Probability density functions of soft information

Abstract

It is well known that soft decoding algorithms are based on the use of reliability measures for binary and nonbinary symbols. We consider an analytical estimation of the distributions of reliability measures after soft decoding of block and convolutional codes or intersymbol interference (ISI) channels. This estimation, based on a simple geometric interpretation of reliability measures, allows the observation of different distributions for correct and incorrect symbols, and could be helpful for the analysis of soft decoding algorithms.