Incoherent Detection of Targets with Unknown Doppler Shift in Spiky K-distributed Clutter

In this paper, the problem of the incoherent detection of targets with unknown parameters in K-distributed clutter is tackled. Multi layer Perceptron (MLP) based detectors are designed to approximate the Average Likelihood Ratio (ALR) based one, which implements the Neyman-Pearson optimum detector for the defined composite hypothesis test problem. Conventional solutions such as the linear law and Constant False Alarm Rate (CFAR) detectors are used as references for evaluating the performance improvement achieved by the MLP solutions. Estimated Receiver Operating Characteristic (ROC) and detection curves show that the MLP detector outperforms the reference ones in all the considered cases of study.