Nonlinear decision rules for robust noncoherent integration

Abstract

New nonlinear decision rules for performing noncoherent integration (NCI) of multiple radar dwells are determined. Conventional noncoherent integration followed by CFAR can be greatly desensitized when the noise energy varies from dwell to dwell. The new rules are robust to dwell to dwell variations in noise energy, providing CFAR and minimizing detection loss. For slow fluctuation target models the locally most powerful decision rule outperforms conventional NCI even in stationary noise conditions. In addition, a composite test is provided that outperforms all individual binary integration (BI) M out of N tests independent of target fluctuation model and is suitable for distributed detection.