Adaptive implementation of optimum MTI processor

The authors propose an adaptive processor implementing an optimum moving target indicator (MTI) in the sense of maximizing the improvement factor (IF). A multichannel scheme for preventing coefficient fluctuation due to the presence of target echoes or impulse interferences is also discussed. The proposed eigenfilter recursively estimates the minimum eigenvector of the input clutter covariance matrix based on the inverse power method (IPM). In the proposed scheme, the IPM iteration is performed in a numerically efficient way using systolic processing architectures. Simulation results demonstrate that the convergence speed of the eigenfilter is faster than that of the prediction error filters and IF values of the eigenfilter always approach the optimum. The multichannel adaptive eigenfilter is shown to be a robust scheme preventing coefficient fluctuation in the presence of target echoes. >