Positioning using Low Earth Orbit mega-constellations

Global Navigation Satellite System (GNSS) is the backbone of numerous critical applications, including autonomous driving, timing and synchronization in telecommunications, power grid, network, financial transactions, scientific experiments, critical infrastructure operations such as energy supply networks, transport infrastructures, search and rescue operations, air traffic management, mapping and surveying, precision agriculture, unmanned aerial vehicle (UAV), etc. However, it is facing inherent challenges, such as limited signal strength, susceptibility to interference, jamming, spoofing, multipath effects, and degraded performance in obstructed environments. These limitations necessitate exploring alternative/complementary solutions to enhance Positioning, Navigation, and Timing (PNT) services. This thesis focuses on leveraging Low Earth Orbit (LEO) mega-constellations to address the shortcomings of legacy GNSS. Their rapid deployment, cost-effectiveness, and global coverage offer significant potential for improving positioning accuracy and robustness, particularly in challenging environments like urban canyons. However, all studies have explored isolated LEO-PNT approaches, providing fragmented insights into specific methods. This research bridges that gap by providing a holistic evaluation of major LEO-PNT methodologies (Dedicated LEO-PNT, LEO-PNT through Signal of Opportunity (SoO), Fused LEO-PNT (Integrated Navigation/Communication)), establishing a robust framework for future advancements. The primary contributions of this work are fourfold. First, it introduces a detailed mathematical model and optimization framework for designing dedicated LEO-PNT constellations. The study optimizes constellation configurations by employing Genetic Algorithms (GA) to enhance global availability and positioning accuracy by minimizing Geometric Dilution of Precision (GDOP). Six hybrid constellation configurations are obtained, which provide 100 % global availability and average GDOP around 1.5 to any latitude/longitude at any epoch. These configurations are used for Position, Velocity, and Time (PVT) estimation of static and dynamic users using the Least Square (LS) solution. Obtained PVT accuracy is further improved by integrating with Inertial Navigation Systems (INS) using an Extended Kalman Filter (EKF). Second, this thesis develops a novel geometric urban canyon model, enabling a statistical performance analysis in terms of availability and GDOP of the positioning service of in-orbit mega-constellations, such as Starlink and OneWeb, in deep urban canyon environments, and compares the performance with the one achieved by GNSS systems. Real-world urban canyon dimension datasets of two major locations, namely the city of London and the Manhattan District of New York City are statistically processed. Probability density functions (PDFs) of building heights and lengths, and street widths in the considered urban areas, are generated and incorporated into the model. The outcomes reveal that LEO mega-constellations outperform GNSS in terms of GDOP and availability in dense urban settings. Third, this developed urban canyon model is further enhanced using 3D ray tracing to construct multipath geometry and assess the key multipath parameters that contribute to measurement errors. The PDF of measurement errors caused by multipath has been estimated for the same constellations. Statistical analysis was performed on the urban canyons of the same locations, which shows that for GNSS constellations, the percentage of satellite links affected by multipath propagation is 50% in London and 70% in Manhattan, which is characterized by deeper urban canyon scenarios. For LEO mega-constellations, the same parameter is reduced to 3% in London and 17% in Manhattan. This effect translates into a reduction of the average pseudorange measurement error due to multipath from 3.5 m and 3.9 m for GNSS to 1.8 m and 2 m for LEO mega-constellations in London and Manhattan, respectively. Lastly, the research explores the integration of navigation and communication in future sustainable LEO-based Non-Terrestrial Networks (NTN), by repurposing the frame structures of established satellite communication protocols, such as Digital Video Broadcasting Satellite Second Generation Extended (DVB-S2X) and 5th Generation - New Radio (5G-NR), for positioning. Both protocols are excellent backbones for modern satellite communications deployments, due to their high spectral efficiencies, therefore, the assessment of their suitability to also provide PNT services is of paramount importance. The detailed frame structure of both protocols is exploited, focusing on the Synchronization Sequence (SS), which is analogous to the Pseudorandom Noise (PRN) codes of GNSS. These SS can be repurposed for signal acquisition and tracking to estimate the Time of Arrival (TOA), similar to GNSS. Furthermore, the open fields within these frame structures can be utilized to transmit ephemeris information using a multi-tier architecture with GNSS as a backhaul. An analysis of the Receiver Operating Characteristics (ROC), the Mean Acquisition Time (MAT) for statistical signal acquisition, and the Cramer-Rao Lower Bound (CRLB) for signal tracking are carried out to evaluate the potential availability and accuracy of pseudorange measurements based on these waveforms. In addition, given that 5G-NR offers a dedicated ondemand Positioning Reference Signal (PRS), a comparative assessment is conducted to analyze its tracking performance and investigate the influence of signal parameters and resource allocation strategies on PRS ranging accuracy. This comprehensive research provides a foundation and trade-off for developing next-generation LEO-PNT systems capable of delivering robust and precise positioning services in diverse and challenging environments.