PASSIVE LOCALIZATION IN QUASI-SYNCHRONOUS SENSOR NETWORKS WITH SENSOR UNCERTAINTY AND NON-LINE-OF-SIGHT MEASUREMENTS by Kaichen Guo Thesis Supervisor: Prof. Shaodan Ma Electrical and Computer Engineering Localization has been one of the challenging problems with the increasing applications in radar, tracking and logistics, etc. Basically, target localization can be classified into two categories, i.e., active target localization and passive target localization. For passive localization, target merely acts as a reflector to forward signal from the source to sensors. However, active target can directly transmit signal to sensors. Due to the simplicity of the signal propagation path, active target localization has been well investigated. However, for passive localization, there are only a few results available. With the adoption of Ultra-Wideband (UWB) technology, there are typical four measurements: Time-of-Arrival (TOA), Tine-Difference-of-Arrival (TDOA), Angle-of-Arrival (AOA) and Received-Signal-Strength (RSS). TOA refers to transmission time between transceivers. TDOA is time difference of pair of TOAs. AOA is angle of arrival which determines the direction of transmission signal incident on an antenna array. RSS is the received signal strength. Among them, TOA is frequently used for superior performance. In the literature, it is usually assumed that receiver positions are precisely known. However, this is not true in general case and a tiny variation in receiver positions would lead to a significant descend in estimation accuracy. On the other hand, in most of the literature, measurements coming from Line-of-Sight (LOS) paths are assumed. Unfortunately, due to practical harsh environments with obstacles in between the transceivers, signals are usually blocked in the LOS path, and reflected, scattered or refracted through Non-Line-of-Sight (NLOS) paths. The NLOS path would introduce certain propagation delay and cause additional uncertainty in the measurements. To improve the localization accuracy, it is necessary to mitigate the NLOS uncertainty and design effective algorithm for localization under NLOS environments. In this thesis, we will particularly investigate the above two problems in passive localization. One is TOA-based joint synchronization and passive localization with sensor uncertainty. Here, we assume there are no accurate sensor positions. The sensor position uncertainty will deteriorate the performance of existing localization algorithms and the main aim of our research is to mitigate this impact as much as possible. Meanwhile, the time offset between transmitter and receiver makes the problem even more difficult to solve. Based on weighted least squares (WLS) criteria, we excavate a valid method to estimate the location of passive target and the time offset between the transceiver simultaneously using TOA measurements. Simulation results show that the proposed method sufficiently reaches the Bayesian Cramer Rao bound (BCRB) and also attains theoretical mean square error (MSE) under small noise region. It also outperforms the other existing algorithm in the literature. The other problem we solve in this thesis is passive target localization in Non-Line-of-Sight environments. Considering the difficulties of acquiring a priori information on the distribution of NLOS errors in practice, we assume that there is no priori information of NLOS errors. Due to the complexity of signal prorogation paths in passive target scenarios and the absence of priori information of NLOS errors, the target localization becomes very challenging. A semi-definite programming (SDP) based algorithm is proposed to estimate the target location. Moreover, inspired by the active localization methods, Min-Max algorithm (MMA) and Quadratic Programming (QP) are successfully extended to solve our passive target localization with NLOS errors. The simulation results reveal the effectiveness of the NLOS mitigation method and low computational complexity of MMA and QP methods.