Huang, Z. (2026). Large-Scale RIS Optimization for Wideband Communications and Positioning [Dissertation, Technische Universität Wien]. reposiTUm. https://doi.org/10.34726/hss.2026.86283
Reconfigurable intelligent surfaces (RISs) can actively shape wireless propagation through programmable electromagnetic responses. Their large apertures and flexible configurations offer significant opportunities for communication and localization, but also challenge conventional narrowband and far-field assumptions. This dissertation investigates RIS configuration from two complementary perspectives: mitigating beam misfocus in wideband communications and enabling accurate near-field localization with large-aperture RISs.For wideband communications, frequency-dependent propagation and practical RIS responses cause different frequency components to experience different focusing conditions, leading to beam misfocus. Under a frequency-flat RIS model, this dissertation develops a semidefinite-relaxation-based near-optimal benchmark and a low-complexity stationary-phase-based method with closed-form phase configuration. It further shows that frequency-flat designs can degrade substantially under practical frequency-selective RIS responses and introduces an optimal phasor (OP)-based framework for efficient configuration under general frequency-selective models. Single-frequency and multi-frequency configurations are compared in terms of sum rate, identifying RIS-supported phase variation, bandwidth, propagation conditions, and the number of RIS states as key factors governing the benefit of wideband-aware optimization.The second part considers localization with large-aperture RISs in the radiative near field. A configuration-state-agnostic framework is developed that relies on conventionally estimated effective BS-RIS-UE channels rather than element-wise RIS states or subarray-specific coding. To characterize the geometric ambiguity caused by the finite RIS aperture, the uncertainty segment (USG) is introduced, and a min-max positioning method is formulated using multiple RIS-assisted USGs. A positioning-oriented RIS optimization problem is further derived by linking delay and Doppler-angle estimation errors to USG displacement and is efficiently solved using Riemannian gradient descent. Numerical results show that the proposed approach outperforms conventional RIS-center anchoring, remains robust under partial blockage, and can achieve approximately 10-cm-level accuracy under sufficiently strong received signals and measurement averaging.Overall, this dissertation shows that effective RIS design must account for both surface response characteristics and the task-specific structure of the propagation environment. The proposed methods provide practical design strategies and insights for wideband communication and near-field localization.
en
Additional information:
Arbeit an der Bibliothek noch nicht eingelangt - Daten nicht geprüft