Dr. Li-Ta HSU, AFRIN, MION, SMIEEE  
Limin Young Scholar in Aerospace Navigation  
Associate Professor and Associate Head,  
Department of Aeronautical and Aviation Engineering,  
The Hong Kong Polytechnic University  

Speech Title：The application of artificial intelligence in satellite navigation
Abstract: 

Satellite navigation systems, such as GPS and BeiDou, are complex infrastructures that play a crucial role in modern navigation and positioning. These systems consist of three interdependent segments: the space segment, the ground control segment, and the user segment. A complex system, long signal travel distances, and low power levels may present various challenges. For example, the GNSS signal that travels more than 20,000 km propagates through various media, such as the ionosphere and troposphere, with weak power, making it vulnerable to different types of interference and distortion. Such distortion could exhibit a significant non-linearity, and artificial intelligence (AI) becomes essential for managing and optimizing complex interactions. This talk will explore how AI is revolutionizing satellite navigation by addressing the existing challenges and enhancing system performance.
One of the primary challenges in GNSS is the accurate determination of satellite orbits. Though the orbit of GNSS satellites in medium Earth orbit (MEO) are well-established, the GNSS satellite orbit accuracy could be compromised due to factors like gravitational perturbation, satellite inertial, and solar radiation pressures, hindering the centimeter level accuracy. Beyond the GNSS satellites, there is a growing interest in utilizing low Earth orbit (LEO) satellites for navigation purposes. LEO satellites offer advantages such as reduced signal propagation delays and potentially enhanced coverage in certain regions. However, their dynamic nature and lower altitudes make their orbits more difficult to track accurately. AI-driven algorithms can significantly improve the MEO and LEO determination by integrating real-time data and predictive modeling, enhancing the precision of both MEO and LEO satellite navigation systems. This capability is crucial for expanding the utility of satellite navigation in diverse applications.
For positioning, an accurate position of the signal source is just a start. Another significant challenge is the weak penetration of GNSS signals due to their low transmission power. These signals struggle to penetrate obstacles like concrete walls, making indoor positioning and the transition between indoor and outdoor environments particularly challenging. This limitation affects the reliability of navigation in urban canyons and complex environments. AI-powered adaptive signal processing techniques, such as beamforming and dynamic power adjustment, can optimize signal distribution and improve positioning accuracy in these difficult environments. By leveraging AI, it is possible to enhance signal reception and maintain reliable navigation even in challenging conditions.
The challenges do not just come from the weak power of the signal. During signal transmission, GNSS is susceptible to various error sources, including ionospheric delays, multipath interference, and Non-Line-of-Sight (NLOS) effects. These errors can significantly degrade positioning accuracy. AI techniques, such as machine learning models, can predict and correct these errors by analyzing historical and real-time data, ensuring more reliable positioning. For instance, neural networks can filter out delays caused by dynamic objects, while machine learning algorithms can model and compensate for ionospheric distortions.
Apart from the accuracy, security is also a critical concern for GNSS, as the signals are vulnerable to spoofing attacks, where counterfeit signals are used to deceive receivers. AI enhances GNSS security through anomaly detection and cryptographic signal authentication, enabling the identification of subtle spoofing patterns and rapid response to potential threats. This capability is essential for maintaining the integrity and trustworthiness of GNSS in critical applications.
In conclusion, AI is playing a transformative role in addressing the challenges faced by satellite navigation systems. AI is paving the way for more accurate, reliable, and secure navigation solutions. This integration of AI and satellite technology promises to enhance the robustness of GNSS and support the growing demand for precise navigation.

Biography: 

Dr Li-Ta Hsu, Limin Endowed Young Scholar in Aerospace Navigation, is currently an Associate Professor and Associate Head in Department of Aeronautical and Aviation Engineering at PolyU, earned his BSc and PhD degrees in aeronautics and astronautics from National Cheng Kung University, Taiwan, in 2007 and 2013 respectively. Winning a Student Paper Award from the Institute of Navigation (ION) in the U.S., he showed promising potential as a scholar early on. Once a visiting scholar at the University College London, and Tokyo University of Marine Science and Technology; a JSPS postdoctoral fellow in the Institute of Industrial Science, The University of Tokyo; a technical representative and a council member of ION; and currently an associate fellow of the United Kingdom’s Royal Institute of Navigation and a senior member of IEEE, Dr. Hsu garnered over ten years of R&D experience in the discipline of aerospace navigation. Since he joined PolyU in 2016, Dr Hsu has undertaken numerous collaborative R&D projects on GNSS positioning for smartphones and intelligent vehicles, especially useful in highly urbanised cities such as Hong Kong, Shanghai and Tokyo. In 2023 to 2024, Dr Hsu was invited to Google LLC as a visiting research scientist, focused on improving the positioning accuracy for billions of Android Phone. Dr Hsu was ranked as the world’s top 2% most-cited scientists since 2022. Dr Hsu has keen research interest in everything related to urban positioning and signal processing, including GNSS real-time kinematic positioning and ways to improve GNSS performance under challenging reception conditions. He also charts new territory in unmanned autonomous systems (UAS) positioning using different signals of opportunity, visual positioning system and LiDAR based simultaneous localization and mapping (SLAM). He has published more than 80 journal papers and is currently an associate editor in Navigation, Journal of the Institute of Navigation and Frontiers in Robotics and AI.