Geometry-Based Channel Quality Prediction under Unknown RIS Configuration

초록

Reconfigurable intelligent surfaces (RISs) are typically studied under cooperative assumptions with known phase configurations and channel state information (CSI). In practice, however, RISs deployed on building façades may operate autonomously, rendering their configurations unknown to user equipment (UE). This paper considers a non-cooperative RIS-assisted vehicular scenario where a mobile receiver communicates with a fixed base station without access to RIS phase information or cascaded CSI. We develop a geometry-based signal model with random RIS phase configurations and show that, despite instantaneous randomness, the mean received signal strength indicator (RSSI) depends deterministically on the UE position. Leveraging this property, we formulate channel quality indicator (CQI) prediction as a nonlinear state estimation problem and propose an extended Kalman filter (EKF)-based framework that fuses noisy position and RSSI measurements. Simulation results demonstrate that the proposed framework achieves nearly a 57% reduction in CQI prediction RMSE as the RIS size increases from N=16 to N=1024, highlighting its effectiveness for RIS-assisted communication without RIS cooperation.

키워드