Dynamically scaled fixed point arithmetic

A technique is presented for dynamically scaling a fixed-point filter at each stage of the iteration such that overflow is no more likely than when floating point arithmetic is used. The results are a generalization of earlier work.<
>