J. O. Coleman

FILES:

colemanCISS98.ps.gz

FORMAT:

gzipped postscript.

AUTHORS:

J. O. Coleman

TITLE:

Systematic Mapping of Quadratic Constraints on Embedded FIR Filters to Linear Matrix Inequalities

ABSTRACT:

This paper develops a flexible operator-based framework for the natural expression of application-specific quadratic constraints on the individual and joint design of FIR filters, along with an associated systematic mapping of those constraints to the linear-matrix-inequality (LMI) form compatible with a rapidly growing family of ultra-efficient numerical-optimization engines. This framework encourages, in particular, the constraint or minimization of the power gains of various embedding system configurations (possibly multirate) in response to specified input spectra--these gains are just application-specific squared L2 norms--and offers a straightforward, systematic approach to the design of real or complex embedded filters with linear or nonlinear phase or N-th band or other structural properties. An example design illustrates.

STATUS:

Presented at the 1998 Conference on Information Sciences and Systems, Princeton University, March 1998.

DATE OF ENTRY:

August 1998.