Covariance Matrix Estimation of the nonhomogeneous and non-Gaussian Clutter

For the problem of the radar targets detection, the clutter data around the cell under test is used for the estimation of the clutter covariance matrix. In this paper, we consider the covariance matrix estimation of compound Gaussian clutter of nonhomogeneous environments, i.e., the training samples used for adaptation do not share the same covariance matrix as the vector under test, and the clutter can be modeled in terms of a compound Gaussian process. The conventional covariance matrix estimations have poor performance under this environment. In this paper, the conjugate prior distributions of unknown parameters of clutter statistics are used to descript the nonhomogeneous environments, and based on Bayesian framework, the minimum mean square error estimation of clutter covariance matrix is proposed, using Gibbs sampler. The computer simulation is used to validate the method proposed in this paper, and the results show that the method of the covariance matrix estimation proposed in this paper is better than conventional ones, especially within a small number of training samples and coherent pulses. The impact of the error of parameters of prior distributions on the detection performance is also analyzed in the end of the paper.