On roundoff error of fixed-point digital filters using sign-magnitude truncation

This correspondence gives the statistical meansquared error at the output of a digital filter due to roundoff accumulation when fixed-point arithmetic with sign-magnitude truncation is used.