Vision Transformers for Efficient Indoor Pathloss Radio Map Prediction
This work addresses indoor wireless network planning, but it is incremental as it applies an existing vision transformer method to a specific domain with optimizations.
The paper tackles indoor pathloss prediction for wireless network planning by using a vision transformer with DINO-v2 pretrained weights to generate pathloss maps from floor plans, showing that extensive augmentation improves generalization and feature engineering helps in low-data scenarios.
Indoor pathloss prediction is a fundamental task in wireless network planning, yet it remains challenging due to environmental complexity and data scarcity. In this work, we propose a deep learning-based approach utilizing a vision transformer (ViT) architecture with DINO-v2 pretrained weights to model indoor radio propagation. Our method processes a floor map with additional features of the walls to generate indoor pathloss maps. We systematically evaluate the effects of architectural choices, data augmentation strategies, and feature engineering techniques. Our findings indicate that extensive augmentation significantly improves generalization, while feature engineering is crucial in low-data regimes. Through comprehensive experiments, we demonstrate the robustness of our model across different generalization scenarios.