Multitarget Parameter Estimation for Cognitive Radar Based on RIS

Cognitive radar can dynamically adjust the resource allocation according to the mission requirements and environments, so as to improve the capability of target detection, identification and tracking. In this paper, a reconfigurable intelligent surface (RIS) assisted cognitive radar is proposed, which controls the reflection coefficients of the RIS elements to alter the spatial energy distribution of signals, and finally reduces the uncertainty in the areas of interest. To enhance the response speed and robustness of the estimation module, a compressive sensing-based target parameter estimation model is developed by incorporating adaptive sparse Bayesian learning. By adopting the differential entropy of the target region as the loss function, RIS can serve as a feedback channel for cognitive radar, thereby enhancing the system’s estimation performance. The cognitive process is implemented by the multiple gradient descent algorithm that combines Frank-Wolfe solver and momentum gradient descent. Finally, simulation results show that the RIS can assist the radar to realize the cognitive function and improve the accuracy of multi-target parameter estimation.