TransPathNet: A Novel Two-Stage Framework for Indoor Radio Map Prediction

Xin Li1, Ran Liu1, Saihua Xu2, Sirajudeen Gulam Razul2, Chau Yuen1
1 School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore
2 Temasek Laboratories @ NTU, Nanyang Technological University, Singapore
Paper Code

Abstract

The accurate indoor pathloss prediction is crucial for optimizing wireless communication in complex environments. This paper introduces TransPathNet, a novel two-stage deep learning framework that leverages transformer-based feature extraction and multiscale convolutional attention decoding to generate high-precision indoor radio pathloss maps. TransPathNet demonstrates state-of-the-art performance in the ICASSP 2025 Indoor Path Loss Radio Map Prediction Challenge, achieving an overall Root Mean Squared Error (RMSE) of 10.397 on the challenge full test set and 9.73 on the challenge kaggle test set, showing excellent generalization capabilities across different indoor geometries, frequencies, and antenna patterns.

Overview of TransPathNet Training Process

Overview of TransPathNet Training Process. The framework employs a two-stage architecture: a coarse stage and a fine stage. TransPathNet, utilized in both stages, integrates the TransNext and EMCAD to produce robust pathloss predictions.

Pathloss Prediction Comparison

Comparison of pathloss predictions under different configurations.

BibTeX

@inproceedings{li2025transpathnet,
  title={TransPathNet: A Novel Two-Stage Framework for Indoor Radio Map Prediction},
  author={Li, Xin and Liu, Ran and Xu, Saihua and Razul, Sirajudeen Gulam and Yuen, Chau},
  booktitle={Proc. IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP)},
  year={2025},
  month={April}
}