Efficient radar signature prediction using a frequency-aspectinterpolation technique based on adaptive feature extraction

A radar cross section (RCS) interpolation technique in both frequency and aspect is proposed for the efficient prediction of radar signatures from computational electromagnetics data. Our approach is based on a multiple-arrival model for the induced current on the target. The model parameters are determined by an adaptive feature extraction (AFE) algorithm, which uses an iterative search-and-extract procedure to find the individual model features. Random frequency and aspect sampling is used to circumvent the ambiguity in selecting the features. Numerical examples are presented to test the interpolation algorithm. It is found that sufficient accuracy in the predicted radar features can be achieved even when the original computed data is sampled at 5:1 below the Nyquist criterion in either frequency or aspect. The algorithm is also applied to efficiently predict the radar images of the benchmark VFY218 airplane at UHF band with good results