End-to-End Learning of Communications Systems Without a Channel Model

TL;DR

This paper introduces a novel end-to-end learning algorithm for communication systems that does not require a differentiable channel model, combining supervised and reinforcement learning techniques.

Contribution

It presents a new training method that enables learning over arbitrary channels without prior models, improving flexibility and applicability.

Findings

Works well on AWGN and RBF channels

Converges faster on Rayleigh block-fading channels

Achieves comparable performance to supervised methods

Abstract

The idea of end-to-end learning of communications systems through neural network -based autoencoders has the shortcoming that it requires a differentiable channel model. We present in this paper a novel learning algorithm which alleviates this problem. The algorithm iterates between supervised training of the receiver and reinforcement learning -based training of the transmitter. We demonstrate that this approach works as well as fully supervised methods on additive white Gaussian noise (AWGN) and Rayleigh block-fading (RBF) channels. Surprisingly, while our method converges slower on AWGN channels than supervised training, it converges faster on RBF channels. Our results are a first step towards learning of communications systems over any type of channel without prior assumptions.