Reconfigurable intelligent surface-aided secret key generation using an autoencoder and K-means quantization

• Proposes a novel RIS-aided secret key generation framework that dynamically tunes reflection coefficients to overcome quasi-static channel limitations, enhancing channel randomness. • Combines an autoencoder with K-means clustering quantization to efficiently convert high-dimensional CSI into random bits, significantly reducing key disagreement rate (KDR). • Simulation results demonstrate superior key generation rate (KGR) and KDR performance across various SNR conditions, validated using the NIST randomness test suite. • Offers a practical solution for physical-layer security in static or slowly varying wireless environments, addressing a critical vulnerability in IoT and resource-constrained networks.