Radar target identification with relative amplitudes and unknown target aspect angles

Abstract

An approach to radar target identification that uses a modified nearest neighbor algorithm and relative amplitude target features is proposed. It assumes that the aspect angles (elevation and azimuth) of the target to be identified are unknown. The relative amplitude features are the ratios of the target amplitude returns of two different frequencies, and they are two-dimensional functions of the target aspect angles. The features are also related to target geometry, wave polarization, and frequency, but are independent of target range and other radar system parameters. The database contains a finite number of samples of reference target features taken at certain target aspect angles. The differences between the aspect angles of the unknown target and those of the samples, called the aspect offsets, cause identification error. An unknown target is considered identified if its nearest neighbor in the feature space is in the set of feature samples from the same reference target in the database. A computer simulation was conducted to determine the misidentification percentage, due to the aspect offsets, as a function of the aspect angle sampling intervals and the number of frequencies, based on amplitude returns from multiple point scatters. >