The development of wireless location techniques has become more and more important in modern society. It is one of the necessary techniques to safeguard the state security and to support a variety of military applications. At the same time, the civilian application of wireless localization has made large influence on people's life. All the aforementioned applications, on one hand, have made the popularization of wireless location techniques; on the other hand, large measurement errors cause standard localization techniques to perform poorly in various complicated environments, and it is necessary to develop new methods to improve the location accuracy.This dissertation focuses on the location techniques in line of sight (LOS) and non-line of sight (NLOS) environments. The main works are included as the following:While the classical multidimensional scaling (MDS) algorithm had been claimed to be infeasible, MDS based algorithms, including modified MDS and subspace approach, have been applied to the range measurements based wireless localization. The classical MDS algorithm is first proved that it is feasible for the wireless localization. Then all MDS-based algorithms are shown that they can be unified under the same framework and they are not optimal according to the Gaussian-Markov theorem. Based on the unified framework, the weighted MDS analysis is applied to obtain the optimum performance.Given time difference of arrival (TDOA) measurements, the MDS algorithm is extended for wireless location using the method as two-step weighted least squares (TWLS) algorithm. The performance of the MDS approach can attain the Cramer-Rao lower bound (CRLB) at moderate noise level, and is robust for large noise at a certain extent.Since the MDS based approaches are sensitive to the outliers, a least absolute deviation (LAD) based cost function of MDS algorithm is presented. The semidefinite programming (SDP) technique is applied for finding the solution. Simulation results show that this method is more robust than classical MDS. In NLOS environment, conventional hybrid TOA (Time of Arrival)/AOA (Angle of Arrival) location techniques would suffer from large errors. After the analysis of the existing scatterers models, an enhanced hybrid TOA/AOA location technique is first presented using the angular spread characteristic. This technique can improve the performance of hybrid TOA/AOA technique in NLOS environment. Then, based on the assumption that the scatterers are single reflection and geometrically fixed during the measurement period, a new approach using scatterer information is proposed for wireless location. Unlike conventional methods, the performance of the proposed approach can be improved by increasing the measurement accuracy. Moreover, the proposed approach is not sensitive to the radius of the circle which the scatterers lie in and can be further improved using more separated multipath signals. If the joint estimation of angle of arrival (AOA) and Doppler frequency of arriving rays can be obtained, an elliptical constraint based method can be utilized to reconstruct the direction of LOS arrival. Numerical examples show that the proposed method can reconstruct the direction of LOS arrival with higher probability and accuracy in serious NLOS environment.Since the root mean square delay spread characterizes the multipath dispersion of radio channel and is dependent on the environment, a root mean square delay spread based equalization scheme is present for wireless localization in NLOS environments. This scheme can obtain better performance than other methods such as constrained least sqauares method and residual weighting algorithm. Without any redundant information about the target position and other measurements, a neighboring base stations (BSs) assisted wireless location method is derived in NLOS environments. It can be reasonably assumed that the home BS is LOS BS and that all other neighboring BSs can be NLOS since they are farther away from the MS. Then the estimator makes use of TOA measurement at home BS to provide range information and attempts to search the azimuth of target utilizing the range measurements of neighboring BSs. The two parameters are then combined to find the target position. This method can improve the location accuracy in serious NLOS environment.Using the motion dynamics and the measurements made at several sampling time periods, a dynamic MDS algorithm is proposed for tracking mobiles. The matrix constructed for employing MDS is extended, and the target position is calculated utilizing the subspace method. In LOS environment, the proposed dynamic MDS has the similar performance as Kalman filtering; however, it outperforms Kalman filtering, conventional MDS methods and TWLS algorithm in NLOS environment.