# Robust Resource Allocation for LEO Satellite-Assisted Secure SWIPT via STAR-RIS under CSI Uncertainty

## TL;DR

This paper develops a robust resource allocation method for LEO satellite SWIPT systems assisted by STAR-RIS, optimizing power transfer and secrecy under CSI uncertainty, and demonstrating significant performance improvements.

## Contribution

It introduces a novel robust optimization framework for STAR-RIS-assisted LEO satellite SWIPT, addressing channel uncertainty and coupled variables with an AO algorithm and penalty strategy.

## Key findings

- STAR-RIS significantly improves harvested power
- Proposed algorithm outperforms baseline schemes
- Robust design effectively handles CSI errors

## Abstract

This paper proposes a robust resource allocation framework for a low Earth orbit (LEO) satellite-enabled simultaneous wireless information and power transfer (SWIPT) system, assisted by a ground-deployed simultaneously transmitting and reflecting reconfigurable intelligent surface (STAR-RIS). We consider a scenario where direct satellite-to-ground links are obstructed, and the satellite serves multiple single-antenna energy receivers, information receivers, and eavesdroppers exclusively via the STAR-RIS. A robust optimization problem is formulated to maximize the total harvested power, subject to secrecy rate requirements, transmit power limits, and STAR-RIS coefficient constraints, under a practical bounded channel state information (CSI) error model. To achieve optimal robust resource allocation, we address the challenges posed by coupled optimization variables and bounded channel estimation errors by first applying the S-procedure to handle robustness against channel uncertainty. An alternating optimization (AO) framework is subsequently proposed, where the active beamforming at the LEO satellite and the passive beamforming at the STAR-RIS are jointly optimized, and a penalty-based strategy is incorporated to enforce the STAR-RIS beamforming design. Simulation results validate the effectiveness of the proposed algorithm and demonstrate that the STAR-RIS architecture achieves substantial performance gains in total harvested power over conventional RIS and other baseline schemes.