Digital filters are given by state equations, and classified into two fundamental groups: the lumped and distributed coefficient types. A coefficient sensitivity is defined in time domain as the upper bound of the output errors. A unified analysis of the coefficient sensitivity is developed in state space. In the analysis, the contribution of erroneous coefficients is considered on the basis of configuration matrices proposed here. The sensitivity obtained is directly dependent on the filter configuration. An algorithm to optimize the configurations of the lumped coefficient type is presented.
|ジャーナル||Technol Rep Tohoku Univ|
|出版ステータス||Published - 1976 1 1|
ASJC Scopus subject areas