Combined wavelet transform and banded-matrix iterative approach/canonical grid methods for solving large-scale electromagnetic problems

Combining the Daubechies wavelet transform and banded-matrix iterative approach/canonical grid method, an efficient algorithm for solving large-scale electromagnetic problems is proposed. Based on the concept of the multiresolution analysis, a banded-matrix is compressed so as to reduce the required storage and speed up the computational time using the sifting window scheme. Using Monte Carlo simulations, the scattering of near-grazing incident waves by the one-dimensional lossy dielectric rough surfaces up to several hundred wavelengths is then investigated.