GeNet: A Graph Neural Network-based Anti-noise Task-Oriented Semantic Communication Paradigm

TL;DR

GeNet addresses semantic communication under unknown channel conditions by decoupling performance from $\mathrm{SNR}$ in an AWGN channel. It converts input images into graphs via SLIC superpixels, uses a graph neural network encoder to extract task-relevant semantics, and a GNN decoder with mean readout to reconstruct semantics for TOC over the channel. The work's key contributions are: (i) the first application of GNNs to anti-noise semantic communication, (ii) SNR-agnostic denoising that avoids training across multiples of $\mathrm{SNR}$, (iii) support for variable image sizes without resizing, and (iv) robustness to geometric transformations without data augmentation, demonstrated on MNIST, FashionMNIST, and CIFAR10. Results show strong performance at low $\mathrm{SNR}$ and improved accuracy with more superpixel nodes, indicating practical potential for noise-resilient TOC in resource-constrained wireless systems.

Abstract

Traditional approaches to semantic communication tasks rely on the knowledge of the signal-to-noise ratio (SNR) to mitigate channel noise. Moreover, these methods necessitate training under specific SNR conditions, entailing considerable time and computational resources. In this paper, we propose GeNet, a Graph Neural Network (GNN)-based paradigm for semantic communication aimed at combating noise, thereby facilitating Task-Oriented Communication (TOC). We propose a novel approach where we first transform the input data image into graph structures. Then we leverage a GNN-based encoder to extract semantic information from the source data. This extracted semantic information is then transmitted through the channel. At the receiver's end, a GNN-based decoder is utilized to reconstruct the relevant semantic information from the source data for TOC. Through experimental evaluation, we show GeNet's effectiveness in anti-noise TOC while decoupling the SNR dependency. We further evaluate GeNet's performance by varying the number of nodes, revealing its versatility as a new paradigm for semantic communication. Additionally, we show GeNet's robustness to geometric transformations by testing it with different rotation angles, without resorting to data augmentation.

Figures (12)

• Figure 1: General semantic communication paradigm. Previous methods need to set specific SNR.
• Figure 2: Our GeNet model architecture.
• Figure 3: Examples of the superpixel graph generation.
• Figure 4: Examples showing the loss of color information.
• Figure 5: Training loss and validation accuracy of GeNet with different GNN models.