Downloads & Free Reading Options - Results

This report investigates the bispectral processing of complex-valued radar target scattering signatures and the impact of such processing on target identification. The concept of 'birange profile' as two-dimensional display of target scattering mechanisms in the range domain is introduced. It is shown that the birange profile can detect responses due to multiple scattering interactions. An algorithm is proposed to estimate the birange profiles of radar targets. Using this algorithm, multiple interactions of real aircraft are extracted, and interpreted in terms of target geometry. An algorithm that uses autoregressive parametric modeling of radar signatures to obtain high resolution birange profiles was introduced and tested using synthesized data and real scattering data. An alternative estimation procedure, which does produce the true bispectrum, was developed. The classification aspect of birange features was investigated using cross-correlation of birange profile of unknown target with that of a catalog target. It was shown that classification using birange features is inferior to classification using impulse response features if the additive noise is Gaussian. Classification with the birange when the noise is non-Gaussian yields relatively reasonable results in comparison with other classifiers. Finally, it was shown that when scattering from extraneous point scatterers is added to the data, then classification with the birange outperforms other classifiers if the signal-to-extraneous scatterers ration is low. Keywords: Identification, Signal processing, Targets, Radar, Statistical.