Minimisation and prediction of the error dynamic range in finite wordlength FIR based architectures: application to the 2-D orthogonal DWT

This paper analyses the effects of fixed-point arithmetic in FIR filter based architectures, based on the roundoff statistical noise model. A novel approach, which allows minimising the architecture's error dynamic range and predicting its value according to the wordlength fractional precision, is suggested. This permits a hardware designer to preset the fractional precision for the sought implementation accuracy. The efficiency of this approach is demonstrated through the 2D orthogonal Daubechies-8 DWT family.