Balanced realization of separable-denominator multidimensional systems (Q1260800)

The authors present a general framework for internal balancing \(n-D\) (separable-denominator) systems and model order reduction using the definition of generalized Gramians proposed by \textit{W. S. Lu, E. B. Lee} and \textit{Q. T. Zhang} [Proc. 18th 1986 Int. Symp. Circuits Systems 79-82 (1986)]. It is shown that the internally balanced realizations for separable-denominator \(n-D\) systems can be obtained by solving certain coupled Lyapunov equations, performing singular value decomposition, and formating the internal balancing transformation matrix. The proposed balancing and order reduction technique is illustrated by applying it to obtain low-order approximations of Gaussian and Laplacian-Gaussian infinite-impulse-response filters.