Low Complexity Autoencoder based End-to-End Learning of Coded Communications Systems

TL;DR

This paper compares deep learning autoencoders to traditional coding methods in communication systems, showing autoencoders can outperform conventional codes in certain conditions and introduces a low complexity autoencoder architecture with improved performance.

Contribution

It presents a novel low complexity autoencoder architecture for end-to-end coded communication systems with superior BER performance over traditional methods.

Findings

Autoencoders outperform convolutional codes in 0-5 dB range.

Proposed autoencoder achieves better BER than 16-QAM with hard decision decoding.

Autoencoder shows improved BER over soft decision decoding in 0-4 dB range.

Abstract

End-to-end learning of a communications system using the deep learning-based autoencoder concept has drawn interest in recent research due to its simplicity, flexibility and its potential of adapting to complex channel models and practical system imperfections. In this paper, we have compared the bit error rate (BER) performance of autoencoder based systems and conventional channel coded systems with convolutional coding (CC), in order to understand the potential of deep learning-based systems as alternatives to conventional systems. From the simulations, autoencoder implementation was observed to have a better BER in 0-5 dB $E_{b}/N_{0}$ range than its equivalent half-rate convolutional coded BPSK with hard decision decoding, and to have only less than 1 dB gap at a BER of $10^{-5}$. Furthermore, we have also proposed a novel low complexity autoencoder architecture to implement end-to-end learning of coded systems in which we have shown better BER performance than the baseline implementation. The newly proposed low complexity autoencoder was capable of achieving a better BER performance than half-rate 16-QAM with hard decision decoding over the full 0-10 dB $E_{b}/N_{0}$ range and a better BER performance than the soft decision decoding in 0-4 dB $E_{b}/N_{0}$ range.