Non-line-of-sight localization in multipath environments

Abstract

Wireless localization is a major challenge for accurately estimating the position of devices that operate in multipath environments. In indoor environments, non-line-of-sight (NLOS) propagation has a significantly negative impact on the performance of conventional localization schemes. These schemes can be divided into two categories. The first is known as the fingerprinting scheme. Under this scheme, localization begins with a training process to construct a database of the features of a predetermined location before the device is used in that location. When a user with the unknown mobile device (MD) sends a positioning query, the localization scheme searches the database and returns the corresponding fingerprints and locations. Fingerprinting can handle the NLOS problem. Doing so requires the pre-calibration of signal characteristics between each Reference Device (RD) and the MD. In addition, the scheme is sensitive to changes in the environment. The second localization scheme is called the geometric scheme and makes use of position-related parameters, such as the angle of arrival (AOA), the time of arrival (TOA), and received signal strength (RSS), to construct the geometric position of possible MDs under the line-of-sight (LOS) assumption. In two-dimensional localization, at least two or three RDs are required for the AOA and TOA localization schemes respectively. However, the geometric approach faces a serious challenge when a signal is disturbed by the multipath effect. Therefore, much research effort has been devoted to tackling the NLOS problem to improve localization. These efforts have focused either on NLOS identification or NLOS mitigation. NLOS identification distinguishes LOS/NLOS range measurement information and uses the LOS information in the conventional methodology to find the estimated location. NLOS mitigation reduces the impact of NLOS paths on localization accuracy by assigning lower weights to longer propagation paths. However, these schemes have not been satisfactory in heavy multipath environments. Other proposed schemes include the use of NLOS paths for localization without the need for mitigation schemes. The schemes work by first constructing the lines of possible MD locations, referred to as the line of possible mobile device (LPMDs), on the LOS or NLOS paths based on pairs of TOA and AOA measurements at both the RD and the MD. The MD’s location can be determined at the intersection points of the LPMDs, causing this technique to be known as line intersection methodology. It is worth noting that line intersection methodology does not work well in a dense multipath environment, especially when the angle between LPMDs is small. This thesis focuses on novel and robust geometrical-based localization schemes in a multipath environment by using just one RD and without the need for any identification and mitigation schemes. Firstly, a robust localization scheme is proposed based on a Gaussian weighting function and proximate points. This scheme uses all measurement data (TOA and AOA) and either LOS or NLOS propagation information to formulate a Gaussian weighting function and proximate points to find the MD’s location without any identification and mitigation schemes. To further improve localization accuracy, an area-based localization scheme which does not require any weighting factor was designed to estimate the MD’s location. To perform a robust localization to handle multiple reflections and diffractions in the dense multipath environment, virtual RD-based and elliptical Lagrange-based NLOS localization schemes are proposed to determine the MD’s location by using one RD and one signal path which undergoes one or more reflections or diffractions. Finally, experiments are conducted to verify the localization accuracy of the virtual RD-based and elliptical Lagrange-based NLOS localization schemes in a heavy multipath environment.