We present a low resolution method for radar target identification. The method employs optimized waveforms in cross correlation with the return from a low frequency chirp to distinguish between similar candidate targets. Target signature waveforms were generated every degree in azimuth for generic wing-body-tail targets using FDTD simulation. The targets were four to five meters in length with approximately two meter wing spans. The transmitted linear chirp had a 20% band width and 250 MHz center frequency. We generated candidate waveforms using a simple chaotic map to cross correlate with the target returns. The waveforms were generated by passing chaotic time series through a bandpass filter. Alternately the waveforms were constructed by concatenating constant amplitude sinusoids whose periods were specified by the amplitudes of the chaotic time series. In both cases waveforms were constructed with the same bandwidth and center frequency as the transmitted chirp. A large number of test waveforms were generated by random variation of the generating chaotic map parameters. These test waveforms were cross correlated with the return waveforms from two similar targets. Waveforms that maximized (minimized) the cross correlation amplitude ratio difference between targets were retained and fine tuned with simplex optimization. Optimization was conducted over target azimuthal windows up to 10° in width. Best maximizing and minimizing test waveforms were retained for each data window. Using this method pairwise discrimination between candidate targets could be achieved over most aspects where the signature of the respective targets are not varying too rapidly with angle.
展开▼
