Investigation On Near-field To Far-field Transformation Method For Near-field Measurements Of Electrically Large Objects

In pattern measurements of large antenna systems as well as the scattering characteristicsmeasurements of large targets such as aircrafts, satellites, ships, far-field condition is necessary for thetraditional far-field measurement, therefore the experimental conditions is impossible to satisfy. Whilethe near-field measurement can avoid this constraint condition, it has been paid widespread attentionrecently. Due to the large amounts of near-field data in near-field measurements of electrically largeobjects, high-efficiency near-field to far-field transformation methods are urgent to be studied.Focused on the near-field to far-field transformation method based on equivalent magneticcurrent and its fast algorithm, the research is carried out as follows:1. The principle and program implementation of the near-field to far-field transformation methodbased on equivalent magnetic current is deeply studied. The selection of basis function, discretizationof integral equation, the over-determined matrix equation solving by the singular value decomposition(SVD) and far field calculations of the equivalent magnetic current are studied. The validation of thecomputer programs is demonstrated by some radiation and scattering examples.2. Generalized Minimum Residual (GMRES) Method is studied to solve the over-determinedmatrix equation. The computational complexity of SVD method is O(M2N+MN2+N3), the GMRESmethod can reduce the computational complexity to O(N2+MN). Numerical results show that GMRESmethod accelerates the calculation efficiency and saves much memory remarkably.3. Fast Dipole Method (FDM) is studied to accelerate the Matrix Vector Product (MVP) inGMRES iterations, a simple Taylor’s series expansion and grouping scheme are used to transform theMVP into an aggregation-translation-disaggregation form naturally, which reduces the computationalcomplexity of MVP and memory requirements from O(MN) to O(M+N). Numerical results show thatFDM can accelerate the calculation efficiency and save much memory.