ICMU2025 in Busan, Korea
Four research presentations at ICMU2025, with Best Paper Candidate award
Publications & Talks Awards
International Conference · 2025
On the Impact of Terrain Types in Deep Neural Network-Based Surrogates of Radio Maps
Abstract
Radio maps, encoding wireless propagation characteristics over a region of interest, have been harnessed as a powerful tool to optimize various aspects of wireless communication systems. Generation of wide-scale radio maps is often a time-consuming process, necessitating extensive wireless propagation simulations and/or measurement campaigns. This hinders its applicability to vehicle-to-everything (V2X) communications, where wireless propagation environment may frequently change over time due to vehicle mobility. Surrogate models by Deep Neural Networks (DNNs) have emerged as a potential solution for low-latency path-loss estimation, yet most of the existing models in the literature are tuned to specific terrain types, lacking generalizabity to different kinds of terrains. In this work, we aim to develop a terrain-aware surrogate model for radio-map prediction that can robustly handle diverse propagation environments. To this end, we first analyze the internal behavior of a state-of-the-art DNN backbone, RadioUnet, to examine how terrain types are reflected in its latent space. Visualizing bottleneck features with UMAP reveals three distinct clusters—urban areas, forested plains, and sloped terrains. We also observe that prediction errors increase in tree-covered areas and near steep slopes or building edges, suggesting the need for terrain-specific modeling. As a first step in this direction, we introduce a simple multitask framework where the model jointly predicts terrain class and path-loss using a shared encoder. This explicit terrain supervision improves prediction accuracy when the dataset or model size is small.
Research Note
A radio map records how much signal strength is lost at each point of a region, and it underpins decisions about base station placement and link adaptation. Producing one over a wide area normally requires extensive propagation simulation or measurement campaigns, which is slow. For vehicle-to-everything communication, where the propagation environment changes as vehicles move, that slowness is disqualifying.
Deep neural network surrogates offer low-latency path loss estimation instead, but most published models are tuned to one terrain type and generalise poorly to others.
This work approaches a terrain-aware surrogate by first examining how terrain is represented inside RadioUnet, a state-of-the-art backbone. Visualising its bottleneck features with UMAP, a technique that projects high-dimensional representations down to a viewable two dimensions, reveals three distinct clusters corresponding to urban areas, forested plains and sloped terrain.
Prediction errors were also found to grow in tree-covered areas and near steep slopes or building edges, which supports the case for modelling terrain explicitly. As a first step in that direction, a simple multitask framework was introduced in which a shared encoder predicts terrain class alongside path loss. Supervising terrain in this explicit way improved accuracy when the dataset or the model was small.
Related Lab Activities
