An improved multilevel Green's function interpolation method with adaptive phase compensation

An improved multilevel Green's function interpolation method (MLGFIM) with adaptive phase compensation (APC) is proposed. The difficulty in applying interpolation approaches to the fast varying phase term in the integral equation kernel for full-wave electromagnetic (EM) simulations is eradicated by using the phase compensation and adaptive direction separation (ADS). The multilevel tree structure in MLGFIM keeps the number of direction separation invariant at all levels, attributing to the recursive interpolation with multilevel phase compensation. The proposed MLGFIM-APC in conjunction with the Lagrange-Chebyshev interpolation yield an O (N log N) CPU time and O (N) computer memory requirement for surface scattering problems. By introducing a transition level, the MLGFIM-APC can adaptively incorporate interpolation techniques of conventional interpolation (CI), transition interpolation (TI) and phase-compensation interpolation (PI), corresponding to electromagnetic simulation of problems of small, moderate, and large electrical sizes, respectively. Large-scale microstrip antenna arrays are simulated to illustrate the accuracy and efficiency of the proposed method. It is found that the CPU time scales better than O(N log N) for these co-planar problems.