Title :
Optimum design, performance evaluation, and inherent limitations of DPCM encoders
Author :
Cyganski, David ; Krikelis, Nicholas J.
fDate :
7/1/1978 12:00:00 AM
Abstract :
The technological feasibility of implementing high-speed analog-to-digital (A/D) converters intended to achieve very high signal-to-noise (S/N) performance as well as long-term stability operation depends critically upon the intrinsic component and circuit imperfections. In studying DPCM techniques for A/D conversion, these effects were included in the design and analysis of two predictive DPCM systems, the first having digital predictor feedback while the second used an analog predictor filter in the feedback loop. In the presence of noise due to quantization and threshold inaccuracies, new optimum linear predictor coefficients that are signal statistics independent were obtained. The noise due to circuit inaccuracies was found to have a pronounced effect on the S/N performance for both the digital and the analog predictor DPCM configurations. On the basis of these newly derived results there are strong indications that, in the presence of circuit imperfections, the S/N performance attainable by a simply oversampled A/D converter constitutes an upper bound on performance for any DPCM system constructed with components of similar accuracy.
Keywords :
DPCM coding/decoding; High-speed and high-accuracy structures and applications; Prediction methods; Analog-digital conversion; Circuit noise; Digital filters; Feedback loop; Frequency; Independent component analysis; Noise shaping; Quantization; Sampling methods; Statistics;
Journal_Title :
Circuits and Systems, IEEE Transactions on
DOI :
10.1109/TCS.1978.1084500