Design of companding quantizer for Laplacian source using the approximation of probability density function

TL;DR

This paper proposes new companding quantizer designs for Laplacian sources using piecewise linear and uniform approximations of the probability density function, aiming to improve signal quality.

Contribution

It introduces novel quantizer models based on approximating the PDF, with performance evaluation through SQNR and approximation error analysis.

Findings

Piecewise linear and uniform approximations effectively model the Laplacian PDF.

Proposed quantizers show improved SQNR compared to traditional methods.

Comparison indicates trade-offs between approximation accuracy and quantizer complexity.

Abstract

In this paper both piecewise linear and piecewise uniform approximation of probability density function are performed. For the probability density function approximated in these ways, a compressor function is formed. On the basis of compressor function formed in this way, piecewise linear and piecewise uniform companding quantizer are designed. Design of these companding quantizer models is performed for the Laplacian source at the entrance of the quantizer. The performance estimate of the proposed companding quantizer models is done by determining the values of signal to quantization noise ratio (SQNR) and approximation error for the both of proposed models and also by their mutual comparison.