Toward an Integrated Framwork for Data-Efficient Parametric Adaptive Detection.

摘要

The conjugate-gradient (CG) algorithm is investigated for reduced- rank STAP detection. A family of CG matched filter (CG-MF) is developed by using the k-th iteration of the CG in solving the Wiener-Hopf equation. The performance the CG-MF detectors is examined for two cases. The first involves an arbitrary covariance matrix. It is shown that each CG-MF detector 1) yields the highest output SINR and smallest MSE among all linear solutions in the Krylov subspace; and 2) is CFAR with non-decreasing detection probability as k increases. The second is a structured case frequently encountered in practice, where the covariance matrix contains a rank-r component due to dominant interference sources, a scaled identity due to the presence of white noise, and a perturbation component containing the residual interference and/or due to the estimation error. It is shown via a perturbation analysis that the (r+1)-st CG- MF detector achieves an output SINR nearly identical to that of the optimum MF detector which requires full iterations of the CG algorithm. Finally, the CG algorithm is used to solve a linear prediction problem in the parametric adaptive matched filter (PAMF). It is shown that the PAMF can be casted within the framework of reduced-rank STAP detection.