LowPower FIRFilters: Part I-Fundamentals
Chip-Hong Chang, Ankita Singla · 2006
Thispaper anditscompanion paper (entitled Part As analternative toCSE approach, algorithms that II- Algorithm) together present techniques forlowpower reducethemultiplication wordlength usingdifferential realization offinite impulse response (FIR)filters using coefficients havebeenproposed in(5)-(8). Thedifferential improved differential coefficients method(DCM). Thispaper coefficients method (DCM)(5)usesdifferential coefficients presents thenecessary foundation andterminology ofthe tomultiply withinputs instead ofthecoefficients directly. DCM.Thecompanion paperdescribes ouralgorithm and Since differential coefficients haveshorter wordlength, the presents design examples. Incontrast totheconventional DCM resulting design canalso useshorter wordlength, andthus thatisformulated atthealgorithm-level, ourmethodis canreduce powerconsumption. However theDCM suffers formulated atthearchitecture-level using dedicated shift-and-fromoverheads since itrequires extra adders tocompute the add-basedcoefficient multipliers in orderto achieve considerable hardware reduction. Byemploying adifferential sumofthestored partial product ofprevious computation in coefficient-part