Decorrelating (DECOR) transformations for low-power digital filters

Decorrelating transformations (referred to as DECOR transformations) to reduce the power dissipation in digital filters are presented in this paper. The transfer function and/or the input is decorrelated such that fewer bits are required to represent the coefficients and inputs. Thus, the size of the arithmetic units in the filter is reduced, thereby reducing the power dissipation. The DECOR transform is suited for narrow-band filters because there is significant correlation between adjacent coefficients. Simulations with fixed coefficient filters indicate reduction in transition activity, ranging from 6% to 52% for filter bandwidths ranging from 0.30 /spl pi/ to 0.05 /spl pi/, respectively, (where /spl pi/ corresponds to half the sample rate). Simulations with adaptive filters indicate reduction in transition activity in the F-block, ranging from 12% to 38% for filter bandwidths ranging from 0.30 /spl pi/ to 0.05 /spl pi/, respectively. The DECOR transforms result in greater energy savings and over a larger bandwidth than existing methods.