Abstract:In modern radar systems, precise target localization using azimuth and velocity estimation is paramount. Traditional unbiased estimation methods have utilized gradient descent algorithms to reach the theoretical limits of the Cramer Rao Bound (CRB) for the error of the parameter estimates. As an extension, we demonstrate on a realistic simulated example scenario that our earlier presented data-driven neural network model outperforms these traditional methods, yielding improved accuracies in target azimuth and velocity estimation. We emphasize, however, that this improvement does not imply that the neural network outperforms the CRB itself. Rather, the enhanced performance is attributed to the biased nature of the neural network approach. Our findings underscore the potential of employing deep learning methods in radar systems to achieve more accurate localization in cluttered and dynamic environments.

What problem does this paper attempt to address?

The paper investigates the problem of target localization in modern radar systems, particularly through estimation of azimuth and velocity. Traditional methods approach the Cramér-Rao lower bound (CRB) using gradient descent algorithms to achieve the theoretical limit of parameter estimation error. The paper proposes a data-driven neural network model and compares it with the gradient descent algorithm in realistic simulation scenarios, demonstrating the superior performance of the neural network in reducing mean squared error. However, this does not imply surpassing the CRB itself, but rather due to its inherent bias. The study emphasizes the potential of deep learning methods in radar target localization in complex and dynamic environments.

Data-Driven Target Localization: Benchmarking Gradient Descent Using the Cramer-Rao Bound