Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Modular Neural Wiretap Codes for Fading Channels

Daniel Seifert, Onur Günlü, Rafael F. Schaefer

TL;DR

This work studies secure communication over finite-blocklength, fading wiretap channels without transmitter CSI. It proposes a modular wiretap coding scheme combining a neural autoencoder-based reliability layer with a 2-universal-hash security layer, and uses a mutual information neural estimator to quantify leakage. Through experiments on multi-tap Rayleigh fading channels, it shows that fading can improve the equivocation rate and reduce information leakage, especially as the number of taps increases and Eve’s channel variance decreases. The findings highlight practical benefits of neural wiretap codes in realistic fading environments, while seed choice for the security layer appears largely inconsequential in the tested setup. The work suggests directions for scaling to longer blocks and improving MI estimation for more complex security analyses.

Abstract

The wiretap channel is a well-studied problem in the physical layer security literature. Although it is proven that the decoding error probability and information leakage can be made arbitrarily small in the asymptotic regime, further research on finite-blocklength codes is required on the path towards practical, secure communication systems. This work provides the first experimental characterization of a deep learning-based, finite-blocklength code construction for multi-tap fading wiretap channels without channel state information. In addition to the evaluation of the average probability of error and information leakage, we examine the designed codes in the presence of fading in terms of the equivocation rate and illustrate the influence of (i) the number of fading taps, (ii) differing variances of the fading coefficients, and (iii) the seed selection for the hash function-based security layer.

Modular Neural Wiretap Codes for Fading Channels

TL;DR

This work studies secure communication over finite-blocklength, fading wiretap channels without transmitter CSI. It proposes a modular wiretap coding scheme combining a neural autoencoder-based reliability layer with a 2-universal-hash security layer, and uses a mutual information neural estimator to quantify leakage. Through experiments on multi-tap Rayleigh fading channels, it shows that fading can improve the equivocation rate and reduce information leakage, especially as the number of taps increases and Eve’s channel variance decreases. The findings highlight practical benefits of neural wiretap codes in realistic fading environments, while seed choice for the security layer appears largely inconsequential in the tested setup. The work suggests directions for scaling to longer blocks and improving MI estimation for more complex security analyses.

Abstract

The wiretap channel is a well-studied problem in the physical layer security literature. Although it is proven that the decoding error probability and information leakage can be made arbitrarily small in the asymptotic regime, further research on finite-blocklength codes is required on the path towards practical, secure communication systems. This work provides the first experimental characterization of a deep learning-based, finite-blocklength code construction for multi-tap fading wiretap channels without channel state information. In addition to the evaluation of the average probability of error and information leakage, we examine the designed codes in the presence of fading in terms of the equivocation rate and illustrate the influence of (i) the number of fading taps, (ii) differing variances of the fading coefficients, and (iii) the seed selection for the hash function-based security layer.
Paper Structure (12 sections, 7 equations, 5 figures, 1 table)

This paper contains 12 sections, 7 equations, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. System Model
  3. Wiretap Coding
  4. Reliability Layer
  5. Security Layer
  6. Mutual Information Estimation
  7. Main Results
  8. Average Error Probability and Information Leakage
  9. Equivocation Rate
  10. Fading Coefficient Variance Analysis
  11. Seed Selection
  12. Conclusion

Figures (5)

  • Figure 1: Modular wiretap code design consisting of the reliability layer $(e_r, d_r)$ and the security layer $(\varphi_s, \psi_s)$.
  • Figure 2: Reliability and security evaluation of the designed WTC for constant rates $R_s=1/4$, $R_r=1/2$, $E_b/N_{0,Y} > E_b/N_{0,Z}$, and ${\omega_{Y}^2} = {\omega_{Z}^2} = 1$.
  • Figure 3: Equivocation rate $R_e$ over blocklength $n$ for varying $E_b/N_{0,Z}$.
  • Figure 4: Information leakage $L$ to Eve over blocklength $n$ for varying $\omega_Z^2$ and $E_b/N_{0,Y}=E_b/N_{0,Z}=0\dB$.
  • Figure 5: Histograms of Hamming distances $d_H$ and Lee distances $d_L$ for the $16$-step quantized encoder output for all possible combinations of message pairs $(m_1, m_2)$ with $m_1\neq m_2$ and random bits $b$, for the WTC with $k=4$, $q=8$ and $n=16$ and ${s=(0,0,0,0,0,0,1,1)}$.

Theorems & Definitions (3)

  • Definition 1: mittelbachSensingAssistedSecureCommunications2025
  • Definition 2: rana2023
  • Definition 3: hayashi2010