Two-Stage Distributionally Robust Optimization Framework for Secure Communications in Aerial-RIS Systems

This letter investigates the robust secrecy design of aerial reconfigurable intelligent surface (A-RIS) systems under multi-source uncertainties spanning mechanical jitter, hardware impairments, and channel estimation errors. Distinct from works relying on ideal assumptions, we address the coupling between uncrewed aerial vehicle (UAV) attitude instability and fastvarying channels. Specifically, we propose a two-stage distributionally robust optimization (DRO) framework to decouple the long-term UAV placement from the per-slot beam forming design. By introducing a lightweight surrogate model for efficient deployment search and leveraging conditional value-at-risk (CVaR)for risk mitigation, the proposed method bridges the timescale mismatch using a Wasserstein ambiguity set for instantaneous channel estimation errors. Simulation results validate that this approach ensures real-time feasibility and improves tail-end secrecy spectral efficiency (SSE), achieving a gain of approximate 68% over conventional robust benchmarks