AI/ML-Aided Processing for Physical-Layer Security

Abstract

This chapter introduces methodologies for preprocessing channel state information (CSI) data in the context of authentication and secret key generation (SKG) for wireless communication systems. It proposes the use of principal component analysis (PCA) and autoencoders (AEs) to effectively decompose the CSI matrix into predictable and unpredictable components. These methodologies are validated using a synthetic dataset generated with the Quadriga channel model. The findings demonstrate that both PCA and AE methods significantly enhance the separability and reduce the correlation among CSI components. © 2024 by The Institute of Electrical and Electronics Engineers, Inc. All rights reserved.