Robust fast maximum likelihood with assumed clutter covariance algorithm for adaptive clutter suppression

The mismatch between the clutter and noise power of the prior knowledge and true interference covariance matrix degrades the performance of fast maximum likelihood with assumed clutter covariance (FMLACC) algorithm significantly. By introducing a scale parameter to flexibly adjust the prior power, the authors propose an algorithm which is more robust to the power mismatch than FMLACC algorithm. They also develop a more straightforward method to derive the maximum likelihood covariance matrix estimator under this scaled knowledge constraint. Moreover, they study the problem of automatically determining the scale parameter. The authors provide two parameter selection methods, the first of which is based on estimating the minimum eigenvalue of the prewhitened sample covariance matrix, and the second is based on cross validation. To reduce the computational complexity, they also develop fast implementations for the parameter selection based on cross validation. Numerical simulations demonstrate the performance enhancement of the proposed algorithm compared with FMLACC algorithm in cases of mismatched prior knowledge.