Deep Joint Source-Channel Coding with Iterative Source Error Correction

TL;DR

This paper introduces an iterative decoding scheme for deep joint source-channel coding that enhances reconstruction quality and robustness against channel noise mismatches by using neural network-based denoising within an iterative MAP approximation framework.

Contribution

It proposes a novel iterative source error correction method for Deep JSCC that improves decoding performance and robustness over traditional one-shot decoding.

Findings

Improved distortion and perceptual quality metrics.

More reliable source reconstruction under channel mismatch.

Effective neural network-based MAP approximation.

Abstract

In this paper, we propose an iterative source error correction (ISEC) decoding scheme for deep-learning-based joint source-channel coding (Deep JSCC). Given a noisy codeword received through the channel, we use a Deep JSCC encoder and decoder pair to update the codeword iteratively to find a (modified) maximum a-posteriori (MAP) solution. For efficient MAP decoding, we utilize a neural network-based denoiser to approximate the gradient of the log-prior density of the codeword space. Albeit the non-convexity of the optimization problem, our proposed scheme improves various distortion and perceptual quality metrics from the conventional one-shot (non-iterative) Deep JSCC decoding baseline. Furthermore, the proposed scheme produces more reliable source reconstruction results compared to the baseline when the channel noise characteristics do not match the ones used during training.