Asymptotic Maximum Likelihood Estimation of Multiple Radar Targets: performance in the presence of model mismatch

This work deals with the problem of estimating the directions of arrival (DOA) and the complex amplitudes of multiple targets present in the same range-azimuth resolution cell of a surveillance radar. The DOA asymptotic maximum likelihood (AML) is first derived by maximizing the asymptotic (large sample size) likelihood function, assuming a deterministic model for the unknown target amplitudes. The performance of the proposed estimator and that of the ML estimator are investigated and compared resorting to Monte Carlo simulation. In particular, here their performances are investigated in the presence of a model mismatch, simulating a scenario where the complex amplitudes randomly fluctuate according to the Swerling I target model. The robustness of the AML and ML estimators is assessed by comparing their mean square estimation errors with the Cramér–Rao lower bound calculated by taking into account of the target amplitudes randomness.