Multistatic target classification with adaptive waveforms

Traditional radar systems rely on a predefined suite of waveforms and post-measurement signal processing to achieve such goals as target detection, classification, and tracking. Cognitive radar (CR) is a newly proposed framework in which the radar actively interrogates the propagation channel and adapts its operating parameters in order to maximize performance. We apply CR to a target classification problem by calculating custom waveforms that maximize the information received from the target echoes. A new MIMO waveform is proposed which maximizes the mutual information between a Gaussian random target and the received data under AWGN. The results indicate that the new frequency-domain formulation offers superior performance compared to both a non-adaptive approach and an ad hoc application of spectral waterfilling to the MIMO setting.