Clutter space-time covariance matrix estimate based on multi-polarised data

The problem of training sample insufficiency is frequently encountered in the space-time adaptive processing and significantly degrades the performance of radar target detection. In this study, the authors propose a clutter covariance matrix estimation algorithm using multi-polarised data in the polarimetric radar system, which can mitigate this problem with an enlarged training sample set. Based on the space-time signal model, they validate that the clutter snapshots for different polarisations share a common spectral structure theoretically. Then, the maximum likelihood estimations of clutter covariance matrixes with multi-polarised training samples are deduced under Gaussian and non-Gaussian statistic. Finally, the performance improvement of our method with limited training samples is demonstrated with simulated data.