IPP-Net: A Generalizable Deep Neural Network Model for Indoor Pathloss Radio Map Prediction

TL;DR

This paper introduces IPP-Net, a deep neural network based on UNet architecture, capable of accurately predicting indoor pathloss radio maps by learning from simulation data and 3GPP models, demonstrating competitive performance in a challenge.

Contribution

The paper presents a novel generalizable deep learning model for indoor pathloss prediction that integrates simulation data and standard models, advancing indoor radio mapping techniques.

Findings

Achieved a weighted RMSE of 9.501 dB in the ICASSP 2025 challenge.

Secured second place overall in the indoor radio map prediction challenge.

Demonstrated the effectiveness of combining simulation and model-based data for accurate predictions.

Abstract

In this paper, we propose a generalizable deep neural network model for indoor pathloss radio map prediction (termed as IPP-Net). IPP-Net is based on a UNet architecture and learned from both large-scale ray tracing simulation data and a modified 3GPP indoor hotspot model. The performance of IPP-Net is evaluated in the First Indoor Pathloss Radio Map Prediction Challenge in ICASSP 2025. The evaluation results show that IPP-Net achieves a weighted root mean square error of 9.501 dB on three competition tasks and obtains the second overall ranking.