GPU-Based Near Real-Time Estimation of the Human Body Penetrating Low-Frequency Magnetic Fields Using Free Space Field Measurements

Low-frequency magnetic fields emanated by wireless power transfer systems expose human beings in their vicinity and induce electric fields into the biological tissues. These potentially harmful induced body-internal electric fields should not exceed the ICNIRP recommended strength limits. The Scalar Potential Finite Difference scheme can be utilized to determine these induced electric fields and thus to assess the human exposure but requires the magnetic flux density distribution in the volume of the exposed body as input. The goal of this work is to determine this magnetic flux density distribution from a few in-situ free-space measurements using graphics processing unit (GPU) accelerated interpolation schemes in near real-time. A single NVIDIA Tesla A100 GPU allows a magnetic flux density distribution reconstruction with an ICNIRP recommended human voxel model resolution of 2 mm in less than 1 second based on 19 sampling points using radial basis functions.