Recurrent MIMO CoFAR Probing Waveform Design for Learning-Aided Clutter Estimation

A cognitive fully adaptive radar system (CoFAR) repre-sents an advanced radar architecture predicated on the principles of sensing, learning, and adaptation. However, the efficacy of such an adaptive radar system hinges upon a comprehensive understanding of its operational environment. The presence of unwanted echoes stemming from clutter can significantly impede the performance of a CoFAR. To address this challenge, this study introduces a sparse Bayesian learning framework aimed at estimating the underlying joint sparse clutter channel impulse response. Additionally, a recurrent transmit probing waveform is devised to minimize the resultant mean square error in subsequent iterations. Leveraging the majorization-minimization framework, we derive a closed-form expression for the overall waveform design vector. Waveform optimization brings the radar a step closer to cognitive mode of operation, enabling it to adaptively learn and adjust its parameters in response to changing environmental conditions. Extensive numerical simulations validate our analytical formulations and illustrate the superior performance of the proposed methodology compared to scenarios where optimal waveform design is not implemented.