Robust Indoor Pedestrian Localization in Non-Line-ofSight Conditions using Ultra-Wideband Active Communication

Abstract

Indoor location-based services (ILBSs) to assist daily life tasks in the area of healthcare, shopping malls, and sports are becoming a trending topic. The key issue for ILBSs is to accurately locate pedestrians indoors. For this purpose, ultra-wideband (UWB) technology with appealing characteristics such as high bandwidth, low cost, and low power consumption is preferable for indoor pedestrian localization at centimeter-level accuracy. However, the channel blocking of UWB active communication, commonly known as non-line-of-sight (NLOS), can drastically impact the localization accuracy with meter-level error. Therefore, many methods of improving UWB-based indoor pedestrian localization accuracy under NLOS using extra perceptions, like signal quality and walking direction, have been proposed. This thesis proposes novel methods to leverage the need for extra perception and address the UWBNLOS issue. Focusing on the ILBS to facilitate daily life tasks, we enhanced the UWB location accuracy in contextual wearable placements of UWB sensors such as smartphone and smartshoe. First, we proposed a robust indoor localization approach for UWB-enabled smartphone which does not require the extra perceptions to mitigate the human body shadowing (UWB-NLOS due to the human own body). Second, we proposed a low-cost footplaced UWB and inertial measurement unit (IMU) fusion technique to address the UWBNLOS due to indoor surroundings as well as the IMU drift issue. For validation of the proposed methods, we conducted real-time walking experiments in a multipath indoor environment and deployed the state-of-the-art Hokuyo Lidar to provide ground truth data. The proposed indoor pedestrian localization approaches can contribute to the adoption of personal devices’ embedded UWB chips and low-cost IoT wearables for pervasive and ubiquitous indoor daily life use cases with centimeter-level accuracy.