Predictor optimization and performance improvement in DPCM
Abstract
The practice of using the signaltonoise ratio improvement (SNRI) or the signaltoprediction error ratio (SPER) as performance indicators for optimizing the predictor in differential pulse code modulation (DPCM) systems is examined. By reconsidering the relationship between signaltoquantization noise ratio (SNR) and SNRI, it is shown that optimizing the predictor by maximizing the SNRI can produce a change in output SNR that is much greater than or much less than the increase in SNRI. The cause of the discrepancy is that optimizing the predictor may produce a change in the prediction error probability density function (pdf), which is the quantizer input. The change in the quantizer input pdf can yield an increase or decrease in the quantizer output noise power even if the quantizer is reoptimized. This effect is ignored in standard DPCM system analyses.
 Publication:

NTC '80; National Telecommunications Conference, Volume 2
 Pub Date:
 1980
 Bibcode:
 1980ntc.....2...36G
 Keywords:

 Differential Pulse Code Modulation;
 Performance Prediction;
 Signal To Noise Ratios;
 Error Analysis;
 Optimization;
 Probability Density Functions;
 Communications and Radar