Networked Signal Processing for Large-Scale Satellite Navigation
This dissertation develops a networked signal processing framework for large-scale satellite navigation, in which receivers or satellites act as graph-connected agents that process local observations and exchange compact messages with neighboring nodes.
Overview
Modern satellite navigation is entering a large-scale networked era. The expansion of multi-constellation GNSS, dense terrestrial reference-station infrastructure, and emerging Low Earth Orbit (LEO)-augmented navigation systems is producing unprecedented volumes of distributed observations. Conventional centralized processing, in which raw measurements are collected and jointly processed at a monolithic analysis center, faces increasing pressure from computation, communication, latency, and resilience constraints.
The dissertation is built around a unifying interpretation of networked signal processing, which connects graph Fourier analysis, distributed graph filtering, and decentralized optimization, providing a common foundation for information aggregation and inference over large-scale satellite navigation networks. First, an error-feedback strategy is developed to improve the communication efficiency of quantized graph filtering by interpreting quantization errors as round-off noise in a state-space system. Closed-form feedback coefficients are derived for deterministic, random-graph, and asynchronous communication processes, and the resulting principles are further extended to communication-efficient diffusion-based decentralized optimization. Second, the framework is applied to ground-based GNSS networks, where time-varying graph models and graph-aware mixing schedules accelerate the decentralized estimation of receiver states and satellite correction products, potentially supporting advanced Precise Point Positioning (PPP) and PPP real-time kinematic (PPP-RTK) services. Third, the framework is extended to spaceborne GNSS networks comprising large LEO constellations, where onboard inference and localized exchange of GNSS information over rapidly evolving orbital graphs enable precise time synchronization and orbit determination despite practical inter-satellite communication impairments. Supported by extensive numerical experiments, this dissertation lays the methodological groundwork for the decentralized architecture of the future satellite-navigation infrastructure.
Presenters
Brief Biography
Xue Xian Zheng is a Ph.D. candidate in Electrical and Computer Engineering at King Abdullah University of Science and Technology (KAUST), under the supervision of Prof. Tareq Y. Al-Naffouri. He obtained a B.E. in Electrical Engineering from the Wuhan University of Technology (WUT) in 2019, and an M.Phi. in Electronic Engineering from South China University of Technology (SCUT) in 2022.
His work has been published in prominent journals including IEEE TSP, npj Wirel Technol, JSAC, and TBioCAS. He has also delivered several oral presentations at international conferences, including ICASSP and satellite conferences organized by the Institute of Navigation (ION).
