In this paper we consider the problem of robust signaldetection in the presence of Gaussian disturbance with unknowncovariance matrix. We design and assess three new robustadaptive detectors, capable of operating in the presenceunknown discrepances between the nominal and the actual steeringvector. Remarkably the new decision rules exhibit a boundedConstant False Alarm Rate (CFAR) behavior and allow, throughthe regulation of a design parameter, to trade off target sensitivitywith sidelobes energy rejection. Finally computer simulationsshow that the proposed detectors achieve a visible performanceimprovement, in terms of target sensitivity, over the traditionaladaptive detection algorithms expecially in the presence of severesteering mismatches.
展开▼