Reduction of floating-point roundoff noise in recursive digital filters with error feedback

Abstract
The reduction of roundoff noise in floating-point recursive digital filters by means of error feedback is discussed. Optimal (minimum-noise) error feedback is formulated, and it is shown to be different from the optimal solution in the fixed-point case. The effectiveness of this scheme is illustrated with design examples employing the direct-form structure

This publication has 7 references indexed in Scilit: