An Adaptive Target Detection Architecture for Mismatched Signals

This letter addresses the problem of adaptive target detection in the presence of possible mismatched sidelobe interfering signals assumed orthogonal to the nominal target signature in the whitened space. To this end, we devise a joint Maximum Likelihood (ML)-Bayesian based detector that simultaneously improves the target detection performance and the rejection capability of the mismatched signals. Specifically, we first inject an orthogonal interfering signal into the null hypothesis of the traditional binary hypothesis test and, then, solve it by means of the latent variable model and the Expectation Maximization algorithm. Finally, we maximize the posterior probability of the hypotheses for decision. In addition, we prove the Constant False-Alarm Rate property of the proposed detection architecture. The illustrative examples conducted on synthetic data corroborate the enhanced detection performance and rejection capability with respect to the state-of-the-art.