CFAR Analysis of Non-Coherent Detectors in Compound-Gaussian Clutter with Inverse Gaussian Texture

In this paper, we investigate the constant false alarm ratio (CFAR) property of non-coherent detectors in correlated compound-Gaussian clutter with inverse Gaussian texture. In order to ensure the CFAR property with respect to the clutter power mean, we give a novel parametrization form of the existing inverse Gaussian distribution. The probability of false alarm (Pfa) expressions of the non-coherent detectors are derived and show that two detectors have CFAR with regard to the scale parameter standing for clutter power mean. It is found that the threshold multipliers depend on the clutter covariance matrix structure and the shape parameter. For realizing the CFAR detection in the overall observation space, the correlated clutter is decorrelated via whitening and the threshold multipliers are obtained via looking up tables for different clutter parameters, radar parameters and the selected Pfa. The experimental results indicate that two non-coherent detectors can ensure the CFAR detection as desired.