Nonintrusive Polynomial Chaos Expansion Method for Uncertainty Quantification in Multivariable Specific Absorption Rate Calculations

The evolution of wireless communication technologies, from 2G to 5G, has continuously raised concerns about health risks from electromagnetic exposure. Accurate assessment of specific absorption rate (SAR) in human tissues from wireless devices is essential. While traditional SAR compliance currently relies on costly dosimetric measurements, numerical simulations face challenges in uncertainty quantification (UQ). Monte Carlo simulation is computationally expensive with slow converge and inadequate for SAR application. The nonintrusive polynomial chaos (NIPC) expansion method offers an interesting alternative. Using the linear regression approach, which relies on oversampling to determine the expansion coefficients, the efficiency of NIPC decreases with increasing number of variables. The Galerkin projection method, although mathematically intricate, provides a robust solution for computing these expansion coefficients. This article identifies the tradeoff between Galerkin and regression-based NIPC methods and highlights their applicability in UQ for SAR modeling in heterogeneous head structures.