Abstract

Overview

The rapid proliferation of wireless technologies necessitates the precise estimation of electromagnetic field distribution, commonly depicted through electric field strength across different geographical areas.

Objective

The research aims to determine the most effective geospatial model for generating a national-level electric field strength map within the 30 MHz-6 GHz frequency range.

Methodology

• Employed five different methodologies for constructing the electric field strength map.
• Four methodologies are based on Gaussian process regression.
• One methodology uses the classical weighted-average method of the nearest neighbor.

Findings

Gaussian process spatial models, also known as Kriging models, generally outperform other methods. However, the performance of classical nearest neighbor models becomes comparable after excluding some outliers, indicating both approaches' effectiveness depending on data quality and outliers.

Conclusion and Future Directions

This research is a foundation for future implementations in mapping EMR exposure. Subsequent research will explore urban infrastructure characteristics and incorporate temporal patterns and environmental data to enhance predictive capabilities. Further studies will also focus on model interpretability and conduct validation studies to ensure real-world applicability.