Adaptive robust constrained matched filter and subspace detection

A minimax methodology for formulating adaptively robust and sensitive matched filter and subspace detectors is provided, where the signal and interference of structured noise are partially known. The signal and interference subspaces are assumed to reside in conical regions of the measurement space, and the gain parameters can have bounded magnitude. It is shown that the minimax approach permits the design of a rich variety of matched filter and subspace detectors that vary in the degree of robustness and sensitivity.