Optimal sequential waveform selection for radar target imaging and classification

High resolution radar technology has renewed the interest in signal design and processing for wideband radar target imaging and recognition. We have used a high resolution radar echo formation model to construct signals that sequentially estimate the target reflectivity function or its class membership in an optimal way.;For the radar imaging problem, we present an optimal signal selection scheme and the associated echo processing method when the target class is known. First, we derive the optimal waveforms for a noiseless observation model. An optimality criterion as a distance measure between the actual and the reconstructed target reflectivity functions is defined. A sequence of signals is obtained that progressively refines the target image. Second, we present a discretized model for the radar echo formation process. This finite-dimensional model is used to derive optimal signal selection strategy in a noisy observation model. The same finite-dimensional model is then used to derive a combined imaging and classification scheme in the case when the target class is initially unknown. The scheme sequentially estimates the target reflectivity function while it simultaneously attempts to identify the target class.;We apply a sequential experiment design procedure to the problem of signal selection for pure radar target classification. The waveforms are designed to minimize the decision time by maximizing the discrimination information in the echo signal. We discuss two strategies for the selection of waveforms in the two-class case. First, when the target environment is assumed fixed during illumination, a library of waveforms is obtained from the reflectivity functions of the possible targets. During actual classification, the sequence in which the waveforms are selected from the library is determined from the noise to clutter power in the range-Doppler support of the target. In the second strategy, when the target environment changes between pulse transmissions, the maximum discrimination information is obtained by a repeated transmission of a single waveform designed from the reflectivity function of the targets.;We perform computer simulations to verify our results and to show the gains in performance that can be obtained from the use of these algorithms.