Opportunity: Physics-based simulation and synthetic data generation for distributed RF sensing

Researchers developing AI/ML methods for distributed radio-frequency (RF) sensing need diverse, accurately labeled datasets to train and evaluate detection, localization, and tracking models. Experimental data collection is costly and time-consuming, often covers limited receiver geometries and propagation conditions, and may lack precise ground truth. These gaps make it difficult to assess whether models will perform reliably in new environments. This project addresses this challenge by developing and verifying a reusable simulation workflow that balances physical fidelity and computational cost. Using electromagnetic solvers such as COMSOL Multiphysics with its RF Module or Ansys HFSS alongside ray-tracing tools – where appropriate, the workflow will generate synthetic sensor data across varied environments and configurations. The ideal outcome is a documented simulation testbed that produces customizable, labeled datasets with known ground truth within each simulated scenario, supporting model training and systematic performance evaluation.