RISCS: RIS-Aided Integrated Sensing and Communications–A Cyber-Physical Security Perspective

Abstract

This article presents a comprehensive framework for modeling, detecting, and mitigating cyber-physical threats in wireless communication systems, spanning two architectures: Reconfigurable Intelligent Surface (RIS)-assisted systems, and RIS-enabled Integrated Sensing and Communication (RIS-ISAC) systems. We first classify common attack vectors, including jamming, spoofing, eavesdropping, and data injection, and evaluate their impact across different system architectures. We then highlight system-specific vulnerabilities and assess corresponding mitigation strategies based on feasibility and implementation complexity. We present a Secure Integrated Sensing and Communication (SISAC)-enabled RIS framework designed to improve localization, sensing, and communication in RIS-assisted systems. Unlike traditional radio localization methods that depend on synchronized Base Stations (BS) and Time of Arrival (TOA) or Angle of Arrival (AOA) estimates, the introduction of RIS offers additional geometric reference points, thereby significantly enhancing localization accuracy. In the sensing domain, we focus on RIS-assisted localization and highlight that the Fisher Information Matrix (FIM) is employed to determine the theoretical lower bound on the estimation accuracy of sensing metrics, which in turn guides RIS placement and configuration for improved coverage. Simulation results demonstrate that RIS directional selectivity significantly enhances the Signal-to-Noise Ratio (SNR) and the secrecy rate for the intended user while degrading performance for unintended locations. Specifically, simulation results quantify the impact of RIS configuration on SNR, MSE, secrecy-related metrics, and localization bounds, while experimental measurements using an RIS-assisted OFDM testbed validate the proposed anomaly detection mechanism under realistic propagation conditions. To validate the proposed framework, we implement a practical RIS-assisted anomaly detection testbed using pilot-aided OFDM and 4096 discrete RIS configurations. A novel detection algorithm jointly evaluates received SNR and Mean Squared Error (MSE) of channel estimation to flag potential eavesdroppers without requiring prior knowledge of their positions. Experimental results confirm that the proposed method enhances physical-layer security by achieving a consistently positive secrecy rate for legitimate users, while degrading the performance of detected eavesdroppers.