Tradeoffs in multivariable filter design with applications to fault detection

This paper investigates sensitivity limitations in multivariable linear filtering. Integral constraints on appropriate complementary sensitivities are developed that give new insights into fundamental design tradeoffs that apply to all linear filters. These results can be used to obtain performance limitations in filtering applications where multivariable structural features are of particular interest. We illustrate our results by exploring design tradeoffs in problems of fault detection and isolation.