Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

APEG: Adaptive Physical Layer Authentication with Channel Extrapolation and Generative AI

Xiqi Cheng, Rui Meng, Xiaodong Xu, Haixiao Gao, Ping Zhang, Dusit Niyato

Abstract

With the rapid advancement of 6G, identity authentication has become increasingly critical for ensuring wireless security. The lightweight and keyless Physical Layer Authentication (PLA) is regarded as an instrumental security measure in addition to traditional cryptography-based authentication methods. However, existing PLA schemes often struggle to adapt to dynamic radio environments. To overcome this limitation, we propose the Adaptive PLA with Channel Extrapolation and Generative AI (APEG), designed to enhance authentication robustness in dynamic scenarios. Leveraging Generative AI (GAI), the framework adaptively generates Channel State Information (CSI) fingerprints, thereby improving the precision of identity verification. To refine CSI fingerprint generation, we propose the Collaborator-Cleaned Masked Denoising Diffusion Probabilistic Model (CCMDM), which incorporates collaborator-provided fingerprints as conditional inputs for channel extrapolation. Additionally, we develop the Cross-Attention Denoising Diffusion Probabilistic Model (CADM), employing a cross-attention mechanism to align multi-scale channel fingerprint features, further enhancing generation accuracy. Simulation results demonstrate the superiority of the APEG framework over existing time-sequence-based PLA schemes in authentication performance. Notably, CCMDM exhibits a significant advantage in convergence speed, while CADM, compared with model-free, time-series, and VAE-based methods, achieves superior accuracy in CSI fingerprint generation. The code is available at https://github.com/xiqicheng192-del/APEG

APEG: Adaptive Physical Layer Authentication with Channel Extrapolation and Generative AI

Abstract

With the rapid advancement of 6G, identity authentication has become increasingly critical for ensuring wireless security. The lightweight and keyless Physical Layer Authentication (PLA) is regarded as an instrumental security measure in addition to traditional cryptography-based authentication methods. However, existing PLA schemes often struggle to adapt to dynamic radio environments. To overcome this limitation, we propose the Adaptive PLA with Channel Extrapolation and Generative AI (APEG), designed to enhance authentication robustness in dynamic scenarios. Leveraging Generative AI (GAI), the framework adaptively generates Channel State Information (CSI) fingerprints, thereby improving the precision of identity verification. To refine CSI fingerprint generation, we propose the Collaborator-Cleaned Masked Denoising Diffusion Probabilistic Model (CCMDM), which incorporates collaborator-provided fingerprints as conditional inputs for channel extrapolation. Additionally, we develop the Cross-Attention Denoising Diffusion Probabilistic Model (CADM), employing a cross-attention mechanism to align multi-scale channel fingerprint features, further enhancing generation accuracy. Simulation results demonstrate the superiority of the APEG framework over existing time-sequence-based PLA schemes in authentication performance. Notably, CCMDM exhibits a significant advantage in convergence speed, while CADM, compared with model-free, time-series, and VAE-based methods, achieves superior accuracy in CSI fingerprint generation. The code is available at https://github.com/xiqicheng192-del/APEG
Paper Structure (31 sections, 37 equations, 15 figures, 3 tables, 3 algorithms)

This paper contains 31 sections, 37 equations, 15 figures, 3 tables, 3 algorithms.

Table of Contents

  1. Introduction
  2. System Model and Problem Formulation
  3. Network Model
  4. Channel Model
  5. Proposed Adaptive PLA Scheme With Channel Extrapolation and GAI (APEG)
  6. Training Phase
  7. Authentication Phase
  8. Proposed DM-based CSI Fingerprint Generation Schemes
  9. CSI Fingerprints as Images
  10. Proposed CCMDM-based Fingerprint Generation Scheme
  11. Proposed CADM-based Fingerprint Generation Scheme
  12. Complexity Analysis of the Proposed Schemes
  13. Complexity Analysis of CCMDM
  14. Complexity Analysis of CADM
  15. Latency and Power Consumption Analysis of CADM
  16. ...and 16 more sections

Figures (15)

  • Figure 1: The system model, where the collaborator's (Jack's) CSI fingerprints are used as the condition to generate Alice's real-time CSI fingerprints, and Eve transmits pilot signals to Bob to impersonate Alice via spoofing attacks.
  • Figure 2: Visualization of the real parts of Alice’s and Jack’s CSI fingerprints generated from the DeepMIMO 'O1' dataset Alkhateeb2019, vertically concatenated and displayed as a heatmap. The lower half shows Alice’s fingerprints and the upper half shows Jack’s fingerprints.
  • Figure 3: The positions of nodes involved, where Alice, Bob, and Jack are posited at (23.5 m, 266.93 m, 0), (0, 0, 4 m), and (23.5 m, 266.73 m, 0), respectively. Eves are randomly distributed within a circle centered at Alice with radius $r$.
  • Figure 4: Illustration of the proposed CADM scheme, where the U-Net framework consists of an encoder and a decoder, each comprising three downsampling and three upsampling modules. The network integrates cross-attention and self-attention modules to fuse features across multiple scales for enhanced CSI fingerprint generation.
  • Figure 5: The visualization of the denoising process during CSI fingerprint generation. The upper layer of (a) (b) (c) is the real part channel of the generated fingerprint, and the layer below is the imaginary part. $t$ is the denoising time step. (a) is the DeepMIMO O1 outdoor frequency domain CSI fingerprint, (b) is the DeepMIMO O1 outdoor angle delay domain CSI fingerprint, (c) is the DeepMIMO I3 indoor frequency domain CSI fingerprint.
  • ...and 10 more figures

Theorems & Definitions (2)

  • Remark 1
  • Remark 2