Fast Prediction of Human Safety Due to RF Exposure

Research into the biological effects of electromagnetic radiation has led to the development of high-performance modeling and simulation software. However, the high computational cost associated with running these tools make them impractical for use in analysis of mission level scenarios. Surrogate modeling provides a method to reduce detailed simulation results to approximate models, and would allow bioeffects to be estimated in near real-time. Without the tools needed to automate the process of training surrogate models from existing simulation software, a barrier exists to propagating results from fundamental research applications to wider use in DoD Modeling, Simulation, and Analysis applications. Nanohmics proposes to develop and demonstrate a novel framework for facilitating the automation of high-quality, computationally-efficient surrogate models from high-fidelity simulations. The framework will provide definitions and implementations for the building blocks of workflows optimized for training and analyzing data-driven surrogate models. These building blocks will capture the underlying operations involved with configuring and running all pre-processing, simulation, post-processing steps, as well as automate tasks associated with sampling the design space. Additionally, Nanohmics will develop interfaces for standard simulation and training tools to provide a ready path for transitioning research application workflows to mission level analysis.