Weighted roundoff noise and pole sensitivity minimization subject to l<inf>2</inf>-scaling constraints for 2-D separable-denominator digital filters

Abstract

The problem of minimizing roundoff noise and pole sensitivity simultaneously subject to l 2 -norm scaling constraints of dynamic range for 2-D digital filters with separable denominator is investigated. A novel measure for the evaluation of roundoff noise and pole sensitivity is introduced and then an effective method for minimizing the measure is explored by converting the constrained optimization problem into an unconstrained optimization problem and by employing an efficient quasi-Newton method. A numerical example is included to illustrate the effectiveness of the present method.