The characteristic basis function method (CBFM) based on large-size blocks with multilevel subdivision is proposed to improve the analysis of electromagnetic scattering from dielectric objects. The small-size blocks strategy is widely used to alleviate the computational burden on generating the characteristic basis functions (CBFs) in the perfectly electrical conducting (PEC) object problems. However, when transplanted into the dielectric objects, this strategy performs poorly according to our empirical results. For adapting the CBFM to dielectric objects, the large-size blocks strategy is proposed for the first time. Primarily, it remarkably reduces the number of the CBFs required for the reduced matrix. Furthermore, the multilevel subdivision on the blocks is adopted to convert the full rank impedance matrices of the blocks into hierarchical matrices, leading to great advantage in two aspects. In the CBFs generation, the self-impedance matrices with hierarchical structure can be easily inversed by a recursive inversion algorithm based on the Sherman-Morrison-Woodbury (SMW) formula, which significantly cuts down the consumption in terms of both CPU time and memory requirement. In addition, the hierarchical structure is taken advantage of by the multiscale adaptive cross approximation (MS-ACA) to accelerate the filling of the mutual-impedance matrices of the adjacent blocks. The numerical results of several dielectric objects are presented to demonstrate the accuracy and efficiency of the proposed method.
10 March 2025 in Microwave and Optical Technology Letters
The improved hybrid discretisation of volume integral equation (VIE) combined with the characteristic basis function method is proposed to analyse the scattering from complex dielectric objects in this study. To mitigate the excessive number of discrete elements which arise from the conformal discretisation on the complex dielectric objects, the hybrid discretisation is developed. For the conventional hybrid discretisation, the conformal Schaubert–Wilton–Glisson (SWG) basis functions are used to discretise the homogeneous regions, meanwhile, the non‐conformal half‐SWG basis functions are employed to discretise the boundary regions which separate different materials or multiscale structures. The hybrid discretisation takes advantage of the merits of both conformal and non‐conformal basis functions and improves the computational efficiency for solving the VIE. In our study, a kind of edge basis function defined within boundary tetrahedral elements substitutes for the SWG basis functions to model the boundary regions, leading to the further reduction on the number of basis functions. Moreover, the characteristic basis functions extracted from the low‐level basis functions are employed to construct the reduced matrix of which the size is considerably smaller than that of the system matrix derived from the VIE. Several numerical results are presented to demonstrate the accuracy and efficiency of the proposed method.