Modular Neural Wiretap Codes for Fading Channels

TL;DR

This paper introduces a deep learning-based finite-blocklength coding scheme for secure communication over fading wiretap channels without channel state information, analyzing its performance under various fading conditions.

Contribution

It provides the first experimental characterization of neural wiretap codes for fading channels, focusing on finite-blocklength performance and security metrics.

Findings

Codes maintain low error and leakage in fading environments

Performance varies with number of taps and fading variances

Security is influenced by seed selection for hash functions

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.