Abstract:Over the last years, several works have explored the application of deep learning algorithms to determine the large-scale signal fading (also referred to as ``path loss'') between transmitter and receiver pairs in urban communication networks. The central idea is to replace costly measurement campaigns, inaccurate statistical models or computationally expensive ray-tracing simulations by machine learning models which, once trained, produce accurate predictions almost instantly. Although the topic has attracted attention from many researchers, there are few open benchmark datasets and codebases that would allow everyone to test and compare the developed methods and algorithms. We take a step towards filling this gap by releasing a publicly available dataset of simulated path loss radio maps together with realistic city maps from real-world locations and aerial images from open datasources. Initial experiments regarding model architectures, input feature design and estimation of radio maps from aerial images are presented and the code is made available.

What problem does this paper attempt to address?

### Problems the Paper Attempts to Solve This paper primarily aims to address the issue of estimating path loss in wireless communication networks. Specifically, the paper tackles this problem through the following points: 1. **Limitations of Existing Methods**: - Statistical path loss models rely solely on the distance between the transmitter and receiver, failing to accurately capture the radio wave propagation in specific environments. - Ray tracing simulations, while capable of high-precision physical phenomenon modeling, are impractical for real-time or large-scale applications due to their long runtime and the need for detailed 3D city models. 2. **Application of Machine Learning Methods**: - Utilizing deep learning algorithms, particularly Convolutional Neural Networks (CNNs), to predict radio maps. Once trained, these models can make accurate predictions within milliseconds, making them suitable for scenarios where statistical path loss models are too inaccurate and ray tracing is too slow. 3. **Lack of Public Datasets**: - There is a current lack of open datasets and code repositories for researchers to test and compare different methods and algorithms. To address this, the authors have released a public dataset containing simulated path loss maps and real-world city maps, along with preliminary experimental results and code. 4. **New Contributions**: - A new radio map dataset is proposed, including the urban geometry of real locations in Berlin and numerous path loss maps generated by ray tracing simulations. This dataset introduces directional antennas for the first time and includes information such as trees and realistic building heights. - The experimental section explores different CNN architectures and compares various input feature encoding methods, including urban geometry, relative positions of transmitters, and their antenna characteristics. - Preliminary research is also conducted to predict the entire radio map from limited information such as aerial images, a challenging problem that has not been sufficiently explored before. In summary, this paper aims to improve the accuracy of path loss prediction through machine learning methods and provides a benchmark dataset to facilitate further research in the related field.

Radio Map Estimation -- An Open Dataset with Directive Transmitter Antennas and Initial Experiments

Radio Map Prediction from Aerial Images and Application to Coverage Optimization

Dataset of Pathloss and ToA Radio Maps With Localization Application

Radio Map Estimation: Empirical Validation and Analysis

EM DeepRay: an Expedient, Generalizable, and Realistic Data-Driven Indoor Propagation Model

RadioUNet: Fast Radio Map Estimation with Convolutional Neural Networks

Radio Map Estimation Using a Generative Adversarial Network and Related Business Aspects

Supreme: Fine-grained Radio Map Reconstruction Via Spatial-Temporal Fusion Network

Radio Map Estimation: A Data-Driven Approach to Spectrum Cartography

High-Efficiency Urban 3D Radio Map Estimation Based on Sparse Measurements

A Radio Wave Propagation Modeling Method Based on High Precision 3D Mapping in Urban Scenarios

A Radio Wave Propagation Modeling Method Based on High-Precision 3-D Mapping in Urban Scenarios

A New Approach to Predict Radio Map Via Learning-Based Spatial Loss Field

RADIANCE: Radio-Frequency Adversarial Deep-learning Inference for Automated Network Coverage Estimation

Model and Transfer Spatial-Temporal Knowledge for Fine-Grained Radio Map Reconstruction

RadioDiff: An Effective Generative Diffusion Model for Sampling-Free Dynamic Radio Map Construction

Radio Map Estimation with Deep Dual Path Autoencoders and Skip Connection Learning

Theoretical Analysis of the Radio Map Estimation Problem

A FL-Based Radio Map Reconstruction Approach for UAV-Aided Wireless Networks

Development of Machine Learning-Based Radio Propagation Models and Benchmarking for Mobile Networks

An Efficient Radio Map Learning Scheme Based on Kernel Density Function