Indoor Environments and Ray-Tracing Data

This dataset contains the data used in our paper "Radio Map Prediction from Noisy Environment Information and Sparse Observations" (https://arxiv.org/abs/2602.11950, submitted to IEEE). The corresponding code to the paper can be found on https://github.com/fabja19/RML_indoor. The code used to generate and process the indoor environments is contained in the repo https://github.com/fabja19/WI_indoor_projects. The dataset contains indoor environments compatible with the ray-tracing software Wireless InSite, radio maps simulated in these environments and images representing the environments in 2D slices at different heights, usable as input images for CNN. We also include 10 copies per environment with noise in the locations of Tx and objects and the material properties, and the corresponding raster images. Furthermore, checkpoints and logs for some of the trained models are contained. If you use the code or dataset in your work, please cite: Fabian Jaensch, Çağkan Yapar, Giuseppe Caire and Begüm Demir, "Radio Map Prediction from Noisy Environment Information and Sparse Observations", arXiv:2602.11950. @misc{jaensch2026radiomappredictionnoisy, title={Radio Map Prediction from Noisy Environment Information and Sparse Observations}, author={Fabian Jaensch and Çağkan Yapar and Giuseppe Caire and Begüm Demir}, year={2026}, eprint={2602.11950}, archivePrefix={arXiv}, primaryClass={eess.SP}, url={https://arxiv.org/abs/2602.11950}, } File Structure dataset_publication.zip contains environment encodings and radio maps used for training and testing CNNs indoor_ckpts.zip model weights and log files for the model trained with binary inputs, with and without SNDA, and the baseline model trained without environment information project_files.zip contains the Wireless InSite projects for the ray-tracing simulations