The core algorithm from my PhD thesis for automatically finding worst-case electromagnetic exposure regions on realistic 3D body surfaces. Presented at SoftCOM 2023.

The Problem

Traditional RF safety standards evaluate power density on flat surfaces, but human bodies are anything but flat. When assessing exposure from 5G and millimeter-wave devices, using planar approximations can miss the actual peak exposure on curved anatomical regions.

The Solution

This algorithm takes a point cloud of any body surface along with power density values at each point, then automatically identifies where the peak spatial-average power density occurs. No manual region selection needed.

Features

  • Works with arbitrary curved surfaces (not just planes)
  • Minimal input: just a point cloud and power density values
  • Automatically finds worst-case exposure regions
  • Includes demo with realistic human head model
  • Jupyter notebook tutorial for getting started

Installation

git clone https://github.com/akapet00/pspd-autodetect.git
cd pspd-autodetect
pip install -r requirements.txt