A generalized forward-backward method for the efficient analysis of large array problems

Development of novel analysis techniques to treat large array problems remains attractive due to the increasing need to employ very large array structures in practical applications. Recently, a forward-backward method (FBM), has been extended to treat large array problems; several unique advantages have been demonstrated. The previous approach of FBM requires monotonic variation on the array surface in order to set up a forward-backward procedure. This limitation is overcome by a generalization of FBM (GFBM) (see Pino, M.R. et al., 1999), which allows arbitrary array elements in general shapes. In contrast to the previous FBM (see Chou, H-T, IEE Proc.-Microw. Antennas Propag., vol.147, no.3, 2000) where the current computation sweeps cell by cell (each cell corresponding to a MoM basis of pulse function) in the forward or backward procedure, the GFBM sweeps the current computation element by element (corresponding to array elements). A simple MoM procedure with very few unknowns (usually only these unknowns to model three elements at most) is employed to obtain the forward current and its backward correction on the selected element. This modification allows arbitrary array elements to be treated accurately. Numerical results demonstrate its validity.