Mobility-Aware Predictive Robust Secrecy Design for STAR-RIS-Assisted V2I Networks Under CSI Aging

This paper investigates predictive robust physical-layer security (PLS) for a simultaneous transmitting and reflecting reconfigurable intelligent surface (STAR-RIS)-assisted downlink vehicle-to-infrastructure (V2I) network under mobility-induced channel aging. In high-mobility vehicular environments, the channel state information (CSI) available at the roadside transmitter becomes outdated due to Doppler effects and feedback delay, which can severely degrade secrecy performance in the presence of passive eavesdropping vehicles. To capture this practical impairment, we model the CSI aging through a Gauss--Markov time-correlation process and represent the residual prediction mismatch by norm-bounded uncertainty sets whose radii are linked to the channel time correlation. Based on the predicted CSI, we formulate a per-slot robust secrecy sum-rate (SSR) maximization problem by jointly optimizing the multiuser transmit beamforming (BF) and STAR-RIS transmission/reflection coefficients under the transmit-power budget and energy-splitting constraints. To solve the resulting non-convex and uncertainty-coupled problem, we develop a predictive robust alternating-optimization framework, where the BF subproblem is handled through S-procedure-based linear matrix inequalities and successive convex approximation, while the STAR-RIS coefficients are updated by feasibility-preserving projected steps. Numerical results show fast convergence and demonstrate that the proposed design outperforms non-predictive, non-robust, random STAR-RIS, and reflect-only RIS baselines, especially under high mobility, long CSI update intervals, and dense eavesdropping deployments.