Distributed Image Transmission using Deep Joint Source-Channel Coding

TL;DR

This paper introduces a deep neural network-based joint source-channel coding scheme for correlated stereo images transmitted over noisy channels, significantly improving reconstruction quality by exploiting source and channel correlations.

Contribution

It proposes a novel deep learning framework with a channel state information aware cross attention module for efficient correlated image transmission over wireless channels.

Findings

Enhanced reconstruction quality by leveraging source and channel correlations.

Outperforms separated coding schemes with capacity-achieving codes.

Effective in noisy wireless environments for stereo image transmission.

Abstract

We study the problem of deep joint source-channel coding (D-JSCC) for correlated image sources, where each source is transmitted through a noisy independent channel to the common receiver. In particular, we consider a pair of images captured by two cameras with probably overlapping fields of view transmitted over wireless channels and reconstructed in the center node. The challenging problem involves designing a practical code to utilize both source and channel correlations to improve transmission efficiency without additional transmission overhead. To tackle this, we need to consider the common information across two stereo images as well as the differences between two transmission channels. In this case, we propose a deep neural networks solution that includes lightweight edge encoders and a powerful center decoder. Besides, in the decoder, we propose a novel channel state information aware cross attention module to highlight the overlapping fields and leverage the relevance between two noisy feature maps.Our results show the impressive improvement of reconstruction quality in both links by exploiting the noisy representations of the other link. Moreover, the proposed scheme shows competitive results compared to the separated schemes with capacity-achieving channel codes.