부산대학교 도서관

The privacy and security of sensitive data are critical concerns when outsourcing computational tasks to external cloud service providers. This paper explores the feasibility of extracting confidential simulated models from method of moments (MoM) matrices. The study focuses on impedance matrices employing Rao-Wilton-Glisson (RWG) functions and the electric field integral equation. Reverse-engineering the simulated geometry can be formulated as an optimization problem. However, the presence of dyadic Green's functions imposes ambiguities when simultaneously reconstructing both the relative angle and dis-tance between RWG source pairs in three-dimensional space. We introduce a local reconstruction method which mitigates these ambiguities. The approach gradually reconstructs the geometric structure locally by employing a reduced set of optimization variables. Numerical examples showcase that the local method can potentially reconstruct the original geometry.