On relationship between traditional and knowledge-based clutter covariance estimate

Recently, the knowledge-based clutter covariance estimate methods are developed to improve the convergence rate. In this paper, the relationship between some widely-used knowledge-based methods and the traditional reduced-order methods are established. It is found that the colored loading (CL) is equivalent to the pre-whitened diagonal loading (DL), and the fast maximum likelihood with assumed clutter covariance (FMLACC) is equivalent to the pre-whitened principal component (PC) method. These equivalences suggest that the convergence rate of the CL and FMLACC method will be on the order of twice the effective rank of the pre-whitened clutter covariance matrix. The conclusion is verified by simulations.