Improving UWB Ranging Accuracy and Security in Complex Multi-path Environments

Abstract

Measurement of the relative distance between devices (ranging) is necessary for a wide range of modern-day applications such as indoor positioning, the use of contactless tokens for executing payment transactions, providing access to infrastructure, passive keyless entry and start systems, unlocking digital devices, and credential verification of electronic passports. UWB technology has gained popularity over the years due to its reliable ranging capabilities, especially in Line-Of-Sight (LOS) conditions. This can be attributed to its high spatial and temporal resolution. Also, the low power consumption rate, low cost, and low interference due to its wide frequency range make it a leading choice for ranging in indoor and industrial environments. However, the widespread presence of Non-Line-Of-Sight (NLOS) and multi-path signals (signals traversing multiple paths between transceivers due to the presence of obstructions) presents a significant challenge to UWB ranging in complex or harsh environments (characterized by the presence of objects of high relative permittivity, e.g., concrete walls, metallic objects, and furniture), affecting ranging accuracy. With UWB, the distance between a transmitter and receiver is calculated by measuring the Time-Of-Flight (TOF) between the two transceivers, ideally under Line-of-Sight (LOS) conditions. Ranging algorithms assume TOF estimations occur under direct LOS; however, when signals include NLOS and multi-path components, this leads to overestimated distances due to inaccurate TOF measurements, ultimately compromising ranging accuracy. The effects of NLOS and multi-path propagation are studied for two main UWB applications: positioning (NLOS identification and mitigation) and secure distance bounding.

The first part of the dissertation proposes a transfer learning-based approach to alleviate the limitations of current NLOS identification works in literature. A myriad of promising NLOS identification approaches have been proposed, including channel statistics-based approaches, non-channel statistics-based approaches, outlier detection-based approaches, etc. However, the key limitations of these approaches are that: 1) their performance is site-specific, i.e., their performance is limited to the environment where data was collected and cannot be extended to other environments; 2) Change in noise/interference pattern – a trained model will significantly degrade in performance in a different environment if the noise pattern or interference is different from the original environment. To overcome these limitations, a robust Transfer Learning (TL)-based NLOS identification approach is proposed, which harnesses transition information obtained via cross-domain mappings of both source and target domains to construct representative homogeneous features. The representative homogeneous features capture discriminative information of both domains, while reducing the distribution divergence between the domains, making it easy to classify LOS and NLOS components from both environments simultaneously. Extensive simulations are presented to verify the robustness of the proposed approach.

The second part of the dissertation focuses on a novel approach to alleviate the limitations of NLOS mitigation approaches in literature. An approach that exploits the differences in time between an established reference index and the index of the maximum peak of the Channel Impulse Response (CIR) for NLOS mitigation is proposed. These differences in time are used to construct a Delay Profile Information (DPI) of the ranging signals, which serves as metadata to complement indoor localization algorithms to improve accuracy in NLOS-dense environments. The proposed approach can be implemented on the fly, that is, it can mitigate the NLOS effects directly on the acquisition of the ranging signals during localization. It does not require an offline phase, which typically involves site surveys or measurement campaigns and training computationally intensive models for NLOS mitigation in the online phase. The performance of the proposed technique is evaluated and compared to state-of-the-art approaches in two real-world NLOS-dense environments.

The third part of the dissertation focuses on solving one key limitation of current secure distance-bounding (DB) protocols in literature, i.e., their susceptibility to NLOS and multi-path propagation, which affects their resiliency in terms of TOF. NLOS and multi-path propagation induce TOF errors, which may lead to false rejection of legitimate provers during distance-bounding, making DB protocols less resilient by increasing their false rejection rate (FRR). To solve this limitation, a timing threshold function is proposed to effectively identify and mitigate the errors introduced by NLOS and multi-path conditions to improve the TOF resilience in DB protocols. Security considerations regarding the proposed threshold function are also considered and presented.

The fourth part of this thesis focuses on the limitations of current DB protocols, i.e., for instantaneous verification of proximity of static provers. The term "persistent distance-bounding" is introduced in distance bounding literature. The core idea is that a persistent distance bounding method would complement existing protocols and give the verifier some indication of whether a mobile prover remains within the determined bound as their communication session continues. This allows the verifier to maintain a session with the prover without continuously running the full distance-bounding protocol. An algorithm termed PiBiF (Position-independent Bounding with integrated Flexibility), which aids the verifier to establish a stable fingerprint to continually check and monitor the bound between itself and a free-moving prover, is proposed. This aids DB protocols to be resilient to prover movements while validating the bound. Security considerations and evaluations regarding PiBiF are also considered and presented.

Date of Award	4 Jul 2025
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Gerhard Petrus HANCKE (Supervisor)