Maximum likelihood angle and velocity estimation with space-time adaptive processing radar

Airborne surveillance radar performance can be improved with space-time adaptive processing (STAP) to cancel ground clutter and interference. This paper considers maximum likelihood (ML) angle and velocity estimation for airborne radar employing STAP. The ML estimator requires a two-dimensional optimization. A computationally efficient quasi-Newton approach is proposed, whereby a positive definite approximate Hessian is formed using only the secondary data processed by three adaptive space-time filters. The algorithm is naturally initialized by the target detection location within a coarsely spaced angle-Doppler filter bank. Monte-Carlo simulations show that the new algorithm nearly achieves the Cramer-Rao bound and outperforms conventional one-dimensional estimators, which suffer from location-dependent biases when employed in STAP scenarios.