Group Equivariant Convolutional Networks for Pathloss Estimation

2025-11-21

AI Summary

To address the reliance on data augmentation and poor generalization in path loss estimation for wireless communications, this paper proposes RadioGUNet—a novel end-to-end deep learning model that integrates group-equivariant convolutions (G-Convs) with a U-Net architecture. RadioGUNet is the first to explicitly encode geometric symmetries—such as rotations and reflections—into path loss prediction via group equivariance, eliminating the need for data augmentation or preprocessing while inherently capturing channel spatial symmetries. Evaluated on the RadioMapSeer dataset, RadioGUNet achieves a 0.41 dB reduction in RMSE over a standard U-Net under comparable parameter counts, demonstrating significantly improved generalization and robustness. The core contribution lies in introducing group-equivariant convolutional layers into radio propagation modeling, thereby unifying physical symmetry priors with deep learning in a principled manner.

Technology Category

• Machine Learning: Calibration & Uncertainty Quantification
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Search and Optimization: Learning to Search

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Systems and Infrastructure for Web, Mobile and WoT: Data management and stream processing for Web, mobile and wireless applications

Abstract

This paper presents RadioGUNet, a UNet-based deep learning framework for pathloss estimation in wireless communication. Unlike other frameworks, it leverages group equivariant convolutional networks, which are known to increase the expressive capacity of a neural network by allowing the model to generalize to further classes of symmetries, such as rotations and reflections, without the need for data augmentation or data pre-processing. The results of this work are twofold. First, we show that typical UNet-based convolutional models can be easily extended to support group equivariant convolution (g-conv). Secondly, we show that the task of pathloss estimation benefits from such an extension, as the proposed extended model outperforms typical UNet-based models by up to 0.41 dB for a similar number of parameters in the RadioMapSeer dataset. The code is publicly available on the GitHub page: https://github.com/EricssonResearch/radiogunet

Problem

• Estimating pathloss in wireless communication using deep learning
• Extending UNet models with group equivariant convolution layers
• Improving accuracy without data augmentation through symmetry generalization

Innovation

• UNet-based framework for pathloss estimation
• Leverages group equivariant convolutional networks
• Extends models to support g-conv operations