One-dimensional Gaussian Density Function Segmentation Based on Piecewise Linear Approximation

Abstract

One of the main challenges in embedding popular audio-visual applications is the hardware implementation of one-dimensional Standard Gaussian density function. The aim of this work is to contribute with an approach to ease this implementation. The degree-l polynomial approximation has been used to segment the considered function. An interesting feature of the proposed approach is that the number of segment decreases for more important error accuracy. The second contribution is the implementation of the proposed algorithm using the content-addressable memory (CAM) method. The synthesis and physical implementation have been done using Cadence tools. The obtained performances are very competitive, especially the small area which depicts linear evolution when the bit-width increases.