Robust IRS Beamforming for IoT-Enabled LEO Satellite Systems via Channel Decoupling under Ephemeris-Aided Statistical Modeling

초록

Low Earth Orbit (LEO) satellite systems empowered by intelligent reflecting surfaces (IRSs) offer new opportunities for seamless Internet of Things (IoT) connectivity, but they face two major challenges: rapidly varying channels caused by satellite mobility and the structural mismatch between the cascaded base station (BS)-IRS-user equipment (UE) channel and the Discrete Fourier Transform (DFT)-based codebook specified by the Third Generation Partnership Project (3GPP). To address these issues, we propose a channel feedback and beamforming framework that decouples the cascaded channel into BS-IRS and IRS-UE components. The BS-IRS link, though fast-changing, follows predictable and periodic satellite trajectories, enabling statistical modeling from ephemeris data. In contrast, the IRS-UE link varies slowly and is typically dominated by a single angular path, which can be effectively represented using the DFT-based codebook. Leveraging this property, we apply sparse recovery to extract directional features, quantize them, and feed them back in a manner compliant with 3GPP standards. This design enables robust IRS beamforming aligned with the dominant UE-side channel direction. Simulation results demonstrate that the proposed approach effectively mitigates channel aging and codebook mismatch, leading to stable beam patterns and improved reliability over conventional methods. © 2014 IEEE.

키워드