Target Recognition with Adaptive Waveforms in Cognitive Radar using Practical Target RCS Responses

In this paper, we utilize high-fidelity electromagnetic-simulated RCS responses in a cognitive radar (CRr) platform performing target recognition. Previous works used arbitrarily generated target responses consisting of a few frequency resonances which are distinct across different targets. However, realistic target responses contain rich frequency components characterized by physical scattering centers of the target. It is therefore imperative to build on prior works by considering practical target responses. We utilize an improved waveform design technique known as probability-weighted energy (PWE) over classical spectral variance methods such as probability-weighted spectral variance (PWSV). Our results showed an improvement in classification performance of SNR and mutual information (MI)-based waveforms used in conjunction with PWE and PWSV update methods over receiver-adaptive wideband pulsed waveform using a CRr platform. In this work, we also consider a more complex case where the target's azimuth angle has some deviation such that the response from that target is not deterministic but rather from an ensemble of different responses as dictated by aspect angle uncertainty.