The performance of three UHF-band antennas that operate near the human body are investigated by using high-fidelity pixel-based anatomical body models (the AustinMan and AustinWoman models) and a fast integral-equation method. The surface-volume integral equation pertinent to the analysis is solved by employing mixed volumetric rooftop and tetrahedral basis functions, an FFT-accelerated iterative algorithm, surface pre-conditioning, and parallelization using multiple workload decomposition techniques. Detuning, antenna pattern changes, and the dependence of the results on the fidelity of the human model are quantified as a function of antenna position for a dipole antenna near the heads, a patch antenna near the wrists, and a spiral antenna along the digestive track of each body model.
