Sensitivity Analysis of Knowledge Aided Detector in Non-Gaussian Clutter

Prior information can be used to improve detection performance of knowledge aided detectors, but the detection performance may be affected by the mismatches between the prior information and current clutter environment. In this paper, the knowledge aided detector in compound Gaussian clutter is considered, for the inverse Gamma distribution is used as the prior distribution of clutter texture component, and the detection performance of this detector is analyzed with different clutter texture component model parameters. First, false alarm rate and detection probability of Swerling I target are given under the condition of mismatched prior information parameters. Second, the impact on the detection performance with clutter texture distribution parameters is analyzed under the conditions of given prior information parameters. Theoretical analysis results show that when the distribution parameters of clutter texture component are located in some area, the detection performance could be better than that with the prior information matchs the clutter environment. The computer simulation validates the conclusion.