Serial vs. Parallel Turbo-Autoencoders and Accelerated Training for Learned Channel Codes

TL;DR

This paper compares serial and parallel Turbo-autoencoders for end-to-end PHY-Layer communication, introduces a faster training algorithm, and proposes a serial architecture inspired by classical Turbo codes, achieving state-of-the-art results.

Contribution

It introduces a component-wise training algorithm that significantly reduces training time and proposes a serial Turbo-autoencoder architecture that outperforms existing neural network based learned sequences.

Findings

Component-wise training reduces training time by nearly tenfold.

Serial Turbo-autoencoders outperform parallel ones and classical coding schemes.

The proposed architectures are the best neural network based learned sequences trained from scratch.

Abstract

Attracted by its scalability towards practical codeword lengths, we revisit the idea of Turbo-autoencoders for end-to-end learning of PHY-Layer communications. For this, we study the existing concepts of Turbo-autoencoders from the literature and compare the concept with state-of-the-art classical coding schemes. We propose a new component-wise training algorithm based on the idea of Gaussian a priori distributions that reduces the overall training time by almost a magnitude. Further, we propose a new serial architecture inspired by classical serially concatenated Turbo code structures and show that a carefully optimized interface between the two component autoencoders is required. To the best of our knowledge, these serial Turbo autoencoder structures are the best known neural network based learned sequences that can be trained from scratch without any required expert knowledge in the domain of channel codes.